.
fl‘

I _.
f. .
fl?
. __.
, r .. y‘mhxm. .
. 32%.?
gnaw .

L.

may»

.
. . 3.
by? .
x.¢.~. 5‘
..

.m. ‘61
. .F\..
. .FL .

4%. . w   33%

.L

11.1..» _

 

1 X3
.msw. p, a». u
u. a .

a .3: I
manna... 3 $14:
R5 ”1.)"
.

THESlS

.200!

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the

thesis entitled

INTEGRATION OF GIS, REMOTE SENSING, AND CENSUS DATA FOR
THE IDENTIFICATION OF LAND USE AND LAND COVER DRIVERS
IN A SEMI-ARID ECOSYSTEM

presented by

Osman C. Wallace

has been accepted towards fulfillment
of the requirements for

 

 

 

M.A. degree in Geography
Q40” /Qvl
V fir protests;

 

0.7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

PLACE lN REFURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 cJCIRC/Dateouepfis-p.“

 

INTEGRATION OF REMOTE SENSING, GIS, AND CENSUS DATA FOR
IDENTIFICATION OF LAND USE AND LAND COVER DRIVERS IN A
SEMI-ARID ECOSYSTEM

A TEMPORAL AND SPATIAL ASSESSMENT OF CHANGES IN A
SOUTHEASTERN ARIZONA COMMUNITY

By

Osman C. Wallace

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF ARTS
DEPARTMENT OF GEOGRAPHY

2000

ABSTRACT
INTEGRATION OF REMOTE SENSING, GIS, AND CENSUS DATA FOR
IDENTIFICATION OF LAND USE AND LAND COVER DRIVERS IN A
SEMI-ARID ECOSYSTEM: A TEMPORAL AND SPATIAL
ASSESSMENT OF CHANGES IN A SOUTHEASTERN ARIZONA
COMMUNITY
By

Osman C. Wallace

Understanding the drivers of landscape change is vital for assessing the total
Earth system and the impacts of ecological and anthropogenic changes at global
scale. Since rangelands cover nearly a half of the global land surface, and because a
large part of rangelands are located in semi-arid ecosystems, they serve as critical
land cover types for determining biogeochemical cycles and energy and gas fluxes.
Satellite reflectance data is often used to inventory biophysical materials and man-
made features on Earth’s surface. However, few studies have examined the utility
of remote sensing for analyzing human influence on grasslands and other range
cover types in brittle, semi-arid ecosystems. By integrating remote sensing,
geographic information systems, and census data, this thesis assesses the impact of
anthropogenic stress in a semi-arid ecosystem and identifies several key drivers of
land use and land cover change in a dynamically growing southeastern Arizona
community. In addition, a framework for a rangeland decision support system is
discussed as a value-added tool for helping ranchers, land managers, and land use
planners to determine rangeland health and forage levels, and for developing

improved and accessible grazing plans for protection of lands from overuse.

ACKNOWLEDGMENTS

I offer my deepest appreciation and admiration to my advisor, Dr. Jiaguo Qi.
His direction, patience, and affable personality made this thesis possible, and better
yet, enjoyable. I greatly thank Dr. Phil Heilman of USDA-ARS, Tucson for
providing essential data sets and sources as well as for his availability for
consultation. Another USDA-ARS staff. member, Mr. Robin Marsett, was
instrumental to this thesis due to his expertise in grazing management, remote
sensing, the study area’s land use history and floristic composition. Dr. David
Skole was instrumental in providing support for the completion of this thesis in a
timely fashion. I also wish to recognize Dr. Robert Walker and Dr. David Lusch

for serving on my committee.

iii

TABLE OF CONTENTS

iv

LIST OF TABLES vi
LIST OF FIGURES vii
CHAPTER ONE
INTRODUCTION 1
1.1 Importance of rangeland land use and land cover change 1
1.2 Remote sensing and its role in socioeconomic analysis 3
1.3 A need for DSS in rangeland management 4
1.4 Research objectives 6
1.5 Study area 8
1.6 Structure of thesis 8
CHAPTER TWO
ISSUES 11
2.1 LULCC dynamics and linkages 13
2.2 Land use history 15
2.3 Significance of pasture and grass 19
2.3.1 Contributions of grasslands to local economy 19
2.3.2 Uses of grass in soil and water conservation 21
2.3.3 Grasslands’ service to people 23
2.3.4 Role of grasslands in wildlife and game preservation 25
2.3.5 Recreational and aesthetic uses of grasslands 27
2.4 Hydrologic resources and water demand 28
2.4.1 Surface water 28
2.4.2 Groundwater and subsurface regime 29
2.4.3 Natural and anthropogenic demand for water 30
2.5 Grazing systems in southeastern Arizona 32
2.5.1 Yearlong grazing 32
2.5.2 Continuous grazing 32
2.5.3 Seasonal grazing 33
2.5.4 Deferred grazing 34
2.5.5 Deferred rotation grazing 35
2.5.6 Rotational deferment 35
2.5.7 Rest-rotation grazing 36
2.5.8 Rotation grazing system 37
2.5.9 Best pasture system 38
2.5.10 Holistic management 38
2.6 The role of fire 40
2.7 Concluding remarks 41

CHAPTER THREE

METHODS
3.1 Imagery collection and processing
3.1.1 Atmospheric correction
3.1.2 Geometric correction
3.1.3 Image co-registration
3.1.4 Image classification
3.2 GIS data
3.2.1 Coverage processing
3.2.2 Attribute data structure
3.2.3 Spatial data model
3.2.4 GIS queries
3.3 Socioeconomic data
3 .4 Data limitations
3.4.1 Temporal limitations
3.4.2 Spatial limitations
3.4.3 Spectral limitations
3.4.4 Classification limitations
3.5 Integration of remote sensing and GIS
3.5.1 Image to coverage procedures
3.5.2 Overlay analysis
3.5.3 Magnitude procedure
3.6 Inferring human-environment relationships
CHAPTER FOUR
RESULTS
4.1 Urban grth and land fragmentation
4.2 Water demand and legal issues
4.3 Grazing management
4.3.1 Mesquite invasion
4.3.2 Impact of grazing on rangelands
4.3.2.1 Infiltration
4.3.2.2 Runoff
4.3.2.3 Erosion
4.3.2.4 Water quality
4.4 Fire suppression
CHAPTER FIVE
CONCLUSIONS
5.1 Future research and decision support systems
5.2 Final remarks

43
43
44
46
47
48
52
52
52
53
53
53
54
55
55
56
58
59
6O
6O
61
61

63
65
72
79
79
82
83
84
85
87
88

92
92
99

LIST OF TABLES

TABLE

Table 1. Sensor characteristics of images.

Table 2. Sensor wavelengths.

Table 3. Improved-DOS haze correction values for 1973 image.
Table 4. Source and years of socio-economic data.

Table 5. Population and housing data totals for county and
census division, 1970-1990.

Table 6. Population and housing data totals for census tracts, 1990-1997.

Table 7. Number of employees and taxable payrolls per economic sector
in Cochise county, selected years.

Table 8. Agricultural patterns in Cochise county, selected years.
Table 9. Total area, in hectares, of urban land cover, selected years.

Table 10. Percent change in urban land cover by census tract,
selected years.

Table 11. Land cover change transformation matrix, selected years.
Table 12. Well attributes per census tract, 2000.

Table 13. Satellite-based sensors along with their spectral region,
pixel resolution, and orbital characteristics.

vi

PAGE

104

104

105

105

105

106

106

107

107

108

108

109

109

LIST OF FIGURES

FIGURE
Figure 1. Political boundary of study area.

Figure 2. Authenticated Spanish and Mexican land grants in
southeastern Arizona.

Figure 3. GIS and remote sensing processing and integration.

Figure 4. Landsat MSS image and metadata, 1973.

Figure 5. Landsat TM image and metadata, 1985.

Figure 6. Landsat TM image and metadata, 1992.

Figure 7. Landsat ETM+ image and metadata, 1999.

Figure 8. Classified land cover map and area totals per census tract, 1973.

Figure 9. Classified land cover map and area totals per census tract, 1985.

Figure 10. Classified land cover map and area totals per census tract, 1992.

Figure 11. Classified land cover map and area totals per census tract, 1999.

Figure 12. Sierra Vista census division.
Figure 13. Registered wells in 2000 per census tract.

Figure 14. Land cover area totals in Sierra Vista census division,
selected years.

Figure 15. A simple three—tier DSS design.

Figure 16. A simple GUI design for a rangeland DSS.

vii

PAGE

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

CHAPTER 1 - INTRODUCTION

1.1 INIPORTANCE OF RANGELAND LAND USE AND LAND COVER

CHANGE

Land cover change may be the most significant agent of global change; it has an
important influence on hydrology, climate, and global biogeochemical cycles
(Skole et a1. 1997). Characterizing the forcing factors that drive changes in
landscapes, and the resultant impacts on biogeochemical and hydrological cycles
and energy and gas fluxes, is essential to understand the total Earth system and the
effects of natural and human-induced changes on the global environment. In
addition, assessing the responses to drivers of changes in land cover and land use,
particularly in those parts of the world that are currently undergoing the most stress,
is an important function for understanding the consequences of land cover and land
use changes as they impact ecological processes as well as evaluating what human
activities contribute to changes occurring on the landscape. Arguably, land cover
change will have greater impact on human habitation, especially in arid and semi-
arid environments, than climate change over the next 20 to 50 years. It is an issue
with far-reaching policy implications, internationally, nationally, and locally.
Indeed, global cover change is as inextricably linked to policy and sustainable
development as it is to basic research needs.

The collective area of rangeland is large. If forested ranges and natural
vegetation in tropical savanna and tundra areas are included, the total land area of
rangelands may be as high as 47 percent of the global land surface (Food and

Agriculture Organization 1995). In the United States, rangeland cover type expands

to even higher spatial coverage; it encompasses one half of the conterminous land
area. Although rangelands of the world have been used by livestock pastoralists for
thousands of years, information about the land’s response to anthropogenic
processes and grazing is limited (Mouat and Hutchinson 1995). The use of this
important and often brittle ecosystem is based on conflicting requirements. On one
hand, the valuable rangeland resource is exploited to the advantage of livestock
producers. On the other, maintenance and improvement of the range condition is
essential not only for the obvious environmental benefits, including controlling
erosion and sediment delivery as well as providing aesthetics and recreation, but
also for determining sustainable optimal grazing capacity and maximum
profitability of range sites while preventing their overuse at any time of the year.

A large part of rangelands resides in the semi-arid zones of the world. These
zones consist of fragile ecosystems that are substantially modified by the activities
of people. For example, increasing human populations in the western United States
during the 19905 has resulted in greater demands on the ecosystems of this mostly
semi-arid zone. Faced with the intricate problem of population demand for limited
resources like land, water, and agricultural production, the challenge to policy
makers, ranchers, land managers, and municipal officials of the region is to manage
these zones so that growth is controlled, productivity is stabilized, and

environmental deterioration is decreased.

1.2 REMOTE SENSING AND ITS ROLE IN SOCIOECONOMIC ANALYSIS

Satellite images offer a unique facility for observing and documenting changing
land cover. Assessing and characterizing how land use and land cover changes are
manifested at multiple space and time scales using multispectral remote sensing
data are critical and vital steps in expanding scientific knowledge of the Earth
system using the unique vantage point of space. Satellite data could be utilized to
gain a better understanding of drivers of land use and land cover change, to relate
observed landscape patterns to processes of change, and to provide the initiative for
strategic planning after gaining a clear understanding of drivers of change.

The large area coverage and frequent acquisition cycle of satellite images in
digital format make earth observation data useful for observing region-size areas
and for integrating with other types of spatial information using Geographic
Information Systems (Wagner and Ryznar 1999). Although remotely sensed data
are inherently suited to provide information on urban land cover characteristics
related to ecological, demographic, socioeconomic, and dynamic aspects of
deveIOped regions at various spatial and temporal scales (Ridd 1995), few studies
have looked at how satellite data may be utilized to document and analyze human
intrusion and influence on grasslands and other range cover types in brittle, semi-
arid environments. Unlike their widespread comfort with municipal economic and
demographic statistics, social scientists are relatively new to the opportunities
provided by the growing archive of satellite-derived earth observational data. This
fact was highlighted by a recent study of the US. National Research Council

entitled People and Pixels (National Resource Council 1998). While traditionally

satellite data were used to address questions concerning vegetation types and
amounts, my study asks a different question. How can satellite data be used to

understand underlying anthropogenic processes that shape the environment?

1.3 A NEED FOR DECISION SUPPORT SYSTEMS IN RANGELAND
MANAGEMENT

Due to the extensive nature of rangelands and the recognized need to manage
them at low cost, remote sensing and remote sensing-based products are considered
to have significant promise for the future (Tueller 1989). As the increasing human
pressure on semi-arid landscape mounts in the western United States, there is a
greater need for an information delivery system to help livestock farmers, land
managers, government officials, environmentalists, and other interested parties to
quickly locate areas of interest based on their respective criteria. For example,
livestock farmers and land managers, utilizing near-real time, satellite-derived maps
of vegetation indices (NDVI, NDSVI) and other biophysical parameters (Fc, LAI,
biomass estimates), might prefer to focus on areas suitable for seasonal grazing,
deferred and/or rotational grazing, as well as those areas needing extensive
protection from overuse in semi-arid regions (Qi et a1. 2000). As a result, ranchers
and managers would know where to graze the cattle and how much fodder pastures
produce per unit area.

Current satellite-based sensors such as TM, ETM+, HRVIR, VISSR, and SAR
have the disadvantages of fixed spectral bands that are inappropriate for within-field
analysis, inadequate repeat coverage for intensive agricultural management, long

time periods between image acquisition and delivery to user, and large pixel sizes

that are unsuitable for precision crop management (PCM) (Moran et al. 1997).
However, a decision support system (DSS) designed to collect and analyze data
from recently launched ASTER and MODIS sensors mounted on the EOS-AM
platform, in addition to other commercial sensors such as EarlyBird, QuickBird,
SIS, OrbView, IKONOS, and Resource-21, with short off-nadir repeat cycle, fine
pixel resolution, and short delivery time from acquisition to user should overcome
such limitations. The only caveat to image acquisition from these sensors will be
the cost. For example, SPOT Image Corp. charges an extra $2000 (nearly twice the
normal cost) for requests of image acquisitions guaranteed on a certain date or in a
narrow time interval due to conflict with other users (Moran et al. 1997). Future
work should focus on determining which remote sensing-based DSS applications
are most economically beneficial and technically feasible, and furthermore, funded
multi-season projects with satellite-based sensors designed specifically to
investigate the economic and scientific viability of remote sensing products for
PCM applications should be given high priority (Hough 1993).

Decision support systems have evolved in response to the need to create
‘mechanisms to develop and transfer technology that meets the needs of individual
landholders and overarching values of society’ (Stuth et al. 1993). The delivery
system, therefore, should provide rangeland managers with a proactive tool to
survey large areas to determine rangeland health and forage levels, and to develop
improved and accessible grazing plans for protection of lands from overuse.
Further, the product should offer additional potential benefits including better

wildlife habitat monitoring (sparrow nesting sites in grass cover), and improved

fire-fuel estimation and watershed management strategies. Finally, the product
must be designed in accordance with input from the end user at every stage of
development. Design and implementation strategies for a web-based decision
support tool essential for monitoring the landscape for sustainable land use
practices, successful regional planning, and effective policy implementations will

be discussed in the final chapter of the thesis.

1.4 RESEARCH OBJECTIVES

The main objective of this study is to demonstrate how satellite data might
contribute to an understanding of human-environment interactions and impacts of
change in land cover in a fragile and semi-arid rangeland environment. By
analyzing remotely sensed data with an appropriate processing model, an objective
method may be developed for relating remotely sensed data from the urban or
urbanizing environments to biophysical parameters relevant to hydrologic and
socioeconomic attributes (Jensen et al. 1994; Phinn 1998; Phinn et al. 1998). To
that end, coupled with demographic and statistical data and a GIS spatial model, an
analysis of multitemporal satellite imagery will be conducted to assist in
qualitatively determining the human drivers and proximate cauzes responsible for
land use and land cover change over the Sierra Vista landscape. Specifically, the
temporal change in economic base of the region and loss of productive range cover
will be evaluated through integration of remote sensing, GIS, and census data.

To sum up the main objective, the study will illustrate that (1) utilizing satellite

technology is an effective approach to document cover change and to determine

anthropogenic causes Of dryland degradation. In addition, using a qualitative
approach, the thesis will infer the main drivers of land use and land cover change in
a region that is undergoing dryland degradation. According to Dregne (1983), the
definition of dryland degradation in North America is:
(a) Substitution of edible grasses by human-based development, as well as by
forbs and shrubs that are not or less edible.
(b) Badly controlled drainage and leaching resulting in 50 percent reduction in
grass and/or crop biomass.

The evidence of dryland degradation will be based on multitemporal,
multispectral satellite imagery covering a time series from 1973 to 1999. The study
will further reveal that (2) continuous urban intrusion and changing economic base
of the region have led to increased fragmentation and degradation of rangelands.
Further, (3) hydrologic and legal issues resulting from increasing residential,
municipal, and industrial demand for water from depleting regional and floodplain
aquifers influence rangeland sustenance. And finally, (4) range management
practices and fire suppression policies function as extemalities that transform
species composition of native rangelands in southeastern Arizona. To a large
degree, the framework for the identification of LULC drivers will rely on bid rent
theory and concept of externalities, i.e., fire suppression policies and range

management practices.

1.5 STUDY AREA

Landsat data of the Sierra Vista census division in southeastern Arizona (Figure
1) provides the basis for the exploration. Sierra Vista (Spanish for "Mountain
View”) is a thriving community of 40,000 people that serves as the regional center
for southeastern Arizona. Located approximately 75 miles southeast of Tucson, the
city is surrounded by Huachuca, Dragoon, and Mule Mountains, and is bordered on
the east by the San Pedro River. Changes in land management techniques, coupled
with tremendous urban growth, have altered the vegetation patterns and dynamics
in the San Pedro River basin (Qi et al. 1999). In the last thirty years, the Sierra
Vista census division has experienced a rather sizable increase in developed land.
According to satellite data, in the period between 1992 and 1999 alone, the total
area of developed land rose from 5,642 hectares to 7,101 hectares — a 26 percent
increase. Between 1973 and 1999, the increase in the urban cover category was 192
percent. Most of the new growth took place over areas of native, perennial
grasslands in the San Pedro River basin. For this study, therefore, there was a need
to map the spatial and temporal distribution of land cover types in order to

understand how human activities affect the ecosystem in this semi-arid region.

1.6 STRUCTURE OF THESIS

Chapter 1 has introduced the importance of rangeland land use and land cover
change, the role of remote sensing as a socioeconomic analysis tool, some
preliminary discussion on decision support tools for rangeland management, as

well as a list of research objectives, and a description of the study area.

Chapter 2 discusses relevant issues related to rangeland ecosystems and
rangeland management in the study area. The topics of discussion include:

o a review of LULCC dynamics and linkages,

0 southeast Arizona land use history prior to 1973,

o the important role grasslands play in the environmental and economic
sustenance of greater Sierra Vista,

0 hydrologic resources and water demand on the Upper San Pedro watershed,
and,

o rangeland management and grazing methodologies, and the role of fire in
southeastern Arizona.

Chapter 3 introduces the data structure of the analysis. Sources and metadata of
census information and satellite imagery, as well as methods utilized for atmospheric
correction, geometric correction, co-registration, incorporation of ancillary data, and
image classification will be discussed. In addition, temporal constraints of remote
sensing and census data, spatial limitations of census-derived information, spectral
limitations of visible satellite imagery, and classification limitations of remotely
sensed data will be given attention. Due to the integrated nature of this research
effort, incorporation of remote sensing, GIS and socio-economic data requires a
discussion of GIS methods including coverage processing, data structure, and the
GIS spatial data model. Additional methodologies covered include the development
of analysis tools consisting of overlay and magnitude GIS algorithms designed for

the convergence of demographic census division and tract coverages with remote

sensing-based land cover classification images. The final section of the chapter
focuses on inferring human-environment relationships from the data sources.

Chapter 4 examines the results of the study. The main body of the chapter will
consist of identification and discussion of land use and land cover drivers for
greater Sierra Vista. Furthermore, contributions of remote sensing and GIS to the
identification of the drivers of change will be given attention.

Finally, Chapter 5 discusses design and implementation strategies of a prototype
web-based DSS for rangeland management. The chapter concludes the study by
summarizing the research initiative and its results, and examines future research

needs and recommendations for rangeland land use and land cover analysis.

10

CHAPTER TWO—ISSUES

Land use and land cover change dynamics and linkages are discussed first.
Definitions of terms as well as the function of land use and land cover change as a
research endeavor for global, regional, and local applications will be highlighted.

This discussion of multitemporal environmental change analysis using remote
sensing does not cover the time period before 1973, as no satellite data over the
Sierra Vista census division were available prior to that date. Therefore, a
clarification of the region’s land use history prior to first date of data acquisition
will reveal human impacts on greater Sierra Vista before the availability of remote
sensing data needed for land use and land cover change analysis. An overview of
pre-1973 human influence on the landscape is essential for understanding the
continuum of human-environment interactions in the region.

The importance of grasslands in our national life and in southeastern Arizona
has been difficult to assess. As a result, these lands have generally been accorded
far less consideration than warranted by the important roles they fulfill. Judgments
of knowledgeable specialists on prudent conservation and management of the
exceedingly varied categories of these land areas are fragmented and dispersed
(Sprague 1974). To comprehend the significance of grasslands, two dimensions of
this resource should be examined, the importance of such lands to the local
economy of Sierra Vista and their function and significance as a cover type in a

fragile, semi-arid ecosystem.

ll

No discussion of environmental analysis in southeastern Arizona can be
adequate without an examination of hydrologic resources and water demand.
Sustenance of the urban and military populations in the Sierra Vista-Fort Huachuca
area and irrigation water for agriculture along the San Pedro River constitute the
two major cultural water uses in the Upper San Pedro Basin. Hydrologic studies of
the Upper San Pedro Basin have shown that continued pumping in the greater Sierra
Vista region at its present rate or higher will most likely threaten surface flows of
the San Pedro River (Lacher 1994). This threat, discussed in detail in later chapters,
is posing serious concerns to both the environmental and economic vitality of the
region.

The vast expanses of rangelands, the numerous benefits that may be derived
from them, and the acceleration of changes imposed on them combine to create
unique challenges for range managers responsible for range mapping and
management. These challenges can ofien be effectively met by remote sensing
techniques. Changes observed in remotely sensed data could complement
ecological assessments of restorable rangelands (Tueller 1989). Restoration of
degraded lands caused by poor grazing methods, as well as selection of
environmentally and economically beneficial grazing methods for effective long-
terrn management require location-specific information and monitoring of the
spatial patterns of change. Regional maps of rangeland vegetation change should
provide range managers with information for setting long-term grazing practices.
Although many studies of production and protection have already been done on

grazing practices in arid and semi-arid rangelands of the west and southwest (Kruse

l2

and Jemison 2000), future research should consider broadening the scope of
livestock grazing systems to be responsive to native ecosystem parameters, the role
of fire, and functions such as critical habitat requirements for biodiversity. Ongoing
biological research should be developed with specific grazing practices in mind. In
addition, grazing management has an economic component that is difficult to assess
within an ecosystem management context. When new practices are implemented,
sometimes involuntarily, the costs of these alternative methods should be
understood. A broader understanding of the relationships of grazing management
practices to overall ranch operations and policies imposed by land management
agencies are needed to make informed decisions on alternative grazing systems.
Selection of grazing methods for use on semi-arid rangelands requires a thorough
understanding of the pros and cons of each practice and the cost to implement it.
Items to consider include land use restrictions, shared uses, types of animals to be
grazed, topography, climate, soils, accessibility to water, and vegetation (Holechek
et al. 1998). Therefore, it is necessary to include a short discussion of typical
grazing methodologies practiced in Arizona to understand their role in the
implementation of environmentally and economically beneficial livestock
management strategies as well as to comprehend their impact on land cover change

in semi-arid grasslands.

2.1 LAND USE AND LAND COVER CHANGE DYNAMICS AND LINKAGES

Land use and land cover change research is an endeavor to understand the
effects of natural and human-induced changes at local, regional, and global scales

within the total Earth system. The distinction between land cover and land use is

13

fundamental. Land cover is the observed physical cover, as seen from the ground
or through remote sensing, including the vegetation (natural or planted) and human
constructions (buildings, roads, etc.) which cover the Earth’s surface (Food and
Agriculture Organization 1997). Water, ice, bare rock or sand surfaces count as
land cover. Land use is based upon function, the purpose for which the land is
being used. Thus, a land use can be defined as a series of activities undertaken to
produce one or more goods or services (Food and Agriculture Organization 1997).
A given land use may take place on one, or more than one, piece of land and several
land uses may occur on the same piece of land. Definition of land use in this way
provides a basis for precise and quantitative economic and environmental impact
analysis and permits precise distinctions between land uses. Land use and land
cover change analysis is regarded as a tool for a more sound scientific
understanding of the Earth system that provides a foundation for sustainable
development—economic development that meets the needs of the present without
compromising the ability of future generations to meet their own needs (Turner et
al. 1995). It examines theoretical linkages between social and natural systems
through land use, direct utilization, resource competition, and pollution (Long and
Harwell 1993). Direct utilization comprises uses that do not lead to land-cover
change from natural vegetation (e.g., recreational use). Resource competition
occurs when humans exploit resources or components of the natural systems
essential to natural-system function and process (e.g., water), while pollution is the
transmission of harmful residuals of human activities, both consumptive and

productive, into the natural area. These impacts are applicable to natural areas close

14

to human settlement and use. Understanding the human and environmental drivers
(i.e., causes of change) of land cover change and its impact at different ecological
scales is the fundamental aspect of the research initiative (Long and Harwell 1993).
Thus, LULCC attempts to answer the nature and extent of land cover and land use

change and the consequences for sustained productivity.

2.2 LAND USE HISTORY

The semidesert and plains grasslands, which make up about 45 percent of the
land cover of southeastern Arizona, (Brown and Lowe 1980) have been the
lifeblood of southern Arizona’s cattle industry. In fact, most of the recorded
changes in the grasslands of southeastern Arizona occurred after the beginning of
large-scale cattle ranching and fire exclusion in 1870 (Bahre 1977).

During the Spanish occupation of the region, livestock, especially cattle and
horses, may have substantially degraded the rangelands adjacent to the presidios
and villages, but there is no evidence that livestock numbers were very large. In
1804, for example, only 3,500 cattle, 2,600 sheep, and 1,200 horses were reported
at Tucson, then the largest Spanish settlement in Arizona with a population of
1,015; at Tubac, the other major settlement, only 1,000 cattle and 5,000 sheep were
counted (McCarty 1976).

The introduction of Old World technology, crops, and livestock by the
Spaniards eventually led to major changes in the grasslands of southeastern
Arizona. Because southeastern Arizona was on the margin of New Spain’s northern

frontier, however, Spanish settlement was sparse and concentrated in the upper

15

Santa Cruz valley. There is no doubt that livestock, especially cattle, horses, and
sheep, were abundant near the presidios and missions, but their effect on regional
grasslands was probably minimal.

After Mexico gained its independence in 1821, several large stock-raising grants
were established between 1821 and 1827 near the present international boundary
(Officer 1987). These grants now represent some of the largest private
landholdings in southeastern Arizona (see Figure 2; McClaran 1995). In their
heyday, thousands of horses, cattle, and sheep were supposedly run on these grants.
According to Bartlett (1854), there were 140,000 cattle on the Babocomari and San
Bemardino grants alone. The primary social driving force for this first significant
land use and land cover change in the region was the stock-raising grants, coupled
with the government initiative to settle what was then the fi'ontier areas, followed
by secondary driving forces of increasing demand for beef attributed to rising
population and income of the central Mexican core.

Beginning in 1846, when war broke out between the United States and Mexico,
and continuing through 1854, when Mexican jurisdiction over the region ended
with the Gadsden Purchase, several military and scientific expeditions and
thousands of immigrants heading for the California goldfields crossed southeastern
Arizona. According to numerous travel accounts, it is estimated that there may
have been as many as 100,000 wild cattle on the abandoned Mexican land grants
during the 18405 and early 18508 (Christiansen 1988). However, considering the
general lack of livestock water developments then (there were no windmills or

stock tanks) and the intermittent nature of most of the streams, it is difficult to

16

believe that large numbers of cattle could have been supported by the grass and
browse in the rangelands adjacent to major sources of perennial water.
Furthermore, large-scale cattle ranching during the 18205 would have been curtailed
by Apache depredations. Nevertheless, probably the most significant human impact
on the grasslands during this period was large-scale ranching.

After the Gadsden Purchase Treaty of 1853, major Anglo settlement of
southeastern Arizona did not begin until the 18705. The subjugation of the Apaches
and the completion of the Southern Pacific Railroad from Tucson to El Paso in
particular facilitated the expansion of the ranching industry and heavy grazing of
the grasslands. By 1885, so much investment capital, both foreign and domestic,
had poured into southeastern Arizona’s ranching industry that cattle numbers
exceeded all expectations. By 1891, cattle numbers in the region had reached
nearly 400,000, before the severe drought of 1891-1893 caused a massive die-off
(Bahre 1977). The drought and its impact on cattle parallel with Malthusian
positive checks on population growth and can be interpreted as triggering events,
however overstocking, not the rising human population of the region, was the
culprit in this case. After the drought, major changes in the grasslands became
apparent, many of which have persisted to the present. The ranching industry had
stabilized by 1900, but overstocking and overgrazing continued (McClaran 1992).

The early twentieth century saw a major nationwide push to conserve range and
forest resources, and in 1907, the United States Department of Agriculture Forest
Service curtailed the free use of forage resources in the newly established national

forests and advocated grazing control and fire suppression. The Taylor Grazing Act

17

was enacted in 1934 to prevent overgrazing and consequent soil and water
deterioration, as well as to stabilize the livestock industry dependent on public
rangelands (Voigt 1976). Grazing districts were established and grazing permits
were issued. The Forest Service, the Bureau of Land Management (BLM), and
other federal agencies charged with range and watershed protection introduced
changes in the grasslands largely intended to improve them for cattle. During this
time period, population growth and demand for land were the drivers for improved
agricultural techniques and development. These chain of events follow closely with
Boserup’s (1965) theory on conditions of agricultural grth that critiques the
Malthusian and neo-Malthusian position which ignores preservation and
improvement of the productivity of land under increasing population. In further
support of Boserup’s theory on land use, innovations such as surface water
irrigation aided by the development of efficient, low-cost irrigation pumps in the
19405 cleared thousands of hectares of grassland for intensive irrigated agriculture.
This agricultural advancement not only increased labor and land productivity, but
also was driven by population pressure on resources.

The rapidly growing population of Arizona, especially since the 19405, has led
to expanded urban and rural development on privately owned lands and is
threatening federal and state trust lands, which are continually being sold to the
public for development. Since 1950 the number of rural subdivisions being built in
southeastern Arizona has exploded, especially in Cochise County, which has both
the largest area of grassland and the largest amount of private land in the state (41

percent)---almost all of it being used for agriculture, ranching, or subdivision

18

development (Hecht and Reeves 1981). According to Census of Agriculture
statistics, cattle numbers in the county have declined from 81333 in 1974 to 69950
in 1999 in part because of tract development in former rangelands, and also because
much of private grazing and agricultural land has been purchased by people seeking
rural retirement or investment opportunities. As a result, between 1974 and 1999,

total farm acres decreased from 2.1 million to 1.2 million.

2.3 SIGNIFICANCE OF PASTURE AND GRASS
An adequate evaluation of the significance of the greater Sierra Vista grasslands
should include the following aspects:
1) The support they provide to the local livestock industry.
2) Their use in soil and water conservation.
3) Their role in regional water resource development and protection in service to
people of nearby areas.
4) Their importance in balanced programs of wildlife and game conservation and
protection.
5) Their contribution to outdoor recreation and aesthetics.

The following sections discuss the above aspects in greater detail.

2.3.1 Contributions of grasslands to local economy
Grasslands in greater Sierra Vista, and rangelands in general, due to poor soil
and dry climate regime, are not basically suited for crop production or other

intensive agriculture but can be used by domestic livestock and wildlife that can

19

convert the forage on this acreage into food products useful for human
consumption. This fact is reflected in agricultural statistics of the region which
reveal that in 1997 the cash sales of livestock represented 64 percent of the total
agricultural income in Cochise County, and on average, three-fourths of the beef
cattle in Cochise County receive the major part of their feed (76 percent) from
pasture (USDA 1997). In addition, cattle and calves were the major source of cash
receipts to farm income. Therefore, grass cover contributes greatly to livestock
feeding and thus indirectly to the food supply. If not for the ruminant sector of the
livestock industry, it would have little market value as an agricultural product.

What will be the effects on the rural economy of the region from reduction of
rangeland grazing? It has been estimated that a 20 percent reduction in grazing
would cause an 11 percent decrease in gross ranch income (Caton 1985). To
simulate the effect of this reduction, all range ranches were studied in nearby
Hidalgo county in New Mexico (Bromley 1988). The dependent ranches generated
over $5 million worth of new money in the county from export sales and spent
almost $2.8 million for goods and services, which included agricultural,
automotive, construction, and transportation services. As a result of economic
interdependence in this rural county a reduction in rangelands caused an 11 percent
reduction in ranch income, and $2 million total reduction in county business
receipts.

Since production of beef cattle and sheep involves approximately the same
combination of inputs for services required from the business community (Bromley

1988), this study gives some insight into the importance of grazing lands. More of

20

this kind of information on interindustry dependence would be usefirl for evaluating
the true importance of rangeland resources to the local livestock industry and

economy.

2.3.2 Uses of grass in soil and water conservation

Hibbert (1981) obtained remarkable evidence of the increase in water yield that
results from converting subhumid watersheds in Arizona from Chaparral to grass.
He demonstrated that eliminating brush with aerial herbicide applications of 2,4,5-T
and fostering growth of grass increased water yield 1 to 4 inches per year. Deep-
rooted shrubs depleted the soil moisture reservoir more than did the grasses,
resulting in greater retention of the winter rains that provide most of the streamflow.
This study emphasizes the excellent attributes of good grass cover in beneficially
modifying land surfaces for desirable water behavior on the land. Additional
benefits include the protection of the soil surface and the reduced surface runoff
under grass that are also associated with great reductions in soil loss and sediment
delivery (Sprague 1974).

The protective action of grass is utilized by many range conservation planners
(Sprague 1974). Planners use grass cover on waterways to provide the most
economical means of conveying surface runoff from fields without incurring
erosion of the channel. Grass maintains good infiltration, curbs runoff, and nearly
eliminates sediment delivery. This has important implications in the transport of
potential pollutants from the land to surface waters. In terms of total mass,

sediment itself is the dominant water pollutant, and something in excess of 4 billion

21

tons of sediment move from the land to water channels during the average year in
western United States (Wadleigh 1971).

Studies at the Campbell Soup Company plant at Paris, TX provide clear
evidence of the exceedingly remarkable effectiveness of grass in rectifying polluted
water at low cost (Thomwaite 1979). The Paris canning plant discharges up to 3.6
million gallons of water a day. This water is contaminated from use in the
preparation of food products and contains from 550 to 900 ppm biochemical
oxygen demand (BOD). Thus the effluent from the Paris plant is about three times
as polluted as raw city sewage. Paris is located in the Houston-Austin soils
association. The soil is a silt and clay loam, similar to White House-Bemardino-
Hathaway association found in greater Sierra Vista, which has been ravaged by
severe sheet and gully erosion while under crop development. The soil is so
erodible that it always should have had grass cover (Thomwaite 1979). This
impervious soil has an infiltration rate of 0.1 inch per day or less when swelling has
eliminated the major cracks that may occur upon drying. Obviously, runoff from
rainfall or irrigation is exceedingly high. Some 400 acres of land were smoothed
with a land plane and terraced at 200- to 300-foot intervals. The smoothed areas
were planted to grass afier a sprinkler irrigation system with underground mains
was installed along the upper side of the smoothed area between the terraces. On
establishment of the grass cover, the canning plant effluent was applied at the rate
of about 0.1 inch per hour.

Since the soil had exceedingly low infiltration capacity, 80-90 percent of the

applied polluted effluent became nmoff at the lower terrace of each smoothed area.

22

Even though most of the polluted water did not enter the soil but was intercepted by
the grass and trickled slowly downslope on the soil surface, 99 percent of the BOD
was removed by the time the water reached the lower terrace. Up to 90 percent of
the phosphorus and nitrogen in the polluted effluent was removed by the time
tricklings reached the lower terrace.

This field application, conducted on a soil association similar to that found in
greater Sierra Vista, is a demonstration of the tremendous capability of grass to
rectify seriously polluted water. The cost for this rectification procedure is
exceptionally modest. The capacity of good grass cover to disperse the entry of
rainfall, to maintain maximal infiltration capacity of soil, to minimize runoff, to
nearly eliminate sediment delivery, and to rectify polluted water so as to provide
continual protection of soil and water resources makes grasses of fundamental

importance in many undertakings for improved environmental protection.

2.3.3 Grasslands’ service to people

Not all the sediment comes from farms and ranches. Much of it originates from
housing and industrial developments, road construction, gullies, and streambanks.
It is not unusual to lose 100 tons per hectare in one rain event from a construction
site that is not adequately protected with grasses and mechanical measures (Sprague
1974). An established grass cover around municipal outskirts would greatly reduce
the problem of runoff and sediment accumulation resulting from urban

development.

23

In addition, researchers have stated that future reuse of sewage effluents will not
be a question of economics but one of necessity in the arid West (Hibbert 1981).
Many cities and industries are using forage plants to recycle municipal sludge and
process wastes. For example, the Flushing Meadows Projects at Phoenix, Arizona
is designed to handle effluent from an activated sludge plant serving the
metropolitan area. The objective is to reclaim sewage water that can be used to
recharge the groundwater. In order to meet this objective, six basins, 20 by 700 feet
in size, were seeded to bermudagrass, while two were left with a gravel base.
Inundations ranged from two days wet and three days dry to two weeks wet and ten
days dry. The vegetated basins gave the best infiltration rate and the gravel basin
the lowest. Water samples from sewage effluent and a 30-foot well on the site
revealed sustained decrease to adequate levels in BOD, ammonium, nitrate,
phosphorus, and bacteria after percolation of effluent through the soil profile
(Sprague 1974).

Other research states that each year about one million acres in Arizona are
currently subject to erosion as they are in transition from farming and forestry uses
to urban, transportation, industrial, and related developmental uses (Rummell
1987). In addition to areas denuded by construction and development, highways
need erosion control both during construction and maintenance. Obviously, the use

of grass is prominent in such control.

24

2.3.4 Role of grasslands in wildlife and game preservation

Wildlife populations are regulated by the availability of food, water, and cover.
The closer these basic components of wildlife habitat occur together, typically the
greater diversity in wildlife species and total numbers. Grassland management
practices and livestock grazing can promote habitat diversity or destroy habitat
diversity, depending on their application. Diversity in both plant species and plant
communities over short distances is the key to healthy wildlife populations. A good
grass cover, along with carefully controlled livestock grazing, is beneficial to the
habitat needs for wildlife (jackrabbits, rodents, and other small mammals) and birds
(sparrows and raptors) (Cook 1966) that typically reside in greater Sierra Vista.

Many range management practices can be used to improve habitat for wildlife
as well as increase forage for livestock. Livestock affect wildlife habitat directly by
removal and trampling of vegetation that could otherwise be used for food and
cover. When defoliation is carried to an extreme, the number of wildlife species
will decline because of loss of diversity in food and cover. Size, pattern, and
location of defoliated areas determine both how and to what extent wildlife is
affected. Moderate grazing (30 to 50 percent use of current year’s herbage
production) in broken terrain generally results in heavy use of lowland areas close
to water and light use of upland areas removed from water sources (Roath and
Krueger 1982). Such grazing favors a mosaic of grazed and ungrazed vegetation,
although the best wildlife habitat near and around the watering areas will receive
heavy use (Cook 1966). Although uniform use of vegetation is desirable fi'om the

standpoint of maximizing livestock production, it can be undesirable to wildlife

25

because of reduced habitat diversity, lack of heavy escape grass cover, and greater
social interaction between domestic and wild species (Brown 1978; Mackie 1978).
However, in many situations water development for livestock has been beneficial to
wildlife in the western United States because it has permitted the use of areas from
which wildlife were previously excluded due to lack of free water. Wood et a1.
(1970) reported that development of water for livestock increased mule deer
numbers, as well as several game and nongame species, in southcentral New
Mexico.

Virtually all the gallinaceous game birds depend heavily on annuals and/or early
successional forbs for food (Greenfell et al. 1980). In addition, annuals are
important foods of many species of rodents and lagomorphs (Wood 1969). Annual
grasses and forbs are particularly important to these categories of wildlife during
the winter because they have large seeds that are high in energy, unlike most scrub
species.

Vegetation requirements for cover of many wildlife species are often much
different than those for livestock feeding. These requirements may also vary
drastically between seasons for some wildlife species. Therefore, diversity in
grassland vegetation structure, vegetation composition, and terrain favors the
highest diversity and density of wildlife. It is important to note that multiple species
of grass, rather than scrub or mesquite mixture, is the definition of grassland
diversity. If carefully controlled, livestock grazing can be a useful tool to obtain

and maintain habitat diversity.

26

2.3.5 Recreational and aesthetic uses of grasslands

The urbanization of the greater Sierra Vista region does have some advantage to
the ranching industry and nearby grasslands. Increased human population provides
considerable opportunity for ranchers who are willing to diversify their enterprise.
Multiple use involves the harmonious use of the range for more than one purpose
(Society for Range Management 1989). The potential for fee hunting is a good
example. Other forms of recreation, such as packing trips, horseback riding,
sightseeing trips, bird watching, and dude ranching are proving to be highly
lucrative in certain areas (Holechek et al. 1998). These enterprises in some cases
bring in far more net income than sale of livestock. It appears that recreational
enterprises will increasingly displace livestock as the main source of income from
ranches in many parts of the western United States (Holechek et al. 1998).

Grass, if used properly, can go a long way toward preventing or abating the
unattractiveness of streams and lakes laden with such pollutants as silt and other
sediments, sewage, and agricultural waste. Grass has been used for years to heal
eroding hillside gullies, to prevent scouring of waterways and drainage systems, and
to tie down soil subject to sheet erosion (Sprague 1974). We are learning more
about grass-covered soils as filter fields to effectively dispose of sewage sludge,
livestock manures, and certain processing wastes—all of which are harmful
pollutants to the environment.

Alternative uses of the various types Of grasslands, i.e., the degree to which
present occupation of land by grass cover and its associated use constitutes the most

effective ecological adjustment of these land types for balanced and prudent

27

management of the total environment, deserves more attention of researchers and

land managers alike.

2.4 HYDROLOGIC RESOURCES AND WATER DEMAND
2.4.1 Surface water

The San Pedro River originates about 30 miles south of the International
Border, north of Cananea, Sonora, Mexico. The river enters the United States near
Palominas, Arizona and flows northward toward its confluence with the Gila River.
Most of its tributaries are ephemeral, flowing only in response to rainfall or
snowmelt events. The San Pedro River is perennial (flowing continuously all year)
for approximately 36 of the 62 river miles in the United States portion of the Upper
San Pedro River Basin. The floodplain aquifer recovers during the winter, with
streamflow returning over most of the river reach in November of December, and
remaining until March or April (ADWR 1990). Outside of the perennial reach, the
San Pedro flows intermittently. Most of the perennial reach of the San Pedro River
falls inside the San Pedro Riparian National Conservation Area (SPRNCA), a
federal reserve managed by the United States Bureau of Land Management (BLM).
The reserve constitutes a narrow band of 56,000 acres of riparian habitat along the
San Pedro River (USDA 1988). The United States Congress established the
SPRNCA in 1988 in an effort to protect the rare riparian habitat from damage due
to increasing demand for water in the surrounding area (National Resource Council

1992).

28

Like most rivers, the San Pedro River has two major flow components: runoff
and baseflow. Runoff occurs after precipitation events or as result of snowmelt, and
lasts a few days until all flow is either lost to outflow from the basin or to bank
storage along river. The highest annual flows in San Pedro River and its tributaries
occur between July and September, and are typically of short duration. Baseflow
results from the discharge of groundwater to the stream and sustains streamflow in
dry seasons. Intermittent flows result from the timing of water uses (by crops and
riparian vegetation) along the stream and from the climatic regime (Putrnan et al.

1988).

2.4.2 Groundwater and subsurface regime

Roeske and Werrell (1973) report that artesian conditions generally exist for
wells penetrating to depths greater than 200 feet. The majority of recharge to the
subsurface regional aquifer occurs along mountain fronts and a small portion via
stream channel infiltration. Direct infiltration over the valley floor is considered
negligible because of high evaporation and low precipitation rates, and, in general,
no groundwater communication exists between adjacent basins (Freethey 1982).

A secondary aquifer exists in shallower floodplain aquifers, deposited above the
valley fill sediments by the San Pedro River and its tributaries. The most important
floodplain aquifer in the Upper San Pedro Basin bounds the San Pedro River.
Recharge to the floodplain aquifer derives from streamflow, upward leakage and
lateral flows from the regional aquifer, agricultural return flows, runoff water

percolation, and underflow across the U.S.-Mexico border (ADWR 1990). In

29

addition to serving as the main conduit for surface water/ ground water interaction,
the floodplain aquifer provides necessary water for phreatophytes (deep-rooted
plants which obtain water from the underlying aquifer) within the riparian zone of
the inner valley and is the major supplier for irrigation wells along San Pedro River
and its principal tributaries (Putman et al. 1988). Groundwater communication
between the regional aquifer and floodplain aquifer is generally from the regional
aquifer to the central floodplain aquifer, but this may reverse in times of low river
flow. Flow between the floodplain aquifer and the San Pedro River is toward the

San Pedro River in perennial reaches (baseflows).

2.4.3 Natural and anthropogenic demand for water

The most important natural depletion of the San Pedro River and its tributaries
is evaporation from bare soils in and along stream courses (Putman et al. 1988).
Water that evaporates from alluvial soil depletes near-surface water storage and
provides a sink for infiltrating water from streams, thereby decreasing subsequent
surface flows and recharge rates to the regional aquifer. Evapotranspiration by
phreatophytes, including several species of cottonwood, Gooding willow, Seep
willow, mesquite, and salt cedar, comprises the single largest natural consumptive
use of water in the Upper San Pedro Basin (Putman et al. 1988). The effect of
phreatophyte consumption of groundwater is similar to that of bare soil evaporation.
Water that is used by the plants is replaced by surface water infiltration. Hence,
that volume of infiltrated water is no longer available in streamflow or for recharge

to the aquifer system. Phreatophytes comprise most of the riparian vegetation and

30

depend largely on the floodplain alluvium for their water supply. Their roots either
extend to the water table or to the saturated capillary fringe just above the water
table (Brown and Lowe 1980).

The Sierra Vista subwatershed contains approximately 6,490 acres of land
irrigated almost exclusively from groundwater wells (Lacher 1994). About 4,610 of
these acres (71 percent) are cultivated as pasture, and the remaining lands are either
inactive croplands or crops that include grapes, pecan, and fruit tree orchards as
well as vineyards (ADWR 1990). Domestic well use increased rapidly from 19505
onward, when high volume pumps became widely available. Most urban and
industrial wells tap the regional aquifer. Most irrigation wells pump from the
floodplain aquifer near the San Pedro River, although some do pump from the
regional aquifer (Lacher 1994).

Water use in the Upper San Pedro Basin is concentrated in two zones and serves
two primary purposes (ADWR 1990): 1) the Sierra Vista-Fort Huachuca area well
field supplies groundwater for municipal, military, and industrial uses. These wells
in the regional aquifer are up to 1500 feet deep; 2) the Palominos-Hereford area
well field supplies water for agriculture. Some of these wells penetrate the regional
aquifer, but most are shallow and only penetrate the floodplain aquifer. Small,
isolated wells outside these two areas pump just enough water for domestic and

livestock use (V ionnet and Maddock 1992).

31

2.5 GRAZING SYSTEMS IN SOUTHEASTERN ARIZONA
2.5.1 Yearlong grazing

Yearlong grazing is a system in which a pasture is grazed “constantly” for all 12
months of a year. That pasture may or may not be grazed in subsequent years and
supplemental feeding may or may not be required for portions of the year that it is
being grazed. This system is more efficient the more uniform the distribution of
forage plants, the better the distribution of water, and the more diverse the plant
communities (Kruse and Jemison 2000). In the past, because the range was
essentially an “Open range”, this grazing method resulted in the inability to
determine grazing capacity for introduced domestic livestock, the inability to
respond to adverse weather conditions (severe drought), and the continuous and
rapidly increasing animal numbers set the stage for severe damage in semi-arid

rangelands.

2.5.2 Continuous grazing

Yearlong grazing of an area (pasture or range) for a 12-month period is not
necessarily synonymous with continuous grazing (Society of Range Management
1998). Continuous grazing refers to grazing the same area throughout a year or that
part of the year during which grazing is feasible. It may be yearlong or shorter
depending on environmental or other restrictions to grazing by livestock. On
annual ranges, Ratliff (1986) showed that cow and calf weights responded more
productively on continuously grazed annual grassland range than with repeated

seasonal or rotated seasonal grazing. Grazing the same area year after year

32

provides the manager with a systematic advantage of using the pasture when it is
optimal for grazing; it requires less handling and moving of livestock; and it may
lessen the need for fencing and water development (Kruse and Jemison 2000).
Often, however, trying to balance plant development with animals’ selective
grazing provides an uneven grazing distribution of palatable plants, resulting in an
overall deterioration in the quality of the prime forage plant community. Ultimately

this creates a less than desirable overall range condition.

2.5.3 Seasonal grazing

Seasonal grazing is restricted to one or more specific seasons of the year
(Society of Range Management 1998) and is that portion of the year during which
grazing is most feasible for livestock production. Ratliff (1986) describes the
forage seasons as determined by environmental conditions that limit, or promote,
plant growth. During the forage production season, plant growth can be so
accelerated that livestock cannot use all the forage produced, assuming an adequate
growing season. Protein and energy supplements are generally not necessary in
these situations. On the other hand, during the dry season when plant growth is
usually limited, there can an insufficient quality of green forage for livestock needs
and protein supplements are often supplied. Seasonal grazing is more appropriate
in high mountain meadows, transitional zones (spring/fall ranges), and areas where
the livestock are trucked in to utilize the forage at peak production and condition,
and range managers develop stocking plans based on annual environmental events

such as seasonal weather patterns for forage plant grth and development (Kruse

33

and Jemison 2000). Predicting whether an adequate amount of precipitation will
provide an adequate amount of forage for the number of livestock of a specific class

within a given year can be difficult.

2.5.4 Deferred grazing
The discontinuance of grazing or deferred grazing by livestock on an area in a

specified period (e.g., during the growing season) promotes plant reproduction,
establishment of new plants, or restoration of vigor to older plants (Kruse and
Jemison 2000). The term often suggests cattle movement but it actually relates
more specifically to plant grth and development. Grazing is deferred for some
range management objective (e. g., until after seed maturity or the establishment of
new plants) and more than not, is used in combination with rotation grazing.
Because it is the forage plant community that is being focused upon, grazing can be
deferred for values other than livestock production (e.g., wildlife habitat). For
range management Objectives, deferment is usually done to enhance the forage
plants within that pasture, but could just as well be done to improve the habitat of a
specific wildlife species. For example, research indicates that a deferred grazing
system can be used to maintain and heighten riparian vegetation including willows
and cottonwoods that provide food, cover, and nesting areas for various bird species

(Clary and Webster 1989).

34

2.5.5 Deferred rotation grazing

This system discontinues grazing on various parts of a range, allowing each part
to rest successively during the growing season to permit seed production,
establishment of seedlings, or restoration of plant vigor. Rotation and deferred
grazing combine to produce a pasture rotation situation, which also can be termed a
“deferred grazing system” in that the livestock are rotated among a set of pastures,
and grazing commences on one while grazing has just been deferred from another
(Kruse and Jemisen 2000). The deferment may be necessary to strengthen the
forage plant community, and there must be an available pasture. In this case, both
the animals and the pasture benefit. Although similar, the deferred rotation grazing
system is an improvement over the deferred grazing system in that another pasture
or two are available. An ideal situation for a deferred rotation grazing system in a
riparian area when wildlife habitat protection is desired would be to have two or
three riparian pastures, one of which does not contain nesting habitat for the target
species (Kruse and Jemison 2000). In this situation, livestock moving into it

benefits the receiving pasture, while the deferred pasture is being rested.

2.5.6 Rotational deferment

In this system one or more parts of the range are rested during the growing
season each year, and use on other segments of the range are rotated. This system
differs from deferred rotation because the livestock are not being rotated
systematically and, therefore, where they are being rotated is less important. It is

the rotation deferment of grazing that is the key. Generally, three or four pastures

35

are required. As with other deferred grazing systems the primary objectives are
improvement of plant utilization, overall range improvement, and maintenance of
livestock production. This system can be utilized more easily to accommodate
wildlife and other values than other systems because it provides more management
options (Clary and Webster 1989). The system functions better when livestock are
not allowed to physically degrade site conditions to such a degree that they have to

be removed. Livestock removal, for the expressed purpose of habitat regeneration,
may involve a time factor that is not in the best interests of livestock production.
However, with a set of pastures available for grazing, all without constraints on use,

rotational grazing of the pastures would offer the most options.

2.5.7 Rest-rotation grazing
In this intensive system of management, grazing is deferred on various parts of

the range during succeeding years, allowing the deferred part complete rest for one
year. Two or more units are required and control by fencing is usually necessary on
cattle range (Kruse and Jemison 2000). Rest can mean deferred grazing or
ungrazed pasture, but in combination with the word rotation it is more appropriately
assigned to a management decision rather than to the grazing animals. Therefore, it
is the loss or lack of grazing that allows a pasture to be rested, and that rested
condition rotates among other pastures or areas of the range during succeeding
years. Resting a pasture for an entire year allows plants to complete a full year
growth cycle without interruption (Ratliff 1986). Altering grazing systems can

have long-term beneficial effects. Hughes (1998) showed that overgrazed BLM

36

allotments that had previously been grazed season-long or with continuous grazing
systems by sheep and cattle responded beneficially following 30 years of rest
rotation grazing in the Mojave Desert of California. In this example, rest appeared

to be the dominant factor while rotation provided grazing options and alternatives.

2.5.8 Rotation grazing system
An orderly sequence of use, when each subdivision is both grazed and deferred

during the same grazing season, is called rotation grazing. This generally refers to
animals being rotated among pastures or to the next pasture in sequence. These
moves are systematic in that the animals are rotated to a new pasture because the
forage in that new pasture is proper for grazing. The pasture from which the
livestock are removed is deferred from grazing. Grazing could commence
systematically, with rotation being initiated by growth and development patterns of
the forage plants within each newly entered paddock (Kruse and Jemison 2000).
An important management aspect of rotation grazing is utilizing a number of
paddocks or subsets of a larger pasture unit, which enhances grazing distribution.
The strength of a rotational grazing system is its flexibility. By utilizing
unbalanced paddocks, grazing could more easily be deferred for reasons other than
grazing or plant development, e.g., wildlife protection and recreation. Therefore,
while even a few paddocks provide for greater flexibility and optional diversity, the

greater the number of small paddocks, the more the diversity and flexibility.

37

2.5.9 Best pasture system

Based on research trials in New Mexico, Herbel and Nelson (1969) advocated a
highly flexible system that provides for always moving to the best pasture in the
system: thus, the “best pasture system”. McCulley (1968) used a comparable
system that he referred to as “repeated seasonal” grazing in trials that he conducted
at the San Joaquin Experimental Range in California. Although these grazing
schemes are not new, they provide the flexibility of rotating between several
paddocks within grazing units, range types, subtypes, or condition classes when it is

most advantageous to livestock, vegetation, or both (Holechek et al. 1998).

2.5.10 Holistic management, holistic resource management, or the Allan
Savory grazing method
According to Savory and Parsons (1980), any of the aforementioned systems
can be interwoven into this method. While Savory suggested that America’s
rangelands (southwestern ranges in particular) were understocked by livestock and
at the same time overgrazed, he was introducing a rotation grazing system that
many range scientists and managers had been looking at and studying for some time
already. This system concluded:
- Continuous, yearlong grazing at moderate rates produces the best
individual animal performance, but range condition declines below its

potential.

38

Range condition can be maintained at an acceptable level if livestock
remove less than 40 percent of the current year’s grth during the
growing season.

Three-herd, four-pasture deferred-rotation grazing at a moderate stocking
rate allows slow range improvement with good individual animal
performance.

High-intensity, low-frequency grazing with six to eight pastures and one
herd provides for faster range improvement, but animal performance
suffers presumably because of the high stock density.

High-intensity, low-frequency grazing systems might produce acceptable
animal performance if grazing deferment periods are shortened.

Increased stocking of 50 to 100 percent can be achieved only by using
some form of range improvement, such as brush control, fertilizer, or

water-spreading.

Savory’s grazing management was holistic in philosophy. It included (a) using

one herd, (b) shortening the grazing and rest periods, (c) managing plant grth

through better harvest efficiency at proper grazing intervals, and (d) applying hoof

and herd impact to alleviate soil surface crusting (Savory and Parsons 1980).

Savory’s animal grazing model gained popularity with many ranchers and some

land managers. His management philosophy is the most popular and widely

practiced among ranchers in southeast Arizona (Marsett 2000). After widespread

acceptance among cattle ranchers, he has broadened his philosophical concepts of

holistic management to encompass more challenging management objectives, such

39

as biodiversity and species variability, on public lands. His model continues to be
controversial among much of the academic community because of its lack of

scientific hypothesis testing.

2.6 THE ROLE OF FIRE

Fire is a pervasive and powerful force in desert grasslands. Its importance in
controlling ecosystem structure and function rivals that of precipitation (Kimmins
1987). As the frequency, season, and behavior of fires have shaped plant
communities, communities have in turn shaped the fi'equency, season, and behavior
of fires. The long-term fire regime is probably more a consequence than a cause of
vegetation patterns; that is, vegetation probably affects fire regime to a greater
extent than fire regime affects vegetation, at least at a coarse level of resolution
(Clark 1990). Nonetheless, as an integral component of rangeland ecosystems, fire
should not be viewed as external to these systems; rather, fire is part and parcel of
community organization and development.

Perhaps nowhere is the role of fire more widely acknowledged than in
grasslands. In fact, some North American researchers have proposed that treeless
grasslands were a product of repeated fires set by Native Americans (Sauer 1944;
Stewart 1951). Fire interacts with ecological factors including topography, soil,
insects, rodents, lagomorphs, and herbaceous plants to restrict woody plant
establishment in grasslands. There is general agreement that fire is necessary
(though usually not sufficient) to control the abundance of woody plants and

maintain most grasslands (Grover and Musick 1990; Wright and Bailey 1982).

40

2.7 CONCLUDING REMARKS

The discussion on greater Sierra Vista’s land use history is intended to convey
the important impact of human activities on the regional land cover, and how
different perceptions and uses of the land have affected the grasslands, especially in
the last 100 years. The resulting landscape fragmentation from human activities
may influence the extent of natural processes like fire and water drainage patterns
as well as the dispersal of grassland plants and animals.

The beneficial role grasslands play in both the economical and environmental
sustenance of rangelands is vital to understanding the issues affecting human and
environment interactions in the region. Grasslands play a vital role in maintaining
ecosystem health, balance, and maintenance. Their beneficial characteristics such
as increasing infiltration, decreasing surface runoff and sediment delivery,
rectifying polluted streams, and serving as cover for wildlife, are a few examples of
the importance of grasslands for environmental protection. And, in addition to
providing the main source of income to the region, the value of grassland is not
only the forage and livestock but also the water, timber, minerals, recreation, and
wildlife.

Hydrological resources and water demand are vital and critical variables in land
use and land cover issues in arid and semi-arid ecosystems. Often, large areas of
farmland revert back to rangeland because of increased irrigation costs, 1055 of
water to urbanization, and lowering of the water table. However, as water depletion

continues, there is a grave danger that rangelands, in turn, could revert to desertlike

41

conditions, largely through human actions, in fragile areas that do not have desert
climates, e.g., the semi-arid region of Sierra Vista.

Livestock grazing is an important human-controlled activity that has
considerable potential to either improve or degrade vulnerable ecosystems.
Techniques developed to utilize grass cover for livestock production are critical in
maintaining biodiversity, promoting healthy range cover, and sustaining profitable

livestock production.

42

CHAPTER 3—METHODS

All images in this thesis are presented in color. A graphical representation of
methods and procedures used in GIS and remote sensing integration is displayed in
Figure 3. The first section of the chapter outlines sources, metadata, characteristics,
processing procedures, use of ancillary data, along with co-registration and
classification procedures of remotely sensed data. The second section discusses
GIS coverage processing, as well as the data structure and spatial data model for the
integration of a well registry database with a GIS as a tool for observing patterns of
water usage. The third section examines the sources and characteristics of census
data. Temporal, spatial, spectral, and classification limitations are discussed in the
fourth section. Integration of GIS and remote sensing, overlay analysis, and the
magnitude procedure code are examined in the fifth section. Finally, the section on
the use of data sources to infer human-environment relationships concludes the

chapter.

3.1 IMAGE COLLECTION AND PROCESSING

Remote sensing derived information is critical to the successful modeling and
monitoring of numerous natural and cultural processes (Jensen 2000). For this
analysis, as typically is the case with most long-term, multi-temporal land use and
land cover studies, images from different sensors at various scales and resolutions

were required for assessing land use and land cover change.

43

Although additional images would have been desirable for a more
comprehensive time series analysis, due to cost limitations and image quality, a
total of four images were selected for the years 1973, 1985, 1992, and 1999. The
source of the images was the Arizona Regional Image Archive located at the
University Of Arizona, Tucson. All the images were scanned during the growing
season and were relatively close to the maximum green-up time in the region
(usually late August through early October). The dates of acquisition were
September 5 1973, September 27 1985, September 30 1992, and October 12 1999.
Figures 4 through 7 display the image data in sequential order. To aid in
summarizing system and optical characteristics of these images, Table 1 lists the
sensor systems and their major characteristics, and Table 2 lists the spectral

wavelengths associated with each sensor.

3.1.1 Atmospheric correction
The atmosphere influences the amount of electromagnetic energy that is sensed

by the detectors of an imaging system, and these effects are wavelength dependent
(Slater et al. 1983). This is particularly true for imaging systems such as the
Landsat Multispectral Scanner (MSS), Thematic Mapper (TM), and Enhanced
Thematic Mapper Plus (ETM +) that record data in the visible and near-infrared
portions of the electromagnetic spectrum. The atmosphere affects images by
scattering, absorbing, and refracting light. Various methods to remove the effects
of scattering from remotely sensed data have been developed. However, a problem

with most of these techniques is that the haze values for each spectral band are

44

selected independently. This can create problems because atmospheric scattering is
highly wavelength-dependent in the visible part of the spectrum and the scattering
values are correlated with each other (Chavez 1988). Chavez’s improved dark
object subtraction (DOS) technique was utilized for atmospheric correction of the
1973 M88 image since this technique allows the user to select a relative
atmospheric scattering model to predict the haze values for all the spectral bands
from a selected starting band haze value, thus the method normalizes the predicted
haze values for the different gain and offset parameters used by the imaging system.
According to the metadata, the 1973 image quality was high and the image was
acquired on a very clear day with no cloud cover. As such, I used a relative
scattering model for a very clear atmosphere. By utilizing band 1(green) as the
starting band, the improved DOS atmospheric correction procedure generated the
predicted and final DN values for haze correction as listed in Table 3. The final DN
values derived from Chavez’s improved DOS procedure were subtracted from the
DN values from each corresponding band of the original image to correct for
atmospheric scattering.

For the TM images of 1985, 1992, and the ETM+ image of 1999, a location-
specific correction method developed to utilize within-image targets was used. This
method, called the refined empirical line (REL) approach, requires only one within-
scene calibration target, minimal field measurements of that target, and a reasonable
estimate of at-satellite radiance (or DN) for a surface reflectance factor (the ratio of
directional reflected and incident radiation at the surface within a spectral band) of

zero (Moran et al. 2000). The estimate of DN for surface reflectance factor of zero

45

is based on an empirical relationship between DN and the surface reflectance factor.
It is derived by the use of the Herman and Browning atmospheric radiative transfer
model (RTM). The implementation of the REL approach was based on data
derived from the Walnut Gulch Experimental Watershed located inside the Sierra
Vista census division study area. The Landsat TM and ETM+ DNs for surface
reflectance factor of zero at Walnut Gulch, estimated fi'om a RTM with inputs
based on on-site measurements of atmospheric conditions taken during the growing
season from June to September, were used for atmospheric correction in TM2,
TM3, TM4, and TMS hands. This procedure was performed independently on each

image.

3.1.2 Geometric correction

The 1973 M88 image was located on path 37, and row 38 of the World
Referencing System One (WRS-l) used by Landsat satellites. The 1985, 1992, and
1999 TM and ETM+ images were located on path 35, and row 38 of the World
Referencing System Two (WRS-2). For Landsat and other satellite images the
systematic distortions, i.e., geometric distortions whose effects are constant and can
be predicted in advance such as scan skew, variations in scanner rrrirror velocity,
and cross-track distortions, are corrected before the data are distributed. In
addition, random or nonsystematic distortions, i.e., distortions caused by variations
in the spacecraft attitude, velocity, and altitude, were corrected, according to
metadata from Arizona Regional Image Archive, by the cubic convolution method

of resampling before distribution. However, all images contain inherent geometric

46

distortions because they record the curved surface of the earth on a flat display.
Areas, distances, and angular relationships are distorted to varying degrees. To
correct the inherent distortion, all images, except the 1999 image that was projected
before distribution, were projected to the Universal Transverse Mercator (U TM)
projection, zone 12. For all images, the NAD83 datum, which was designed to fit

the surface well in the North American region, was specified.

3.1.3 Image co-registration

Image-to-image registration is the translation and rotation alignment process by
which two or more images of like geometry and of the same geographic area are
positioned coincident with respect to one another so that corresponding elements of
the same ground area appear in the same place on the registered images (Chen and
Lee 1992). This type of correction is necessary for multitemporal remote sensing
analysis as it promotes the comparison of two or more images obtained on different
dates to see if any change has taken place in them.

Since the 1999 image was orthorectified before distribution, it was used as the
rectified scene to register the 1973, 1985, and 1992 images. Before registration, the
1973 M88 image spatial resolution was adjusted from 80 to 30 meters (TM and
ETM+ ground resolution) by resampling. For each rectification, roughly 30 ground
control points were selected at easily identifiable locations such as road
intersections, edges of fields, as well as the intersections of utility corridors. The
precision and accuracy of co-registration were enhanced by utilizing ERDAS

remote sensing software that uses a chip extraction algorithm that zooms in and

47

performs subpixel sampling. The root mean square (RMS) error, a measure of
distortion of each control point, for all the GCPs was acceptable at less than the
threshold value of 0.05 pixel, and consequently, precision error was less than one
pixel size of 30 meters.

After all the images were co-registered, a GIS coverage of the study area, i.e.,
the census division of Sierra Vista consisting of tracts 14 through 20, was used to
clip the original images to create a multitemporal set of land cover data bounded by

the census division extent.

3.1.4 Image classification

Remotely sensed data of the Sierra Vista census division was analyzed to
extract useful thematic land cover information. The multispectral classification was
performed using the supervised algorithm approach. Known a priori land cover
types were identified by combination of previously published vegetation maps,
previous fieldwork data, recent personal field observations (utilized only for the
1999 image classification), and personal interviews with US. Department of
Agriculture personnel with over 30 years of classification experience in the region
(Marsett 2000). The specific sites in the remotely sensed data represented
homogeneous examples of these known land cover types. These sites were
collected by on-screen selection of polygonal training data, and conversely, by
seeding an x, y location specific to a cover type in the image space using the cursor.
In addition, feature space plots depicting the distribution of all the pixels in the

scene using red and NIR bands were especially helpful in separating mixed

48

mesquite, oak woodland, and riparian classes. Automated in ERDAS software,
spectral characteristics of the training sites were evaluated, multivariate statistical
parameters were calculated in each training site, and every pixel both within and
outside these training sites was evaluated and assigned to the class of which it had
the maximum likelihood of being a member.

Before the supervised classification algorithm was run on the atrnospherically
and geometrically corrected raw images, image data were merged with topographic
information. A 1990 digital elevation model (DEM) was registered to the images to
aid in montane forest and oak woodland mapping in the Huachuca Mountains
region of the scene. In this region, species that have very similar spectral
characteristics (e.g., mesquite, aspen, and evergreens) occupy quite different
elevation ranges. Mesquite is commonly found in lower elevations whereas aspen
is typical of higher terrain. Use of ancillary DEM data as a discriminatory tool thus
aided in defining and classifying the various associations between these land cover
types present in the scenes and their habitats.

The classification scheme is based on the resource-oriented USGS Land
Use/Land Cover Classification System Level I (Anderson et al. 1976) with slight
modifications (addition of mixed mesquite and desertscrub, and subtraction of
tundra and ice) to accommodate the native vegetation of the region. Following is a

name and description list of land cover classes used in the classification:

49

 

Forest

Vegetative communities comprised principally of trees potentially
over 10m in height and typically characterized by closed or multi-
layered canopies. Species in this category are evergreen (with the
exception of aspen), largely coniferous (e.g., ponderosa pine), and
restricted to the upper elevations of mountains that arise off the
desert floor.

 

Oak Woodland

Vegetative communities dominated (>3 0% total cover) by
deciduous trees (Quercus spp.) with a mean height usually between
6 and 15m. Tree canopy is usually open or interrupted and
singularly layered. This cover type often grades into forests at its
upper boundary and into semi-arid grassland below.

 

Mixed

Vegetative communities composed of leguminous trees (Prosopis
spp.) whose crowns cover 15% or more of the ground often
resulting in dense thickets (30-75% total cover). Historically
maintained maximum development on alluvium of old dissected
flood plains; now present without proximity to major
watercourses. Winter deciduous and generally found at elevations
below 1,200m.

 

 

Grasslands

 

Vegetative communities dominated by perennial and annual
grasses (>3 5% total cover) with occasional herbaceous species
present. Trees and shrubs do not exceed 20% of the total cover.
Generally grass height is under 1m and occur at elevations
between 1,100 and 1,700m, sometimes as high as 1,900m. This is
a landscape largely dominated by perennial bunch grasses
separated by intervening bare ground (45-50% total cover) or low-
growing sod grasses and annual grass with a less-interrupted
canopy. Serrri-arid grasslands are generally positioned in elevation
between evergreen woodland above and desertscrub below.

 

50

 

 

Desertscrub

Vegetative communities comprised of short shrubs with sparse
foliage (>3 5% total cover) and small cacti that occur between 700
and 1,500m in elevation. Within the San Pedro river basin this
community is often dominated by one of at least three species, i.e.
creosotebush, tarbush, and whitethorn acacia. Individual plants are
often separated by significant areas of barren ground (40-45%
total cover) devoid of perennial vegetation. Many desertscrub
species are drought-deciduous.

 

Riparian

Vegetative communities adjacent to perennial and intermittent
stream reaches. Trees can potentially exceed an overstory height
of 10m and are frequently characterized by closed or multi-layered
canopies depending on regeneration. Species within the San Pedro
basin are largely dominated by two species, i.e. cottonwood and
Goodding willow. Riparian species are largely winter deciduous.

 

Agriculture

Crops actively cultivated (and irrigated). In the San Pedro River
basin these are primarily found along the upper terraces of the
riparian corridor and are dominated by hay and alfalfa. They are
minimally represented in overall extent (less than 3% total cover)
within the basin and are irrigated by ground and pivot-sprinkler
systems.

 

Urban

This is a land-use dominated by small ejidos (farnring villages or
communes), retirement homes, residential neighborhoods,
commercial buildings, industrial sites, as well as airstrips and
buildings inside Fort Huachuca.

 

Water

Sparse free-standing water is available in the watershed. This
category would be mostly represented by perennial reaches of the
San Pedro and Babocomari rivers with some attached pools or
repressos (earthen reservoirs), ponds near recreational sites such as
parks and golf courses, and sewage treatment ponds east of the city.

 

Barren

A cover class represented by large rock outcropping, abandoned
mines (including tailings), and other surfaces that are largely absent
of above-ground vegetation.

 

 

Clouds/Shadows

 

1985, and 1992 images.

 

51

 

The classified images represent categorized pixels in an image representing the
11 land cover classes discussed above. These images covering the years 1973,

1985, 1992, and 1999 are displayed in Figures 8, 9, 10, and 11, respectively.

3.2 GIS DATA

A political boundary coverage consisting of the seven census tracts comprising
the Sierra Vista census division was utilized for analyzing the human influence on
the local semi-arid ecosystem. The coverage was obtained from the USGS National
Biological Information Infrastructure (NBH), a national biological information

partnership.

3.2.1 Coverage processing
The coverage, geocoded before distribution, was imported into Arc/Info,
topologically cleaned to remove gaps at nodes and breaks in arcs defining the tract

polygons, and projected to UTM zone 12 using NAD83 datum (Figure 12).

3.2.2 Attribute data structure

The well attribute data dictionary, obtained from the Arizona Department of
Water Resources, was entered into a GIS by (1) creating a new INFO data file to
hold the attributes, (2) adding the attribute values to the newly created INFO data
file, and (3) relating the attributes in the INFO data file to the polygon attribute

table of the census coverage.

52

3.2.3 Spatial data model

The census division coverage served as a logical construct for the storage and
retrieval of water usage data relating to land use and land cover change. This
simple model consisted of seven census tracts with a data dictionary of attributes

from numerous wells located in the census division.

3.2.4 GIS queries

Some of the important variables of water usage in the greater Sierra Vista are
the number of wells, well depth, water level, and primary well usage. The ADWR
(1990) reported 5,298 registered wells on file with the ADWR in the Sierra Vista
census division. This number increased by 1031 to 6,329 by the year 2000. The
year 2000 data included geo-referenced UTM coordinates for each registered well
in the division. The locations and the total number of registered wells for each
census tract are included in Figure 13.

The 2000 data dictionary for the wells point coverage provided input for simple
GIS queries that calculated average well depths, average water levels, and usage

percentages of registered wells for each census tract.

3.3 SOCIOECONOMIC DATA
Every effort was made to collect demographic data as close as possible to the

years of image acquisition. The sources of census data used in this study are listed

in Table 4.

53

Table 5 lists the population and housing data for the 1970, 1980, and 1990
census for Cochise County and the Sierra Vista census division. Tract data for the
Sierra Vista census division were only available for 1990 and 1997, and are listed in
Table 6. New tracts for the census division were designated by the Census bureau
for the 1990 census due to the influx of population during the 19805.

County business patterns were collected to augment population and housing
data for detection of urbanization patterns. The number of businesses and taxable
payrolls per economic sector were collected from the same years of image
acquisition, and are listed in Table 7.

Table 8 is data collected from the agricultural sector of Cochise County.
Specifically, farm-level information includes number of farms, total area, average
size, average market value, and percent of land area in the county used for farming
activities. In addition, county totals of different farming systems, i.e., cropland,
cropland pasture, woodland pasture, pastureland farms, in addition to number of
cattle farms and cattle totals are listed to monitor land use change. Irrigated farms

and irrigation totals in acre-feet are included for detection of water usage patterns.

3.4 DATA LIMITATIONS

Several data limitations were discovered during data collection and processing.
Temporal constraints of remote sensing and demographic data, spatial limitations of
designated census areas, spectral limitations and choice availability of remote

sensing data, and classification limitations were the major topics of concern.

54

3.4.1 Temporal limitations

Due to budget constraints, lack of low-cost availability of remote sensing data
was a limitation to a more broad time series analysis. Because image acquisition
during the peak green-up time was essential to differentiate cover types (e.g.
mesquite vs. riparian, grass vs. scrub, etc.), only images taken from late August
through early October were sought. Available low-cost images acquired during
these times of year were limited to 1973, 1985, 1992, and 1999.

Census of population and housing were only applicable to years 1970, 1980,
and 1990. Obviously, data for the evaluation of spatial variations and shifting
patterns in distribution of population and housing in 10-year periods between
census tabulations were not available for analyzing human influence on the
ecosystem. Similarly, except for 1992, agricultural statistics did not coincide with
the years of image acquisition. County business patterns were available for every
year; therefore this data source did not pose any temporal constraints to the

analysis.

3.4.2 Spatial limitations

As mentioned, the census tracts 14 through 20 for the Sierra Vista census
division were implemented starting with the 1990 census. Pre-1990 demographic
data were available only for the census division without any delineation of tract-
level spatial coverage. Therefore, the analysis was limited to examining population
and housing units at only the larger-area census division for 1973 and 1985. For

1992 and 1999, population and housing units were available at tract level. In

55

addition, agricultural statistics and economic business patterns were available only
at the county level. However, because Sierra Vista is the major economical and
agricultural hub of the southeastern Arizona region, the majority of exchange of
goods and services, as well as sales of agricultural products take place within the

census division (ADWR 1990).

3.4.3 Spectral limitations

Two important temporal resolutions should be held constant when performing
change detection using multiple dates of remotely sensed data. First, the data
should be obtained from a sensor system that acquires data at approximately the
same time of day. Landsat MSS, TM, and ETM+ data were acquired around 10:30
AM. for the study area. This eliminated diurnal sun angle effects that can cause
anomalous differences in the reflectance properties of the remotely sensed data.
Second, whenever possible it is desirable to use remotely sensed data acquired on
anniversary dates (same dates of the year for each image). Using anniversary date
imagery removes seasonal sun angle and plant phenological differences that can
destroy a change detection analysis (Jensen et al. 1994). Although anniversary
dates of images in the study were not identical, sun angle and phenological
differences are minimal since the maximum date difference between the images was
a matter of weeks.

Ideally, remotely sensed data are acquired by the same sensor system that
collects data with the same instantaneous field of view (IFOV) on each date.

However, the M88 sensor has a different IFOV (resulting in 80-meter spatial

56

resolution) than TM and ETM+ sensors’ IF OV (30-meter spatial resolution).
Therefore, it was necessary to resample the M88 image to the uniform 30-meter
pixel size. This does not present a significant problem as long as it is recognized
that the information content of the resampled data can never be greater than the
ground resolution of original sensor system.

Different sensor systems do not record energy in exactly the same portions of
the electromagnetic spectrums (i.e., bandwidths). For example, the Landsat MSS
sensor records energy in four relatively broad multispectral bands, and the TM
sensor in six relatively narrow optical bands and one broad thermal band.
Therefore, only bands that approximated each other across the three sensors were
selected for change detection. These included bands green, red, and NIR of the
M88 sensor, and green, red, NIR, and SWIR bands of TM and ETM+ sensors (see
Table 2). In order to improve the accuracy of the training sites, however, a
hyperspectral sensor consisting of numerous narrow bandwidths capable of
precisely differentiating plant phenological cycles and numerous urban and non-
urban cover types would have been preferred for better classification outputs.

An analog-to-digital conversion of the satellite remote sensor data usually
results in 8-bit brightness values ranging from 0 to 255. Ideally, the sensor systems
collect the data at the same radiometric precision on different dates. Unfortunately,
the radiometric resolution of the early MSS sensor was only 6-bit (as compared to
8-bit TM and ETM+ radiometric resolution), and therefore had to be decompressed
to 8 bits for change detection purposes. However, the precision of decompressed

brightness values can never be better than the original, uncompressed data.

57

Finally, the 1985 and 1992 images exhibited slight cloud cover and resultant
shadow effects, although the cloud cover was much less than the prescribed upper
limit of 20 percent of the scene (Jensen et a1. 1994). Consequently, cloud cover and

shadows had to be masked out before supervised classification was performed.

3.4.4 Classification limitations

Classification accuracy assessment of remote sensing-derived land use/land
cover maps and associated area totals per census tract would ideally require
comparison of two sources of information: (1) the remote sensing-derived
classification and (2) reference test information. For this study, it would have been
desirable to utilize Digital Orthophoto Quadrangles (DOQs) in assessing the
classification accuracy of land cover maps derived from Landsat M88 and TM data.
The cost of such an endeavor was prohibitive, however, because, based on previous
land use studies of the region (Kepner et a1. 2000), in addition to obtaining the
orthophotos, it is recommended that the suitability of DOQs in distinguishing
between different land cover classes must be assessed using high-resolution
airborne color video data at further cost. Although it is desirable to collect the
reference information as close to the date of data acquisition as possible, DOQs for
the 1973, 1985, and 1999 images were not available, and the DOQs for 1992 were
available in either early April or late October. These time periods are undesirable
for interpretation because they are too distant from the maximum green-up time
when discrimination of vegetation cover of the region would be most confidently

assessed. Because reference information is required but was not available for

58

accuracy assessment of supervised classification, several important classification
evaluation statistics including the KHAT statistic, which is the measure of the
difference between actual agreement of reference data and automated classifier and
the chance agreement between reference data and a random classifier, as well as
producer’s accuracy, and user’s accuracy could not be calculated. However,
because known a priori land cover types were identified by a combination of
previously published land cover maps, personal observations, previous fieldwork
data, and interviews with USDA classification experts of the region, I am confident
that the overall accuracy of the classifications is no less than 80 percent. In
addition, deriving human drivers of land use and land cover change typically
depends on detecting changing trends of land cover and land use, rather than a
precise quantitative assessments of land cover maps derived for more systematic
vegetation analyses (Jensen 2000). Further, because accounts and trends of land
use and land cover change in the region were similar in detail to the results of
classification maps produced, such as decreasing grass cover and increasing urban
and mixed mesquite cover (Bahre and Shelton 1993), the classified maps were

deemed appropriate for land use and land cover change analysis.

3.5 INTEGRATION OF REMOTE SENSING AND GIS
The following subsection discusses conversion of image data to GIS vector
coverage, overlay analysis, and magnitude procedures built for generating land

cover area totals per census tract for all years of image acquisition.

59

3.5.1 Image to coverage procedures

In order to discriminate census tract area totals of land cover change from one
classified image to the next, a GIS procedure using Arc/Info’s programming
language Arc Macro Language (AML) was developed to sum land cover totals per
census tract.

The first step in the procedure consisted of converting the classified images to
vector data models. Using this format, feature boundaries are converted to straight-
sided polygons that approximate the original regions. Vector data formats have the
advantages of relatively lower data volumes, better spatial resolution, the
preservation of topological data relationships, and better overlay searching
capabilities. In addition, because the census tract coverage was also in a vector
format, a consistent data model was necessary to spatially interrelate multiple layers

of vector coverages.

3.5.2 Overlay analysis
After all the classified images were converted to vector coverages, the census

tract coverage and classified land cover coverage were interrelated by an overlay
analysis. In a map overlay, a new map was created that shared the space division of
both source maps. Every new polygon created on the map has a new attribute
record in an expanded attribute database associated with the map overlay. This
means that the overlay map is searchable by either of the sets of regions used to
create it. This capability was deemed necessary for summing land cover totals on a

tract-by-tract basis.

60

3.5.3 Magnitude procedure

Spatial magnitude of land cover classes per census tract was calculated by
writing a GIS procedure using Arc/Info’s AML. The code for the procedure is
listed in the Appendix A. The first step in the procedure consists of a frequency
function that calculates land cover type area totals per census tract polygon. These
totals are then related to an INFO file that consists of only census tract polygon
information (tract ID’s and tract area totals). Classified cover totals are then sorted
and summed up per cover type and checked against census tract area totals. Finally,
a data file is written that lists land cover totals per census tract. The procedure was
performed for each year of image acquisition. Land cover area totals per census

tract are listed in Figures 8 to 11.

3.6 INFERRING HUMAN-ENVIRONMENT RELATIONSHIPS
The integration of remote sensing, GIS, socio-economic, and agricultural data

created a data model necessary to construe anthropogenic influences on the Sierra
Vista ecosystem. The supervised classification maps derived from remotely sensed
data provided a visual and statistical interpretation tool for temporal and spatial
monitoring of natural land cover as well as human land use patterns over the study
area. As evidenced from both remotely sensed and agricultural census data, land
cover categories that experienced positive and negative areal changes provided
important clues to the identification of drivers of land use change. The census

division and tract data provided means to analyze the role of human settlement

61

patterns, land-based economic activities and their impact on land cover change.
The well registry database was given geographic properties and as a result could be
mapped for visual processing and served as a vital instrument for determining
patterns of water usage and the impact of water demand on land use and land cover
change. Urban land area totals derived from classified land cover maps, coupled
with population, housing, and agricultural statistics, served as important indicators
for water demand as well. Finally, the temporal and spatial characteristics of
vegetation cover types derived from remotely sensed data, with extra attention
given to areas distant from the influence of urban intrusion, provided important
links between grazing techniques and rangeland management strategies and their

impact on the fragile semi-arid ecosystem.

62

CHAPTER F OUR—RESULTS

Land use and land cover change and global environmental change form a
complex and interactive system linking human action to use/cover change,
use/cover change to environmental feedbacks, and environmental feedbacks to their
impacts and human responses (Turner et al. 1995). Further complicating this
system is the fact that the linkages occur at different spatial and temporal scales.
Decreasing native grass cover in greater Sierra Vista, for example, has immediate
impacts on land productivity, vegetation changes (increase in distribution and
frequency of mesquite and scrub species) and soil erosion, mid-term impacts on
landscape fragmentation, and possible long-term impacts on regional environmental
change.

Varied human driving forces (e.g., population, development), mediated by the
socio-econorrric setting (e.g., market economy, resource institutions) and influenced
by the existing environmental conditions or context (e. g., semi-arid climate regime),
lead to an intended land use (e.g., livestock rangeland) of an existing land cover
(e. g., semi-arid grassland) through the manipulation of the biophysical conditions of
the land. These manipulations (e.g., grazing) are the proximate causes of change to
distinguish them from the underlying human forces of change and are the most
immediate activities or actions that create change. These actions may convert the
existing cover, in this case by grazing practices and lack of effective fire policy, or
modify it through introducing exotic grass species into existing rangelands and

failing to effectively manage invasive species. Either cover change (conversion or

63

modification) is further affected by biophysical forces that change it unless human
inputs are used to maintain the converted or modified cover (physical maintenance:
e.g., removal of scrubs, weeds, and invasive species). Changes in the landscape
may themselves affect the driving forces and the social setting in which they
operate, and these effects may alter the intended land use.

The causes and consequences of land use and land cover change are research
paths that aim toward understanding the social, economic, and policy factors that
manipulate land use, occasionally addressing environmental components of the
problem, but are rarely directly relevant to regional and global change studies. A
major exception is work drawing on integration of remotely sensed imagery, census
data, and GIS, used as tools to link land cover to land uses, proximate sources of
change, and the underlying human causes to improve knowledge of carbon flows
and biodiversity change. The regional analysis of the Sierra Vista census division
revealed three important social drivers of land use and land cover change. These
are urban growth and land fragmentation, water demand and legal issues, and
grazing management and fire suppression. The information and interpretations
generated by these drivers could offer lessons with which to frame land use and
land cover dynamics in rangelands, particularly for transitory stages of change, and
provide a factual basis on which to develop and test models of land use and land
cover change by improving our understanding of current and future use/cover
activities. In addition, the drivers serve as the primary causality factors of land
cover change and are important indicators of anthropogenic processes that directly

link to patterns observed on the landscape. Understanding these processes as they

64

pertain to semi-arid ecosystems is critical for sensible management of economic
development and natural resources in drylands. After the discussion of drivers of
change and their relation to patterns observed on the landscape, the chapter
concludes with a short discussion on integrating spatial information and technical
considerations involved in the development of a decision support system for

promoting informed range management practices.

4.1 URBAN GROWTH AND LAND FRAGMENTATION

Results of the area totals derived from remotely sensed imagery classifications
reveal a sustained increase in urban cover type (Tables 9 and 10, Figures 8 to 11,
14). Overall, the census division urban cover increased fi'om 2433.9 hectares in
1973 to 7100.7 hectares in 1999, a 192 percent expansion. A tract-level inspection
reveals more specific trends. Examining Tables 9 and 10 discloses that tract 15
experienced the least amount of urban expansion, probably because this tract is
located in the inner core of the city of Sierra Vista. Tract 14 is comprised mostly of
Fort Huachuca, a US. Army base. The temporal and spatial expansion of built
cover in this tract is more stable and constant, and is characteristic of military
installations that typically have regulated planning, design, site feasibility, and
construction requirements (Balbach 2000). Tracts 16, l7, l8 and 19 show a large
surge of growth in built cover between 1973 and 1985, and level off to a lower and
more stable increase between 1985 through 1999. Tract 20 was virtually
undeveloped until 1999 when it experienced a 95 percent increase in built land.

Tables 5 and 6 show similar trends in totals of population and housing units for the

65

county division and census tracts. The most densely populated tracts coincide with
the urban core census tracts of 15, 16, and 18.

But what does urban growth have to do with changes in grass cover besides
moderate increments of land transition? The transformation matrix (Table 11)
shows that grassland to urban net gain was only 5 percent from 1973 to 1985, and 2
percent from 1985 to 1992 as well as from 1992 to 1999. As discussed next, the
main urban impact on the grasslands of greater Sierra Vista is largely an economic
phenomenon.

The major force governing land use in the region is economic or location rent.
The term rent in this context has a specific meaning that differs from that implied
by the everyday use of the term. Economic or location rent is the surplus income
that can be obtained from one unit of land above what can be obtained from an
inferior unit of land (Chisholm 1962); for the most part, it is measured against land
at the margin or limit of major agricultural cultivation, i.e., grass rangeland. For our
purposes, we can define economic rent as a measure of the level of return that the
market at large (all the potential bidders for land) would expect a particular piece of
land to produce. It is basically a measure of the advantage, as the bidders see it, of
one piece of land over another. This implies that pieces of land differ in some
respect and that such differentiation is reflected in higher or lower returns per unit
of land. In this simplified model, the causes of differentiation are the friction of
distance and spatial variation in the quality Of the natural resources. However, the
major variable is no longer simply the transport cost of agricultural commodities to

transshipment points (as is the case in many traditional von Thunen-based location

66

theory models), but rather demand for accessibility to urban markets and amenities
that has occurred on a large scale in the Sierra Vista census division.

Urban land invariably commands a higher value than rural land, and where the
two type of use are in direct competition, urban uses generally win. But land that is
expected to become urbanized has a higher value—an anticipated value——and this
has a considerable effect on the type of land use practiced in rural areas of the
census division. In the typical von Thunen model, lands adjacent to the urban
market tend to be farmed at the highest intensity. In reality, however, such land is
most likely to become urbanized and thus has the highest anticipated value. Under
these circumstances, a rancher is unlikely to invest large amounts of capital and
labor in livestock production when, by waiting a little while, very large financial
gain might be achieved by selling the land to property developers. The speculative
value of land for agricultural purposes, therefore, is lower very close to an
expanding urban center and increases with distance as the likelihood of urban
encroachment declines. As evidence, the value of agricultural land per farm in
Cochise County has decreased on average from $ 795,292 in 1987 to $ 545,528 in
1997 (Table 8). In addition, Table 8 also reveals that the region has experienced a
steady decline in total agricultural and pastureland area. In 1981, Santa Cruz and
Cochise counties were first and second, respectively, in the sale of remote
subdivision lots in Arizona. Tighter temporarily imposed restrictions on developers
in the early 19705 forced many major rural subdivisions into bankruptcy (Hecht and
Reeves 1981), leaving behind thousands of cleared hectares with eroding, unused

bulldozed roads.

67

The encroachment of urban land use on the hinterlands is a testament to models
developed by Douglas (1994) and Sinclair (1967) that emphasize the role and
impact of urban settlements and settlement-related activities in promoting land
cover change and conversion of land from rural and natural uses to urban usage.
The typical scenario of land exchange involves the sale of agricultural land to
speculators who wait until encroaching development is imminent before selling the
land at a larger profit to subdivision developers. This relationship is illustrated in
the classified imagery as well as in the GIS wells coverage of tract 20 (Figures 11
and 13). Tract 20 is adjacent to the urbanized tracts and is a transitional zone in the
sense that urban land rent is greater than nonurban land rent, though not by enough
to lead to massive development (notice unpaved road and utility corridors in Figure
7, and the large spatial coverage of wells in Figures 13). Land speculation in this
zone is likely in its later stages. The transition zone is likely to be partitioned into
subcategories such as that zone where urban land rent is greater than the cost of the
sum of the compensation to agriculture and speculators’ earnings (Thrall 1987). It
is important to note that between the city core and the transition zone, people are
willing and able to pay the cost of converting nonurban land to urban land use;
hence, the land here is urban. This is equivalent to requiring that infrastructure is in
place prior to development and that the new residents at the urban margins pay the
cost of the infrastructure.

Land development in tract 20 started recently with low-density housing
developments often with unpaved roads, water wells, and underground septic

systems. This zone of transition is now leading to urban development to the extent

68

that the infrastructure of highways and other public goods are being installed as the
zone is being developed. This particular transition zone is mostly reserved for high
to median-income housing for retirees and white-collar workers (Marsett 2000).

Tract 20, evidenced by land cover totals in Figure 8, was dominated by grass
cover in 1973. The majority of the grass, however, is being converted to urban land
use and land cover through the mechanism of economic rent and land speculation,
which in turn is being fueled by a steady influx of new residents to this particular
tract. The increasing trend in population is attributed to the census division’s
locational advantage for a booming tourism industry as well as its locational
advantage as a sun-belt community for retirement. Although the economic rent
variable is identified as a contributor to urban growth, another question remains in
the explanatory power of urban growth and land fragmentation as a driver of land
use and land cover change. What role do biophysical factors, including global
climate change, play in augmenting urban expansion and outbidding of agricultural
lands?

McCarthy and Lindberg (1966) developed a physical optima and limits model
on the spatial pattern of agriculture. Two key variables are employed: temperature
and moisture. Over a particular area, there is an optimal combination of both inputs
for a particular crop (i.e., grass rangeland). Outward from this, hoWever, the
restrictions imposed by the physical and environmental limits of the two variables
make conditions less and less favorable for agricultural resources. Although in
need of further research, the most serious problem associated with the greenhouse

effect, brought by the ability of the atmosphere to be selective in its response to

69

different types Of radiation (mainly short-wave solar radiation) which are
augmented largely by carbon dioxide and methane gases that absorb solar energy, is
the increasing dryness often accompanied by rising air temperatures in many areas
(Kemp 1990). Reduced precipitation following changes in circulation patterns, plus
the increased rates of evapotranspiration caused by higher temperatures would
create severe moisture stress for grasslands in already fragile ecosystems.
Ultimately, under extreme but likely conditions, physical limits will be reached that
make livestock production undesirable. Of course, it is always possible, even
within the context of environmental limits, to provide production. Costs, however,
rise to infinity near extreme limits, meaning that a real technological barrier exists.
In fact, the optima provide the rancher free of charge with the maximum benefits
that nature can provide in terms of temperature and moisture. To move away from
the optimum imposes costs and recourse to scarcer resources. If global warming-
induced low precipitation and high temperature patterns become the norm in
southeastern Arizona (as suggested by the recent lack of snow cover, and
consequent lack of recharge in regional aquifers), it is ventured that the price of
agricultural land will fall at an even faster rate, creating a clirnatologically
intensified, greenhouse-based expansion of the transition zone between urban and
agricultural land open for speculation and development. As non-urban land prices
fall, speculators will likely opt to purchase more and more rangeland property for
future urban uses. Therefore, global environmental factors in addition to land rent
will contribute to urban expansion. Given the characteristics of spatial economic

production and consequent labor inputs discussed below, this chain of events will

70

only lead to a cycle of depletion of additional grass and other natural cover coupled
with a rampant pattern of urbanization.

Sierra Vista serves as a commercial center for southeastern Arizona and
northern Mexico, and is rapidly attracting capital, banking, and additional
commercial interests (Clark 2000). The large influx of new firms into the region is
likely to create external econorrries of scale, defined by Dicken and Lloyd (1990) as
“savings that a plant or firm gains from its connection with other plants or firms.”
One source of such external economies is explicitly spatial. By clustering in close
spatial proximity to other activities, it is believed, firms will benefit from a
particular kind of external economy of scale called economies of agglomeration. In
this context, scale economies achieved during the early phases of growth are passed
on as external economies expand. The cumulative process of development has a
multiplier effect by which direct effects (production and sales) are followed by
indirect effects (increase in labor pool, local consumption of goods, rise in real
income, and increase in the number of services) in a chairrlike sequence as
expansion induced in one sector has repercussions on other sectors, though the
effect becomes less and less pronounced as distance from the original stimulus
increases (Dicken and Lloyd 1990). This phenomenon is readily observed in Table
7. Contract construction, retail trade, transportation and public utilities all
experienced tremendous sequential increases in the number of employees and
payrolls in selected years. The service sector, a crucial sector of external economies
of scale, experienced an even more impressive boom (payrolls for the first quarter

rose from almost $2 million in 1973 to over $44 million in 1999).

71

But agglomeration may also generate diseconomies. There may be a point at
which an expanding urban agglomeration becomes incapable of maintaining its
efficiency although sustainable steady growth continues. Problems such as
congestion and clogged transportation arteries, soaring land prices, pollution, and
administrative overload begin to transform urbanization economies into
diseconomies. However, the most crucial affect of agglomeration in Sierra Vista
from a local and regional viewpoint, fueled by both economic and environmental
stimuli, may be its strong detrimental effect on the ecosystem resulting from
cumulative encroachment into open spaces of grass cover. The landscape
fragmentation that has resulted from extensive abandoned rangeland combined with
increasing urban and rural settlements isolate areas of desert grassland and hinder
the dispersal of species and the spread of fires. There is a need for more research in
determining optimal sizes for agglomerations just as for individual plants, sizes at
which public utilities and services are provided at optimum levels of efficiency, and

the environment is protected from mass-scale encroachment of built land.

4.2 WATER DEMAND AND LEGAL ISSUES

Although groundwater is renewable through natural recharge, overpumping can
temporarily deplete the resource. If groundwater-pumping withdrawals exceed
aquifer recharge, then groundwater is being “mined” or overdrafted from the basin
(Lacher 1994). Further, groundwater overdraft conditions may be evident in
groundwater declines that form a large “cone of depression” in the aquifer around a
pumped well or in the area of an active well field. If pumping continues to exceed

aquifer recharge over time, the cone of depression will continue to expand outward

72

from the wells. The rate of water level decline and growth of the cone of
depression depend on the aquifer’s hydraulic characteristics, the surrounding
geohydrology, the rate and volume of water pumped, and the rate of recharge to the
aquifer (Lacher 1994). The cone of depression will continue to grow until
equilibrium is achieved and aquifer recharge equals aquifer discharge.

Currently, although groundwater is renewable through natural recharge,
groundwater pumping withdrawals exceed aquifer recharge in greater Sierra Vista
region due in large part to increased municipal and industrial demand for water
(ADWR 1999). The same ADWR report also reveals that total volume of
recoverable groundwater held in storage in the upper San Pedro basin has decreased
by 13 percent in the regional aquifer since 1990. Due to technological innovations,
improved well system design for groundwater extraction (such as turbine and
centrifugal pumps with electric engines) has allowed for deeper well depths. The
average depth for all wells in the Sierra Vista census division is 275 feet, a depth
that penetrates well into the regional aquifer located under the region’s substrata.
These deep artesian wells have taken a toll on the groundwater supply. ADWR
(1990) reports the existence of a cone of depression approximately 4 miles long and
2.5 miles wide (10 square mile size) in the Sierra Vista-Fort Huachuca area, with
long axis parallel to Huachuca Mountains (N W-SE direction). The cone’s extent in
the area north and east of Sierra Vista is ill defined because it falls within the firing
range on Fort Huachuca where few wells are accessible. It has been reported that
artesian pressures throughout the region have been declining due to prolonged

pumping (Vionnet and Maddock 1992). The water levels in two wells located at

73

the 1990 center of the cone of depression showed a net decline Of 50.2 and 48.2
feet, respectively, during the period 1980 to 1988 (ADWR 1990). Net decline rates
over an area of 25 square miles centered at Sierra Vista ranged from 0.4 to 3.9 ft/yr
(averaging 1.4 ft/yr) for the same period. Schwartzman (1990) reported the
existence of a second smaller (3-mile long) cone of depression near Huachuca City
along the Babocomari River (northwest of Sierra Vista). Further, evidence indicates
that perennial reaches of the Babocomari became intermittent after cultural
development (Brown et al. 1981).

Shallow wells typically found on the San Pedro River floodplain aquifer have a
potential impact on the regional aquifer as well. Decreasing water levels around a
well form a roughly circular cone of depression, deepest at the well and lessening
with distance from the well (Lacher 1994). As pumping time and frequency
increases, the volume and radius of the cone increase and its edge may reach the
strearnbed if the well is shallow. If streamflow is present, the well will draw water
directly from the stream. If no surface water is present in the vicinity of the cone of
depression, the cone continues to increase in size in the floodplain aquifer. If the
cone reaches the interface between the floodplain aquifer and the regional aquifer,
the well will begin drawing water from the regional aquifer. In Sierra Vista,
groundwater pumping alone accounts for the wide spread water table decline
(Putman et al 1988).

In addition, I would add the recent dry climatological pattern over the region to
the contributors of declining regional aquifer. Typically, precipitation in

southeastern Arizona is bimodally distributed with about 50 to 60 percent of the

74

annual total falling in the summer monsoon season, and 21 to 35 percent occurring
in the winter months (Lacher 1994). Spring and fall are usually dry. Winter storms
originate mainly from mid-latitude Pacific (cyclonic) fronts, producing several days
of rain, moderate winds, and mountain snow. Over the last 10 years, the
precipitation pattern over the Sierra Vista census division has become event-driven
with scattered and sporadic shower and thunderstorm activity rather than the normal
frontal activity associated with cooler temperatures (Mauget and Upchurch 1999).
Isolated precipitation events along with diminishing winter snow cover on the
Huachuca Mountains cause a decrease in the amount of recharge level necessary to
equal aquifer discharge since direct infiltration over the valley floor is considered
negligible because of high evaporation and low precipitation rates. In effect, the
drying climatological regime causes further imbalance on the supply-side of the
regional aquifer equilibrium.

Hydrologic model studies provide further warnings of additional groundwater
depletion in the near future. Putrnan et al. (1988) projected that pumpage in the
Sierra Vista-Fort Huachuca area will triple fi'om 1986 to 2006, with a maximum
annual rate of groundwater level decline projected at 6 ft/yr by the year 2006.
Pumping estimates for the future were based on population projections and
consumptive use figures based on the 1980 census (228 gal/day/person). The
projected pumpage in the Sierra Vista subbasin accounted for the BLM land
exchange (establishment of the SPRNCA in 1988), which includes several land
grants along the San Pedro River and additional farmland near the International

Border. To make matters worse, however, the conceptual model did not take into

75

account the dry climatological regime and characterized the aquifer system as being
unconfined and regularly recharged by the Huachuca Mountains front recharge.

An aquifer’s capacity to transmit water, or hydraulic conductivity, and the
volume of water an aquifer can release or take into storage, described by the
storativity value, are the primary scientific descriptors of an aquifer (Putman et al.
1988). These two values are used to estimate regional groundwater flow rates,
groundwater storage volume, and the effect of a pumping well on the aquifer and
other nearby wells. If regional aquifer overpumping continues, the ability of the
aquifer to transmit and release water to the surface will diminish due to negative
impact of overpumping on artesian pressures (Putman et al. 1988). The total
volume of recoverable groundwater in an aquifer is determined by the geology and
chemistry of the aquifer material, which in turn determine the number and size of
open pore spaces that exist. Under Sierra Vista, water held in storage in the
regional aquifer resides in these pores, and is released to the surface by an upward-
declining pressure gradient through the upper basin fill which is composed mostly
of permeable and porous material including gravel, sand, silt, and poorly cemented
clay (Lacher 1994).

Increasing rates of well drilling and water demand due to residential and
commercial development in the region, evidenced by increasing population, urban
area, and housing units (Tables 5, 6, 9, and 10, Figure 13), are drivers of land use
and land cover change in the Sierra Vista region because decreasing amounts of
regional ground and surface water has a direct, frontal impact on agricultural use of

land. Over the last 50 years, depletion of groundwater reserves has led to an initial

76

increase in rangeland at the expense of farmland (Cox et al. 1983), and the same
cycle is presently leading to potential depletion of grass cover on rangelands. The
very nucleus of the range ecosystem is composed of the interrelations among
grasses, soils, climatic variables (precipitation and temperature), and surface and
groundwater. Annual and perennial grasses both have fibrous root systems, with
the densest concentration in the top 15 to 20 centimeters of soil (Hyder 1973). On
the other hand, competitors such as burrroweed have few feeder roots in the top foot
of the soil, but have a taproot that extends much deeper than the grass roots.
Because of root system differences, depleting near-surface soil moisture caused by
urban water demand, and potentially global warming, shrub and weed species have
the competitive edge during the summer growing season. Furthermore, annual
species such as tobosa grass, frequently encountered on heavier soils subject to
flooding (Anderson et al. 1963), are disappearing due to lack of baseflow (discharge
of groundwater to the stream which sustains surface flow in dry seasons) and
competition with cholla, burroweed, mesquite, and phreatophytes.

As the water demand in greater Sierra Vista increases, the validation and
quantification of surface water and groundwater rights are coming under the
scrutiny of local, state, and federal institutions. On June 12, 1980, the Arizona
legislature enacted a comprehensive groundwater management code governing the
allocation and use of groundwater. This legislation was a direct product of
Arizona’s continued overdraft of groundwater (Lacher 1994). The 1980 code
created four Active Management Areas (AMA’s) with specific regulations on

groundwater pumping and well construction, and two Irrigation Non-expansion

77

Areas (INA’s) in which expansion of irrigated agriculture was prohibited. Because
greater Sierra Vista falls outside of AMA’s and INA’s, however, the beneficial use
rule applies to groundwater rights. That is, anyone may pump water from below his
property as needed for a “beneficial” use on that parcel of land. The only
restrictions that apply to drilling wells or pumping groundwater are the
requirements of well registration and conformity to well construction standards. As
seen in Table 12 and Figure 13, the freedom of “beneficial” use has been taken to
an extreme by residential developers who drill wells at a significant depth below the
upper reaches of the regional aquifer (presently around 200 feet for artesian
conditions). Further, proliferations of wells in tract 20 exemplify the unregulated
nature of water management in the region.

Due to urban water demand and its negative environmental effects, there is an
urgent need to provide research and information for a decision on whether the
ADWR should designate Upper San Pedro Basin as an AMA under the 1980
Groundwater Management Act. The 1980 Act allows the director of the ADWR to
form an AMA if any of the following conditions exists: 1) active management
practices are necessary to preserve the existing supply of groundwater for future
needs, 2) land subsidence or fissuring is endangering property or potential
groundwater storage capacity, or 3) use of groundwater is resulting in actual or
threatened water quality degradation (Lacher 1994). It is humbly recommended
that ADWR should designate this particular basin as an AMA given the recent

unregulated urban demands on water resources and the propensity of most

78

developers to overlook harmful environmental effects of unrestricted well
construction.

In 1988, Congress passed the San Pedro National Conservation Area Act. This
motion established the Conservation Area out of public domain lands managed by
the BLM “in order to protect the riparian area and the aquatic, wildlife,
archeological, paleontological, scientific, cultural, educational, and recreational
resources of the public lands surrounding the San Pedro River in Cochise County,
Arizona.” The initial reservation comprised 56,000 acres and provided for the
subsequent acquisition of additional parcels of land. The Act required the Secretary
of the Interior to develop a long-term management plan and it expressly reserved
water rights. Additional land purchases of subsequent threatened areas, especially
those that are currently in danger of being developed adjacent to the river on tract
20 (Figure 11), should provide adequate protection from urban encroachment.
Leasing of such lands for recreational purposes, wildlife and game management, as
well as public domain grazing could provide revenue for additional restorative
efforts and help promote sound range management practices on regulated federal

lands.

4.3 GRAZING MANAGEMENT

4.3.1 Mesquite invasion
One of the most dramatic changes in the grasslands in greater Sierra Vista in the
last 30 years has been the rapid increase of scrubby trees and shrubs (Bahre and

Shelton 1993). These woody invaders, whose spread has been influenced, if not

79

caused, by overgrazing and fire suppression, deserve inclusion into the discussion
of grazing impacts and rangeland management.

The recent increase (19805 onward) in dense phreatophyte cover in the San
Pedro riparian corridor is probably attributable to an increase in the percentage of
vegetation made up by mesquite and tarnarisk (Putman et al. 1988). As early as
1965, Hastings and Turner reported a dramatic increase in mesquite populations
along the floodplains and an invasion of tarnarisk. At present, mesquite invasion is
widespread in the Sierra Vista census division. Examining the transformation
matrix in Table 11, we see that nearly 20 percent of grass cover in 1973 became
mixed mesquite by 1985. The net gain, however, was negative since 25 percent of
mesquite was transformed to grass. A 13 percent net gain by mesquite occurred
between 1985 and 1992. This percentage rose to 22 percent between 1992 and
1999. Figure 14 also demonstrates the invasive nature of mesquite. Total area of
mixed mesquite in the census division increased from 11,843 hectares in 1973 to
18,548 hectares in 1999, a 57 percent increase. Most of the dramatic increases took
place in tracts 14, 17, and 20 (Figures 8 to 11). The majority of lands on these
tracts are located outside the 5-kilometer urban buffer comprised of tracts 15, 16,
and 18. Because a vast majority of large-scale and hobby ranches are located in
these rural areas, the proliferation of mesquite could have a direct relationship with
grazing strategies and land management practices utilized by range managers and
hobby farmers.

In the southwestern United States, range productivity on most sites can be

greatly increased by control of mesquite, and tarbush, and seeding with Lehmann

80

lovegrass and fourwing saltbush (Holechek 1998). Mesquite competes seriously
with warm-season perennial grasses. The inverse relationship between density of
mesquite and perennial grass production has been widely recognized (Reynolds and
Martin 1968). The roots of mesquite and acacia plants are generally more extensive
than those of herbaceous grass species, and allow the plants to tap soil water
sources at greater depths. Other species, such as creosote bush, have a spreading
root system that enables the plants to exploit water sources for extensive areas in
the soil surface around the plant. Anderson et al. (1963) concluded that mesquite
and acacia stands exceeding 15 to 25 trees per acre should be removed or thinned
before reseeding with grass, and that recovery of perennial grasses under improved
grazing management was faster on mesquite-free range than where mesquite was
not killed.

The causes identified for the mesquite increase include (1) a reduction in the
frequency and intensity of wildfires due largely to overgrazing following settlement
and fire suppression; (2) a decline in natural perennial grasses, which, when healthy
and dense, can reduce mesquite seedling establishment; (3) increased dissemination
by livestock and/or Merriam kangaroo rats of scarified mesquite seed; (4) hoof
damage to ground cover and soil compaction by livestock resulting in reduced
moisture in the upper layers of soil, which hinders grass establishment and growth;
and 5) land clearing and cultivation (Bahre and Shelton 1993). These management-
related causation factors have contributed greatly to the demise of grass cover in the

Sierra Vista census division.

81

The rural census tracts, namely tracts 14, 17, and 20, have the most noticeable
loss of grass cover in the classified imagery (Figures 8 to 11). Tracts 14, 17, and 20
lost 78, 42, and 36 percent of grass cover, respectively, between 1973 and 1999.
Although other causes previously discussed, such as urbanization and water
demand, contributed to the decrease, it is safe to venture that a large portion of grass
cover loss in these tracts was due to lack of effective livestock management

practices.

4.3.2 Impact of grazing on rangelands

Livestock affect watershed properties by consumption of plant parts and
through the physical action of their hooves. Accelerated erosion occurs when lack
of proper range management leads to destruction of the vegetation cover that retards
soil loss from the forces of water and wind. In addition, accelerated erosion is the
most severe consequence of overgrazing due to the fact that, especially in arid and
semi-arid regions, replenishment of lost soil is a very slow process. Severe erosion
also leads to additional environmental problems such as decrease in water quality
and infiltration, and increase in surface runoff (Holechek 1998). The condition of
the soil and the vegetation complex on which precipitation falls has a major
influence on the quality and quantity of water available to the semi-arid ecosystem.
Therefore it is necessary to examine the relationship between range management
and grazing impacts on the ecosystem and related land use and land cover change,
with particular emphasis on infiltration, runoff, erosion, and water quality. As a

management and decision support tool, remote sensing and remote sensing-based

82

products could provide valuable information on monitoring and assessment of
ecological impacts attributable to grazing methods as well as providing value-added

assistance for future protection of rangelands from overuse and neglect.

4.3.2.1 Infiltration

Grazing and browsing animals reduce water infiltration by removing protective
plant materials and compacting the soil surface by hoof action. The negative
impact of heavy grazing on water infiltration is well documented. After reviewing
grazing impacts on infiltration, Gifford and Hawkins (1978) concluded:

1. Ungrazed plots have higher infiltration rates than those of grazed plots.

2. Moderate and light grazing intensities have similar infiltration rates.

3. Heavy grazing causes definite reductions in infiltration rates over
moderate and light grazing intensities.

Although it has been speculated that under some conditions the hoof action of
grazing animals will loosen the surface of compacted or crusted soils, actual
research shows just the opposite effect (Warren et al. 1986). Several studies in
Texas and New Mexico are consistent in showing the concentration of hoof action
under short-duration grazing reduced infiltration compared to continuous grazing
(McCalla et al. 1984; Thurow et al. 1986). Heavy stocking consistently reduced
infiltration during the grazing season, but this appeared to be alleviated by winter
freeze-thaw activities. Other studies are consistent in showing that grazing systems

other than short-duration grazing have little influence on infiltration rate but that

83

reductions occur when stocking rates are increased from moderate to heavy (Wood
and Blackburn 1981; Pluhar et al. 1987).

A few studies have evaluated the influences of grazing on soil structure. In
New Mexico, lightly grazed, heavily grazed, and severely grazed ranges had pore
spaces of 68 percent, 51 percent, and 46 percent, respectively (Flory 1936). It was
also found that heavy grazing degraded soil structure by reducing the percentage of
water-stable aggregates compared to moderate grazing, and that heavy grazing
increased soil compaction (bulk density) more than moderate grazing (Wood and
Blackburn 1984). In the same study, different grazing systems (continuous, Merrill
three-herd/four-pasture, and high intensity-low frequency) had a similar effect on

water-stable aggregates and bulk density.

4.3.2.2 Runoff

Livestock grazing increases nmoff by reducing infiltration. Increased surface
runoff due to heavy grazing is usually associated with water quality degradation
because of increases in sediment and other pollutants (animal wastes, agricultural
chemicals, and decayed vegetation).

It is well documented that heavy grazing increases runoff compared to moderate
grazing (Sharp et al. 1964; Hanson et al. 1970). On the other hand, protected areas
generally have the least runoff. Branson and Owen (1970) showed that runoff
increases as vegetative cover and mulch decreases and the amount of bare soil
increases. The same study documented the inverse relationship between runoff and

vegetation cover.

84

Limited data indicate that moderate or light grazing can increase groundwater
and runoff compared to no grazing, without having a detrimental impact on the
watershed or water quality (Hanson et al. 1970; Lusby 1970). Under moderate or
light grazing intensities, adequate vegetation is maintained to protect the site, but
excessive vegetation that causes water losses by transpiration and evaporation is
removed. More research is needed on the potential of controlled grazing to increase
water yields. Information regarding influences on groundwater would be
particularly useful for the southwestern United States, where many underground

aquifers are being rapidly depleted.

4.3.2.3 Erosion

Sediment yields in runoff and streamwaters are commonly used as a measure of
erosion on rangelands. Heavy grazing accelerates erosion by reducing plant cover
that protects the soil and retards overland flow. Several studies have documented
higher erosion under heavy grazing intensities than under moderate intensities, and
concluded that moderate grazing will not cause watershed damage on most
rangelands (Thurow et al. 1986; Pluhar et al. 1987). In many cases watershed
recovery can be accomplished by changes in grazing practices. In New Mexico,
Aldon (1964) reported that sediment loads were reduced more than 75 percent due
only to better livestock control (change from season-long to winter grazing) and
reduced grazing intensities.

The influence of short-duration grazing on sediment production compared to

other grazing methods has been a concern. Available research is consistent in

85

showing that short-duration grazing increases sediment production compared to
moderate continuous grazing (McCalla et al. 1984; Thurow et al. 1986). The
reduced vegetation cover associated with short-duration grazing in the studies cited
previously appeared to cause the higher sediment production.

The most detailed evaluation of hydrologic responses under short-duration
grazing was reported by Warren et al. (1986). They studied infiltration and
sediment production on a silty clay soil (similar to the association in greater Sierra
Vista) using a short-duration grazing system with moderate, double-moderate, and
triple-moderate stocking rates. Short-duration grazing at all intensities reduced
infiltration and increased sediment production compared to no grazing. These
deleterious effects were increased as stocking rate increased. The damage was
augmented when the soil was moist at the time of trampling. Thirty days of rest
was insufficient to allow hydrologic recovery. Another part of the study evaluated
seasonal changes in infiltration and sediment production under short-duration
grazing at a moderate stocking rate. The infiltration rate declined and sediment
production increased following the short-term intense grazing period inherent to
this system. These effects were most severe during drought and dormancy, due to
reduced vegetation standing crop. It was also found that there was no definite
hydrologic advantage of increased stocking density via manipulation of pa5ture size
and numbers.

Although limited data are available, the amount of vegetation required to protect
different types of rangeland needs more study. Complete protection of the soil

requires about 550 kg per hectare of plant material (Osborn 1956). Ground cover

86

levels of 30 to 40 percent appear adequate for flat, arid areas, and levels as low as
20 percent will protect most soils from wind erosion (Branson et al. 1981).

The key to maintaining healthy hydrological conditions on rangelands is
through grazing practices that develop and maintain a good grass cover. Perennial
grasses, because of their high basal area and excellent soil binding properties, play
the critical role in watershed stability (Holechek 1998). Moderate stocking rates in
conjunction with grazing practices such as rotational deferment and rest rotation
grazing that promote even livestock distribution over the range appear to be the best
approaches to maintaining a good perennial grass cover. The success of any
grazing program geared toward watershed maintenance and enhancement is best
measured by the residue of living and dead vegetation it maintains on the site
throughout the year. A good residue of forage left ungrazed prior to initiation of
new growth in the spring may appear to be wasteful from a forage standpoint.
However, in the long run, the land manager will be rewarded by higher forage
production and less variation in the forage crop between years due to increased soil

moisture and mineral supplies available for plant growth.

4.3.2.4 Water quality

Fecal wastes from livestock grazing can be a sizable pollution problem in range
watershed management. Fecal colifonn bacteria counts in water have been used as
an indicator of infectious bacterial contamination (Wadleigh 1971). However, the

coliforrn bacteria themselves are not pathologically harmful. Livestock operations

87

have caused increased coliform bacterial pollution in rangeland washes and streams
(Gary et al. 1983).

The extent of the bacterial pollution depends largely on livestock numbers,
timing of grazing, frequency of grazing, and access to the stream. Tiedemann et
al.’s study (1987) showed that fecal coliform bacteria levels tended to increase as
intensity of livestock use increased. This study presented evidence that livestock
removal may not provide an immediate solution to elevated fecal coliform bacteria
levels in streams. Grazing strategies that disperse rather than concentrate livestock,
such as rotational deferment and rest rotation grazing, appear best when fecal
contamination is a concern. Practices that improve livestock distribution and attract
livestock away from streamside areas are also recommended in Tiedemann et al.’s
study.

Treated sewage sludge as a soil amendment on rangelands does not appear to
deleteriously affect the quality of runoff water (Aguilar and Loftin 1991), and in
some situations, applications of municipal sewage sludge increases herbage yields

and reduces detrimental runoff and sedimentation.

4.4 FIRE SUPPRESSION

Fire has historically been common in most desert grasslands. Before 1882, fires
were extensive, sometimes covering hundreds of square miles, and several lines of
indirect evidence suggest that fires occurred at least every 10 years (Bahre 1977).
Shrubs were inconspicuous in desert grasslands before 1880, which suggests that
fires occurred frequently enough to prevent widespread shrub development. Most

desert grassland shrubs are susceptible to fire, at least as seedlings (J .H. Bock and

88

Bock 1992). For example, velvet mesquite plants usually did not resprout
following fire unless stems were larger than 1 centimeter in diameter when they
burned (Glendening and Paulsen 1955). Furthermore, many woody species do not
produce seeds until they are at least 10 years old, and their seeds on the soil surface
are easily killed by fire (Martin 1975). Thus, a fire frequency of once every 7 to 10
years appears to maintain relatively shrub-free grasslands.

Considerable evidence suggests that widespread livestock grazing reduced fine
fuel, and therefore fire frequency, in southeastern Arizona after 1880 (Bahre 1977).
Historical accounts and direct evidence of reduced fire frequency in nearby pine
forests of Huachuca Mountains document reduced incidence of fires concomitant
with the buildup of the livestock industry (Martin 1975). In fact, forest
administrators encouraged overgrazing to reduce fire hazard and promote tree
grth (Bahre 1993).

Many grass species common in southeastern Arizona including blue and hairy
grama, plains lovegrass, as well as the introduced lehmann lovegrass, recover much
more quickly after a fire than large, decadent scrub and mesquite species because
their rhizomatous roots are located below the soil surface, and thus usually escape
lethal temperatures (Glendening and Paulsen 1955). However, large bunchgrasses,
such as threeawns, and stoloniferous species like buffalo grass and black grama, are
damaged more than smaller bunchgrasses because more fuel is present, fire duration
is longer, and heat penetration is deeper into plant tissue (Holechek et al. 1998).

Burning may significantly affect nutrients in range soils. Burning sometimes

increases the supply of nitrogen, phosphorus, and sulfur available for plant growth

89

(Holechek et al. 1998). Nitrogen is frequently limiting, especially on brush-
supporting soils. Addition of even small amounts of available nitrogen may have a
profound effect on revegetation (Hobbs et al. 1991). Although fire survivors have
access to more resources (e.g., light and soil nutrients) than they had before the fire,
it must be noted that a plant species need not survive a fire to reap its benefits.
Species that produce abundant and widely dispersed seeds capable of establishing in
the high-light, fluctuating-temperature environment characteristic of recent burned
rangelands may also benefit from fire. Examples include annual grasses such as
needle grama, threeawn, and Lehmann lovegrass (Hobbs et al. 1991).

After a fire event, the resprouting vigor of woody shrubs decreases with
decreased soil moisture content; resprouting is generally less common following
fires that occur during the growing season compared to dormant-season fires
(Wright and Bailey 1982). Blue paloverde, burroweed, ocotillo, oneseed juniper,
and snakeweed rarely sprout following fire (Wright and Bailey 1982).

The long-term absence of fire may produce dramatic changes in community
structure and function, particularly if soils do not limit shrub establishment
(Holechek et al. 1998). In the absence of fire, desert grasslands may develop dense,
woody overstories that significantly reduce herbaceous grass production. The
resulting lack of fine fuel reduces fire intensity and frequency, and the community
changes from grassland to shrubland (Brown and Lowe 1980). After this threshold
between grassland and shrubland has been crossed, land management strategies
should be reevaluated. Once woody plants dominate a site, fire alone cannot return

it to the earlier composition since fine fuel is too scarce and discontinuous to

90

produce fires of sufficient intensity to kill woody plants. Thus, woody plants
become permanent occupants of the site. Without herbicides or mechanical shrub
control the change is irreversible. The net result of the absence of periodic fires is a

reduction in herbaceous grass production.

91

CHAPTER FIVE—CONCLUSIONS

5.1 FUTURE RESEARCH AND DECISION SUPPORT SYSTEMS

Reversing the problems of grazed ecosystems requires a better ability to make
good management decisions. The world’s biological resources are being
overexploited because local ecosystems, especially agroecosystems, are exporting
resources to distant markets (McNeely 1990). Cities and foreign markets gain the
benefits while local communities experience the environmental costs. Whether
local or global, decisions always have a political component and always contain an
element of subjective balancing. Inconsistencies and disagreements will be greatest
when the subjective component of decision-making is greatest. Such problems are
minimized when different parties (ranchers, land managers, public policy officials)
share information, have common methods of defining problems, and agree on the
broad objectives of sustainable development (McNeely 1990). In this circumstance,
disagreements can focus on and resolve the value judgments without having to
argue about what ought to be the objective underlying facts.

Decision support systems (DSS) offer a mechanism for improving the
objectivity of decision making, especially where complex interactions are involved.
An open and distributed web-based DSS for rangeland monitoring should aim to be
flexible, with an integrated design for accessing, retrieving, and generating images
and reports, as well as being able to accommodate decision models for conducting

further analyses including sensitivity analysis using remote sensing algorithms such

92

as numerous green and recently developed senescent vegetation indices, fractional
cover, and biomass estimates.

The concept of a DSS has evolved over the last 15 years. As other computer-
based tools have developed, the potential for integrating a wide range of data
sources has become obvious. These include databases and graphically oriented
systems such as remote sensing and data visualization. Furthermore, the limitations
of some of the information sources and computer-based presentation techniques
have been recognized; hence, non-computer elements are now commonly accepted
as part of an overall DSS (Stuth and Lyons 1993). These features become
particularly important when dealing with users and uses where computers or
Internet connections are not available (e.g., ranchers in remote locations).
Therefore, the major goal of a DSS must be to improve decision-making through
conversion of variety of data sources into value-added information and provide
users at all levels of technical access and training with the means to assess
alternative outcomes against each other more objectively and comprehensively than
could be done previously.

The technological development of remote sensing and data visualization began
years ago as separate endeavors, but the two now meet at a time when their
unification is an important objective for the development of decision support
systems for natural resource management. Coupled with a distribution network
utilizing the World Wide Web, we are now able to develop distributed, open, web-

based DSS with data acquisition, data management, and data utilization capabilities.

93

For base tier data establishment, the most reliable source is satellite imagery.
Displayed in Table 13, ASTER and MODIS sensors mounted on the EOS-AM
platform, in addition to other commercial sensors, provide fine pixel resolution,
short off-nadir repeat cycles, and quick delivery from acquisition time; the SPOT 4
VEGETATION sensor Offers 48-hour repeat cycles because of its wide 2,250 km
swath width although the spatial resolution is rather coarse (1.36 square kilometers).
As such, this sensor should be utilized for regional rangeland analysis, while
ASTER and MODIS sensors are ideal for detailed vegetation analysis desirable for
most small- to medium-scale ranching operations. Although ETM+ sensor
mounted on Landsat-7 has ideal pixel resolution (30-meter), its repeat cycle is 16
days, and delivery time is generally 2 weeks to one month. Therefore, this sensor
may not be suitable for most rangeland DSS applications that require expeditious
decision-making capabilities for determining optimum stocking rates in suitable
areas.

Ideally, all the data acquired should be organized, managed, and used in an
integrated manner, although segmentation of information resources is quite
common. Since data sets are typically application-specific, they are often designed
and operated independently of each other. However, there is an emerging
awareness of the need to use a general purpose OODBMS (Object-Oriented
Database Management System) to unify information management. The goal of
such systems, e.g., ORACLEI, is to increase the efficiency of information

processing by maintaining greater data integrity, less data redundancy, and more

 

' Use of ORACLE trademark does not imply endorsement.

94

efficient data sharing and output. Greater data integrity means that if a data element
(e.g., pixel reflectance value in an image) is stored in more than one place, its value

should always be the same. When updated in one place, the data element’s other
locations will be automatically updated. Duplication of storage of the same data
element should be kept to a minimum to reduce redundancy. Faster output and data
sharing refers to the access of the same data item by more than one application.
Obviously these three principles are closely related to one another and serve to
expedite more efficient information processing. Major attractions of OODBMS are
their functionality since they are applicable to both tabular and spatial data, and the
fact that systems are comprised of objects and each object has its own distinct
behavior. Certain behaviors in objects may trigger behavior in other objects.
Objects are abstractions of real world entities, such as grass cover, soil moisture,
images, information requests, etc. All these objects have their own behavior and
characteristics and can influence one another in any manner that may occur
naturally or abstractly in our own minds. As OOP (Object-Oriented Programming)
handles objects in the way we treat them in our Own thought processes, we should
be more comfortable in understanding and implementing systems when they are
developed in an object oriented fashion.

In OOP, the behavior of an object is represented by its state changes (changes in
object property values) and by its communication with other objects, which
normally triggers new behavior in those objects. Communication is accomplished
through message sending. For example, a management plan object may send an

“area” message to a specific grass cover object. The grass cover object would

95

invoke its “area” method, which would retrieve its area value from the object-
oriented database, which in turn was calculated by invoking a GIS process. As can
be seen, both the database and GIS processes are methods of an object. The object
itself forms the link. Therefore, specialized processes in raster data visualization
(calculation of vegetation indices, fractional cover and biomass algorithms),
topological calculations (parcel by parcel land cover totals, route analysis), and
databases (retrieval of image and tabular data from related entities) will themselves
be comprised of objects with behavior and inter-object communication.

A system can be built from a collection of objects, perhaps several hundred,
with each object doing a very specialized job. These objects must be classified
within a system under “classes” that form the abstraction from which objects can be
created. For web-based applications, this requires the functionality and capability
of an intemet-savvy, object-oriented 4th generation programming language such as
Java. Java is a platform-independent language and is incorporated into all of the
major Web browsers (Internet Explorer, Netscape). Therefore, Java-based
applications may run as stand-alone executable programs on any platform (PC,
UNIX, and LINUX) or as applets on the Web. Its powerful networking capabilities
allow developers to create three-tier architecture system designs with databases,
servers, and clients. Third-party data visualization library and methods packages,
such as J WAVE, provide Java with server-side capability to manipulate and process
orthorectified multi-band images to calculate vegetation indices, fractional cover,

and biomass estimates.

96

The following is a simplified design of a three-tiered, distributed DSS created in
Java (Figure 15). The middleware (a Java-CGI application residing on a web
server), on a daily basis, calls two objects, search and input, that establishes a
connection to a distributor’s FTP site searching for the most recently available,
sequentially-ordered raw images. If the images are present, the web server retrieves
them. The database is opened for input via the recently developed ActiveX data
object control (a predefined set of objects for accessing database servers), and the
raw images are sent to the database by invoking a “send” message. In the database,
the image objects are sorted by sensor, area coverage, and date. After the database
connection is closed, the raw data are now stored and wait to be invoked by a user-
defmed client message received by the web server. The users on the client browser
may specify to view the whole image, a subset of an image, an animation sequence,
algorithm-derived indicators in either image or tabular format, or request a FTP to
their IP addresses. These specifications, depending on user choice, become data-
strearned image, subset, algorithm, or delivery methods handled by objects and
pointers programmed in the web server. Obviously, these entities must be
programmed to handle native data structure of the images. Once the web server
receives the requests from the client browser, it invokes server objects that retrieve
and process the raw images stored in the database. Outputs are sent with data
coordinates to the browser to allow for client-side GUI interaction and FTP requests
are sent to the user’s IP address. Figure 16 provides a simple diagram of the DSS

from the perspective of user-GUI interaction.

97

Level One algorithms in Figure 16 (NDVI, NDSVI) detect vegetation by
examining red, NIR, and SWIR regions of the electromagnetic spectrum. Level
Two products are based on NDVI-related algorithms that identify the fractional
cover and the leaf area index of vegetation per unit of land. Biomass maps are used
to estimate total dry plant weight per unit area, and time series include animation
maps produced to aid managers in detecting spatial and temporal change by quick
visual assessment. Forage production estimates will be derived from both Level
One and Level Two indicators. Other products that can be created from Levels One
to Three include environmental indicators such as forage weight, canopy cover,
wildlife habitat and fuel loading.

Because most decision support systems are user-specific decision tools, one of
the major goals of a rangeland DSS is to meet and exceed range managers’
requirements for geospatial information. A rangeland DSS should utilize high-
quality vegetation data and be designed as an aide for effective assessment of
vegetation conditions. As such, it must assess large areas with varying species
composition and frequency, and provide timely and valuable indicators of
vegetation for user-defined areas of interest. Furthermore, requests for archived
data, comments and feedback regarding improvements in GUI functionality and
design, and user satisfaction surveys should be given prompt attention to improve
the quality of the end product. By listening to user input, comments, and
recommendations, a powerful product could be developed to aid range managers to
quickly and accurately assess vast areas to determine rangeland health, forage

levels, and optimal grazing plans.

98

5.2 FINAL REMARKS

Between 1973 and 1999, the rate of conversion of grasslands in the Sierra Vista
census division has increased (Figure 14). According to satellite data, grasslands
were reduced by 27.5 percent between 1973 and 1985, 24 percent between 1985
and 1992, and 29 percent between 1992 and 1999. Overall, 61 percent of the grass
cover in 1973 was eradicated by 1999. During the same time period, mixed
mesquite and scrub area totals increased by 57 and 21 percent, respectively. Of
local and regional significance is not just the area converted but also the area
modified or changed in ecological condition. Grasslands in the region are being
degraded through mismanagement, with such consequences as soil erosion, changed
floristic composition, and diminished productivity. In addition, population growth
and water demand driven by the census division’s locational advantage as a tourism
and retirement spot are not only leading to conversion of land cover from grass to
urban, but are also additional contributors to land degradation and fiagmentation.

Of the major physical processes of global change—climatic change, alteration
of biogeochemical cycles, and land use—the most influential agent of future change
in grasslands will be land use. The most probable outcome in southwestern United
States semi-arid ecosystems is an extensive transformation of the grasslands, either
by degradation through unsustainable use or by conversion to urban use and cover.
These changes will be principally, but not exclusively, driven by the requirements
of the human population. Further, these land use changes will have far more
extensive and degrading impacts on the grasslands than the forecast consequences

of climate change.

99

Over the last 20 years, the rate of conversion of grasslands to croplands has
slowed in the developed world (Meyer and Turner 1994). In the developing world
the rates are high and increasing, but quite variable from one country to another.
The rates of conversion are highest in grassland-rich nations and are small or
negative in tropical nations where forests are being converted to croplands and
grasslands. Unsustainable pastoral use is already a major ecological problem in
many areas of the developed and developing world. Because the brunt of the
human population grth is yet to come for many of the developing nations, global
demand for the products of the land is likely to continue accelerating for the
foreseeable future. For developed countries, environmental protection coupled with
economic sustainability is the key issue. The capacity of the land—and of the
ecosystems more generally—to sustain that demand will remain an issue of
fundamental importance. The level of concern that current trajectories of change
have elicited reflects the possibility that much land transformation in some sense
constitutes land degradation, whether that is defined as a decrease in the capacity of
the land to meet demands placed upon it or is given some other meaning.
Environmental change in the aggregate or in particular cases, such as in greater
Sierra Vista, raises serious questions about ecocentric versus anthropocentric views
of nature. Therefore, the analysis of land use and land cover change inithe Sierra
Vista region is an attempt for better scientific knowledge than we now possess of
the physical and anthropogenic extent, character, and consequences of land

transformation.

100

One of the many useful applications of remote sensing is to inventory
biophysical materials and man-made features on the surface of Earth. Some of the
data are static, that is, they do not change over time. Conversely, some biophysical
materials and man-made features, such as vegetation cover and built area, are
dynamic, changing rapidly. It is important that such changes he inventoried
accurately so that the physical and human processes at work can be more fully
understood. In this study, remotely sensed data were used to transform satellite-
collected reflectance data into land cover information using image-processing
techniques that included supervised classification. Drivers of land use and land
cover change were inferred by examining the link between changing rates of land
cover information and socio-economic urban growth models, indicators of water
demand, and grazing management strategies. Integration of remote sensing, GIS,
and socio-economic data was vital in detecting and quantifying changes in land use
and land cover, monitoring urban growth and water usage, and analyzing the impact
of grazing strategies on the land surface.

Quantitative capabilities of remote sensing for accurately capturing land cover
change detection were limited due largely to lack of available reference test
information (DOQs and other field data). Comparison of ground truth with remote
sensing-derived classifications would have allowed a meaningful quantification of
land cover classification accuracy. Furthermore, the temporal nature of population
and housing data prevented the analysis from detecting urban growth variations
between the 10-year census collection gaps. A higher time frequency of image

acquisition, preferably on an annual basis, coupled with annual tract-level census

101

and agricultural data, would have provided a more detailed assessment and
identification of drivers of land use and land cover change.

To understand the total Earth system and the effects of natural and human-
induced changes on the global environment, it is critical to see the Earth as an
intricately coupled system involving the interactions of land, oceans, atmosphere,
ice, biota, and most importantly, man. A sound scientific understanding of the
Earth system provides a foundation for sustainable development—economic
development that meets the needs of the present without compromising the ability
of future generations to meet their own needs. The unique capability for remote
sensing of the Earth from space, coupled with in situ observations of human
settlement patterns, provide the data needed on global, regional, and sometimes
local scales to fuel that understanding. Land use, i.e., how the land is being used by
human beings, and land cover, i.e., the biophysical materials found on the land,
provide valuable insight on the complexity, fragility, and constraints of a particular
ecosystem and its interaction with humans. Understanding this relationship requires
the identification of interlinked factors and drivers that have a direct environmental
and social impact on a given ecosystem. The drivers of change must be identified
before the level of environmental and economic awareness can increase high
enough to implement sustainable land use planning and ecosystem protection.

Lack of land use planning that recognizes the integrity of the ecosystems should
be given the highest ranking of environmental priorities. Because land use
decision-making is often fragmented, with numerous local and regional units having

planning authority, there is a need for both qualitative and quantitative approaches

102

to provide integrated and shared solutions. Knowing where and why land use
conflicts currently exist and how they have impacted our environment and quality
of life is the first step toward resolving these problems and preventing similar
impacts in the future. To assist local, regional, and global planning, inferential
insights into the driving forces of land use and land cover change is necessary to
provide a portfolio of evaluative and predictive tools. To this end, there is an
unprecedented need to determine mechanisms to develop and transfer technology
that meets the needs of land managers, ranchers, planners, and values of society.
Dissemination of temporal and spatial land cover and vegetation indication
information over an open, distributed decision support system will increase the
capabilities Of concerned parties to make informed environmental and economic

decisions.

103

TABLES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year 1973 1985 1992 1999
Platform Landsat 1 Landsat 5 Landsat 5 Landsat 7
Sensor MSS TM TM ETM +
Altitude 900 km 705 km 705 km 705 km
Swath Width 185 km 185 km 185 km 185 km
30m (120m3 0m ( l 20m30m(15mpanchromatic
Spatial Resolution 80 m thermal band) herrnal band) Om thermal)
Radiometric Resolution 6-bit 8-bit 8-bit 8-bit
Number of Spectral Bands 4 7 7 8
Table 1. Sensor characteristics of images.
Sensors
MSS TM ETM+
Blue 0.45-0.52 0.45-0.52
Green 050.6 0.5206 0.52-0.6
Wavelengths Red 0.6-0.7 0.63-0.69 0.63-0.69
(in micrometers)NIR 0.7-0.8;0.8-l.l 076-09 076-09
SWIR 1.55-1.75;2.08-2.35 1.55-1.75;2.08-2.35
Thermal 104-12.5 10.4-12.5
Panchromatic 0.5-0.9

 

 

 

 

 

 

 

 

 

 

 

Table 2. Sensor wavelengths.

104

 

 

 

MSS DN values generated for haze corrections using Band 1 input
Atmospheric condition: very clear; Date of Acquisition: 09/05/1973

 

 

Predicted Final
1 18.110 17.000
2 9.290 9.095
3 5.234 5.443
4 2.173 0.470
Table 3. Improved-DOS haze correction values for 1973 image.
Source Collection Years

 

Census of Population and Housing (US. Dept. of Commerce)

1970, 1980, 1990, 1997

 

Census of Agriculture (US. Dept. of Ag.)

1974, 1987, 1992, 1997

 

 

County Business Patterns (US. Dept. of Commerce)

 

 

1972, 1985, 1992, 1999

 

Table 4. Source and years of socio-economic data.

Population

Housing Units

Census division
County

Census division
County

 

 

 

 

 

 

 

 

 

1970 1980 1990

20422 33939 41325
61910 85686 97624
7140 12203 16830
14378 32564 40238

 

 

 

Table 5. Population and housing data totals for county and census division,

1970-1990.

105

 

 

 

 

 

 

 

 

 

Pop. Pop. Pop.l990 Sq. Housing

Tract 1990 1997 Mi. Units
0014 8689 9819 68.3 1987
0015 7949 10225 4043.8 3 896
0016 7263 9341 2459.0 3265
0017 7631 9686 231.4 3384
0018 4471 5778 1021.2 1661
0019 2121 2524 297.0 789
0020 4250 4949 75 .6 1848

 

 

 

 

 

 

 

 

 

Table 6. Population and housing data totals for census tracts, 1990 and

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1997.
Years 1973 1985 1992 1999
Number of employees
Total 10901 12914 17091 21019
Agric.,forestry,fish,mine D 226 254 195
Contract construction 661 986 858 1 154
Manufacturing 1391 1375 1243 1242
Transportation/public util. 671 1125 1105 1379
Wholesale trade 318 507 504 609
Retail trade 3364 4376 5729 7277
Finance,insur., real est. 701 699 748 903
Services 1928 3148 6642 8245
Taxable payrolls (Jan-Mar.)
($1000)
Total 16951 40503 69442 97062
Agric. ,forestry,fish,mine D 1 181 1360 931
Contract construction 954 3011 2885 4840
Manufacturing 2797 6552 5857 6769
Transportation/public util. 1357 6407 7951 12193
Wholesale trade 483 1574 1886 3411
Retail trade 3436 9361 14464 20297
Finance,insur., real est. 959 2365 2880 4269
Services 1920 9272 32149 44319

 

 

 

 

 

"D" denotes figures withheld to avoid disclosure of operations of individual

reporting units.

Services sector includes hotels, business, computer, auto, recreation, health,

education, social, and management services.

Table 7. Number of employees and taxable payrolls per economic sector
in Cochise county, selected years.

106

 

 

 

 

 

 

 

1974 1987 1992 1997
All farms 721 836 831 824
Acres 21 12344 2077793 1891664 1260021
Average size 2930 2485 2276 1529
Average market value($) 328857 795292 731623 545528
Percent of land area 52.80% 51.94% 47.28% 31.50%
Cropland farms 466 478 501 446
Cropland acres 149556 139086 120472 116018
Cropland pasture farms 174 203 233 185
Cropland pasture acres 149556 139086 27060 26353
Woodland pasture farms 35 27 15 20
Woodland pasture acres 35023 4597 2022 1988
Pastureland farms 598 597 607 583
Pastureland acres 1927765 1962436 1780389 1148704
Land irrigated 110485 49012 52434 63252
Land in inigated farms 670039 486679 510020 222816
Irrigation water (acre-ft) 329528 154700 174318 1071 14
Sprinkler irrigation acres 14823 24813 29819 27114
Cattle and calves farms 484 544 522 517
Cattle and calves 81333 77788 65289 69950

 

Table 8. Agricultural patterns in Cochise county, selected years.

Total Area of Urban Land Cover, hectares

1973 1985 1992 1999
Tract 14 1,316.88 1,447.65 1,665.99 1,689.75
Tract 15 375.48 471.87 475.29 470.70
Tract 16 174.60 465.12 578.88 582.80
Tract 17 77.76 555.39 748.35 938.79
Tract 18 60.84 642.06 724.05 747.63 .
Tract 19 29.16 465.57 725.61 1,258.47
Tract 20 399.24 550.98 723.33 1,412.55

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 9. Total area, in hectares, of urban land cover, selected years.

107

 

Percent Change in Urban Land Cover
1973-1985 1985-1992 1992-1999
Tract 14 9.93%! 15.08°/ 1.439
Tract 15 25.67% 0.72% 6.97%]
Tract 16 166.39% 24.46% 0.68%]
Tract 17 614.24% 34.74% 25.45%]
Tract 18 955.33% 12.77% 3.26%]
Tract 19 l496.60°/ 55.87°/ 73.42%]
Tract 20 38.01% 31 .28°/ 95.280/1

 

 

 

 

 

 

 

 

 

 

 

 

Table 10. Percent change in urban land cover by census tract, selected years.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1985
Grass Scrub Mixed Mesquite Barren Urban
Years Cover Classes % Hectares % Hectares % Hectares % Hectares % Hectares
Grass 50231150570 20.16 4617.18l 19.74 4521.51 1.41 322.92 6.83 1564.56
Scrub 15.13 2021.49 48.59 6490.71 29.081 3884.94 2.27 302.581 2.46 328.77
1973Mixed Mesquite 25.15 2971.44 18.83 2225.61 36.88l 4357.71 8.81 1041.30 4.60 543.15
Barren 5.01 156151672 521.01 2.03 631817300 2275.02 2.47 76.95
Urban 2.25 54.7 4.781 116.37 4.7 116.64 0.75 18.36 85.59 2083.32
1992
Grass Scrub Mixed Mesquite Barren Urban
% Hectares % Hectares % Hectares % Hectares % Hectares
Grass 59.11 9899.19 7.66 1283.04 23.03 3856.59 7.24 1211.76 2.86 479.52
Scrub 6.47 923.85 55.60 7943.13 30.44 4349.61 4.14 592.02 2.91 415.53
1985Mixed Mesquite 10.21 1484.01 20.39 2963.34 62.37 9065.97 0.73 106.11 3.15 458.19
Barren 5.83 234451251 503.01 7.99 32121719 2892.06 1.62 65.16
Urban 1.06 48.96 3.47 159.75 2.87 131.76 2.41 110.881 90.14 4145.31
1999
Grass Scrub Mixed Mesquite Barren Urban
% Hectares % Hectares % Hectares % Hectares % Hectares
Grass 54.73 6929.46 7.8 997.83 29.30 3709.17 4.24 537.12 3.76 476.55
Scrub 2.54 331.92 62.60 8168.581 25.35 3307.32 2.45 319.681 4.44 579.33
1992Mixed Mesquite 7.07 1334.88! 27.05 5104.62 55021038320 1.20 226.89 3.27 617.581
Barren 6.43 317701816 896.85 12.20 602.73 57.93 2860.83 4.83 238.50
Urban 1.781 100.53 2.74 154.80 3.49 196.74 0.81 45.81 90.95 5131.081

 

 

 

 

 

 

 

 

 

 

Table 1 1. Land cover change transformation matrix, selected years.

108

Average well depth (ft) Averwater level (ft) Average pump rate (gal./min.)

 

 

 

 

 

 

 

 

 

 

 

 

 

Tract 14 281.881 182.69 293.91
Tract 15 277.60 257.38I 483.33
Tract 16 546.06 403.53 280.21
Tract 17 236.81 145.45 345.30
Tract 18 567.02 374.14: 173.71
Tract 19 226.76 125.41 28.00
Tract 20 269.71 174.66 54.281
Census Division 275.26 182.5 99.56
Table 12. Well attributes per census tract, 2000.
Sensors (Platforms)
MODIS (EOS-AM) ASTEQEOS-AM) VEGETATION (SPOT 4)

Spectral Region
(micrometers)

Pixel Resolution

Orbital Characteristics
Off-Nadir Repeat Cycle
Time of Data Acquisition

Delivery time from
acquisition

 

0.66-0.87(2 bands),
0.47-2. 13(4 bands),
0.42-0.94(12 bands)

0.52-0.86(3 bands),
1.6-2.43(6 bands)

0.4-0.89(3 bands),
1.58-1.75 (lband)

 

0.25 km (visible, NIR)

15 m (visible, NIR)

1.36 km (visible, NIR)

 

Polar-orbiting, sun-

Polar-orbiting, sun-

Polar-orbiting, sun-

 

 

 

 

 

synchronous Synchronous synchronous

1-2 days 5 days A8 hours

10:30 AM 10:30 AM ate-moming
8 hours AS hours 48 hours

 

 

 

Table 13. Satellite-based sensors along with their spectral region,
pixel resolution, and orbital characteristics.

109

FIGURES

 

ARIZONA

in...

Tucson

 

Slerra Vlsta .HCOCHISE  
_ ensus D1vrsron -coUNTy3zg 4

\p

 

 

 

 

Figure 1. Political boundary of study area.

110

 

.c-i
94
- Florence
. fifi
I. l _ \ 3"%
L" \3‘ 5mm 0
‘ ‘8 \0. x‘"
t" \f ‘ T- ’:
“"ia \. \9-
(a \3 ~:/ \
'\’o \93, x 0
‘33- ‘r‘: ‘3‘
- ‘ $.59» 2‘ ‘.

‘\.-. V /'\‘\N
“cm a "\ \. IE SanSimono‘

. San Ignacio dc la Canon (a. 1820) \‘.

I‘.~
fl

“ f: LLuhMadaMfBlaHoatNoJ)
' (a. 182!)
"”° ~ . . 3. Maria Saundra: del Car-e- fl
‘ \ (Dilemma) (ca. 1826)
I 4. San lo“ de Sonoiu (ca. 1821)
5. San Ignacio dd Babocomari (ca. 1827)

5 6. San Juan del lac Boqumu y‘Nog-ra
(ca. 1827)
7. 5» Rafael do 1. z“): (a. 1821)
a. San Rafael del Valle (ca. 1827)
' 9. San Bemardino (a. 1821)
L3 1.0 3.5 39w
9

/ l Douqu o . . ‘

 

 

 

 

Figure 2. Authenticated Spanish and Mexican land grants in southeastern Arizona.

111

 

 

   

 

 
  
 

 

  
 
    

 

 

  
  

 

 

 

 

    
 
 

Raw image AWOSPFCTIC Geometric
from ARIA COHCCUOH correction
GI: Image Image
grr classification coregistration
GIS
classification
vector coverage Magnitude Datafile:
1 I II Land cover
totals per
procedure
census tract
Overlay Relate Well attributes
procedure procedure INFO file

 

 

 

 

 

 

A

 
   

 

GIS census

 

    
      

 

 

Clean & vector
Final GIS / Project coverage from
census vector USGS NBII

 
 

 

      

coverage

  

 

Figure 3. GIS and remote sensing processing and integration.

112

 

 

 

Landsat MSS image of Sierra Vista Census Division, 1973

3500000

3480000

 

 

Scale
f 2 Meters
10000 0 1 0000
Imago MOM-ll
Doll of Aunt-mon- 09/05/1973
Datum: M083
Sch-wick GRS 1990
Units: motor!
Fin! Sin: 80
Reflection: UTM
Zone: 12
Organization: uses
Dopenmonr EROS Dou Comer
Address: Simx Felt. SD 57195

 

Figure 4. Landsat MSS image and metadata, 1973.

113

 

 

 

Landsat TM image of Sierra Vista Census Division, 1985

 

 

Scale
— — Meters
10000 0 10000
Inogo Mondale
Date of Acquisition 09/2 7/ 1985
Datum: NAD83
thcroid: 605 1980
Units: motor:
":01 Size: 30
Projection: uTM
Zone: 12
Organization: University of Arizona
annnmom: Oflico of Arid Lands Studios
Addrou: Tucm. AZ 86719

 

Figure 5. Landsat TM image and metadata, 1985.

114

 

 

 

Landsat TM image of Sierra Vista Census Division, 1992

 

 

Scale
r _ __ . Meters
10000 0 10000
Image Matador.

Date of Acquisition: 09/30/1992

Datum:

Organization:
Mflml:
Address:

NADB3
GRS 1980
mm

30

Space Int-giro
Crammer Service
Lam-m. MD 20708

 

Figure 6. Landsat TM image and metadata, 1992.

115

 

 

 

Landsat ETM+ image of Sierra Vista Census Division, 1999

3500000

3490000

 

 

Scale
Meters
10000 0

Image MM
Data of Anquieition; 10/12/1999
Datum: NADBS
Spheroid: GRS 1980
Units: meters
Pixel Size: 30
Projection: UTM
Zone: 12
Orwnlzation: USGS
Department: EROS Dela Canter
Addie“: Sioux Fella. SD 57198

 

Figure 7. Landsat ETM+ image and metadata, 1999.

116

 

 

 

Sierra Vista Census Division Land Cover, 1973

 

 

 

 

 

 

 

3500000 - - 3500000
3490000 -l - 3490000
3400000 - - 3400000
Scale
E = Malere
10000 0 10000
Area louh in Square Meter:
Land Cover Categories Tract I4 Tract 15 Tract 16 Tract 17 Tract 18 Tract 19 Tract 20
- Fume! 37249200 0 o o o o o
- “Woodland 19979200 0 o 0 o o o
- MiaadMeequlte 88919900 421200 1108800 12949200 1713800 4963800 31370400
5 era-clam 97888000 902900 4309200 33888800 7905800 11714400 73094400
- Saubiand 82357200 39800 309800 24194800 1000800 1422000 24606200
- mm o o 0 1893800 0 0 2788400
- Agrsctutur. o o 0 298200 0 0 1713800
E: Barren 10989200 32400 201800 11917200 194400 144000 9020800
- w... 244900 8 c 72000 0 o o
J: Urban 13188800 3764800 1748000 777900 808400 291900 3992400

 

Figure 8. Classified land cover map and area totals per census tract, 1973.

117

 

 

 

Sierra Vista Census Division Land Cover, 1985

 

 

 

 

 

 

 

- 3500000

1- 34900110

- 3480000

Seal.
5 = m
IMO 0 10000
A4119 Totals in 501-n Mot-u

Land Cover Catomrloa Tue! 14 Tract IS Tract 18 T1801 1? Treat 18 TIM 19 T1808 20
- Form 30504800 0 0 0 0 0 0
- 0.111410811181111 15993000 0 o o 0 0 0
- Mind Mosquito 104014800 1081100 747900 13011500 14371100 8298200 1110911100
a 8188818118 80339800 47700 898200 29088200 1718100 1258200 73130300
- Duo-numb 80318700 135900 1195200 21788100 1730700 3402000 34300800
- Riparian o 0 0 534800 0 0 2471500
- 441110011014 0 0 0 1521900 0 0 1818800
a 8mm 14840100 5400 183800 14112000 139800 178400 10287800
- Wm 242100 0 0 79100 0 0 0
a cum. 8982200 0 0 0 0 871400 0
E Urban 14478500 4718700 4881200 5553900 8420800 4855700 5809800

 

Figure 9. Classified land cover map and area totals per census tract, 1985.

118

 

 

 

Sierra Vista Census Division Land Cover. 1992

 

 

 

 

 

 

 

 

3500000 4 - 3500000

3490000 4 - 3490000

3400000 .. .. 3480000

Nu TotalsinSquroMoton

um Covorcmoviu ram 14 mm 15 1188118 71.81 17 11m 18 Tract 18 mm 20
- rm 33109900 0 0 o 0 o 0
- 088W 12489500 0 0 o 0 0 0
- mm mm. 130785500 82800 729000 13834200 1701000 8311700 35484300
a aw 38450900 44100 598700 27784800 414000 888800 80724800
- noun-8m 80080400 132300 1.30100 17823500 1881900 4080900 25920000
- mun-1 o 0 0 490500 0 0 1514700
- Wlfllm 0 0 0 1511100 0 0 0
D Barron 17910000 5400 31500 15580800 208800 253800 1441:1500
- W410: 244500 0 0 79800 0 0 0
E 018811. 0 0 0 o o 0 0
E 010411 18850900 4752800 5788800 7483500 7240500 7258700 7233300

 

Figure 10. Classified land cover map and area totals per census tract, 1992.

119

 

 

 

Sierra Vista Census Division Land Cover. 1999

 

3W-

3‘90000 .1

 

 

 

 

 

 

 

Scale
r = Met-9
10000 0 10000
A208 Total- in Stu-1m Motor-
WCOVM cuegorim rm 14 Tract 15 Tract 18 Tract 17 Inc: 19 71.8119 Mano
- Forest 27548800 0 0 0 0 0 0
- 0211 Woodland 18899900 0 0 o 0 o 0
- Mixed Mewuite 119379800 139500 881300 20388100 1489800 2175300 41081400
: 0184:1488 21948400 12800 897900 19740800 729100 88800 48813700
- Deurucmb 113187700 139500 219800 18522800 1452800 9588500 24997500
- 81.1.1... 0 0 0 1998300 0 0 4827800
- 491mm 0 0 0 1339200 0 0 o
: Barren 11527200 18900 89300 14338100 299700 48300 13844900
- er 241300 0 0 70500 0 0 0
E Clouds 1103400 0 o 0 0 0 0
a 11111411 18897500 4707000 8828000 9387900 7478300 12584700 14125500
Figure 11. Classified land cover map and area totals per census tract, 1999.

120

 

 

 

 

5000 0 5000 10000 Meters

 

Figure 12. Sierra Vista Census Division (Tracts 14-20).

121

 

 

Registered Well Distribution by Census Tract, 2000

 

 

Registered Well Totals by Census Trad

Trad 14: 141
Trad 15: 26
Trad 16: 18
Trad 17: 132
Trad 18: 58
Trad 19: 240
Trad 20: 785

Swroa: Arizona Dept. (1 Water Resumes Well Regstry

 

Figure 13. Registered wells in 2000 per census tract.

122

 

wcmmrouhmmmmrnmm

 

 

.1985
.1992

 

 

 

 

 

 

   
 

fl _ V\\\\\§

  

7/fl/flw/é

'— — *“—'—_———— __ B1973

 

y/Ax//////7///////////////////////z

nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn
ooooooooooooooooooooooooooooooooooo
nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn

 

      

   

 

§§§§
Egg/g

nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn Coco‘s-ocooovunoOIII-Io
-----------------------------------------------------------

 

 

Egg/5%

        

 

  

§
7/4/AWW/l

 

20000
15.11” >—
M

toil-‘18»

 

mm“

Fond

 

 

, selected

lVlSlOIl

Figure 14. Land cover area totals in Sierra Vista census (1

years.

123

 

 

 
  
 

  

Web client
Purpose: get information
from the user on a web
page, send it to the
application server to
process the information

      
     
 
 

Middleware (Java-CGI)
Purpose: build web pages
based on requests from the
users via a graphical
interface, establish database
and remote connections

 
   
   
     
   
 
  
 

  

Database Server
Purpose: handles server
connections and perform
SQL statements, receive
queries specified by user

 
     
   

 

Figure 15. A simple three-tier DSS design.

124

 

 

 

 

 

Enter affiliation;
Select date(s), sensor, location
(lat/long, region, or ranch)

 

 

 

 

Request handled by worker
threads made up of pointers
programmed in the web server
to retrieve selected file(s) from
an OODBMS (ORACLE)

 

 

  

List of available
images (in HDF
and [MG format)

 
     

 

 

 

Select image(s) from list

 

 

 

 

Request handled by worker
threads made up of pointers
built into an object that handles
image input/output

 

 

 

 

      

Display
RGB/greyscale
image

......................................

 
   

 

Sent as data-streamed JAVA
objects that recognize data
coordinates and allow for
client-side GUI interaction

 

 

 

 

 

 

 

   
       
    

Level II
Biomass

Fc, soil
moisture, LAI

 
   
   

Level I
NDVI,
NDSVI

  
  

   

 

 

 

 

 

Server worker threads
Select site or click on image must be programmed to
for subset handle [MG and HDF data
structures
Level 111
Annual
biomass
Prompt for production,
indicators animation \
FTP/Download; digital

   

formats: .eOO, geoTIFF,
.shp, .gif, .bsq; tabular
formats: .txt, .dat

      
  
   
 

 

 

 

   

 

Figure 16. A simple GUI design for a rangeland DSS.

125

 

APPENDIX

/* AML tabulates land cover classes by census tract in Sierra Vista division
/* written by Osman Wallace

/* begin date: 09/08/00

/* end date: 09/08/00

&setvar year = [response 'enter year for tabulation (eg, 73): ']
intersect sv_ctracts3 sv%year°/oclass classtract%year°/o poly .1
&setvar parc = classtract%year%.PAT

&setvar infi = MAGNITUDE

&call freql

&data arc info

arc

select %infi%

&call zero_out

select JUNK

RELATE %infi% BY SV_CTRACTS3-ID

RES GRID-CODE = 1
CALC $1FOREST = AREA

as

RES GRID-CODE = 2

CALC $10AKWOODLAND = AREA
as

RES GRID-CODE = 3

CALC $1MESQUITE = AREA

as

RES GRID-CODE = 4

CALC $1GRASSLAND = AREA

as

RES GRID-CODE = 5

CALC $1DESERTSCRUB = AREA
as

RES GRID-CODE = 6

CALC $1RIPARIAN = AREA

as

RES GRID-CODE = 7

CALC $1AGRICULTURE = AREA
as

126

RES GRID-CODE = 8
CALC $IBARREN = AREA
as

RES GRID-CODE = 9
CALC $1WATER = AREA
as

RES GRID-CODE = 10
CALC $1URBAN = AREA
as

select %infi%
calc CHECK = 0
calc DIFF = 0
calc CHECK = FOREST + OAKWOODLAND + MESQUITE + GRASSLAND +
DESERTSCRUB + RIPARIAN + AGRICULTURE + BARREN ~
+ WATER + URBAN
calc DIF F = AREA - CHECK
select JUNK
erase JUNK
Y
select %infi%
calc $PRINTER-SIZE = 200
&setvar ws = [show workspace]
output %ws%\%infi%%year%RPT.dat init

print 'Tract Area Forest Oak Mesquite Grass
Scrub ~
Riparian Agriculture Barren Water Urban Check
Diff
print

*W4,TRACT,*W14,AREA,*W14,FOREST,*W14,0AKWOODLAND,*W14,ME

SQUITE,*W14,GRASSLAND,*W14,DESERTSCRUB,*W14, ~
RIPARIAN,*W14,AGRICULTURE,*Wl4,BARREN,*W14,WATER,*W14,UR

BAN,*W14,CHECK,*W14, DIFF

Q STOP

&end

&return

/******************************************************************

&routine freql
/******************************************************************
frequency %parc% junk
sv_ctracts3-id
grid-code
end
area
end
&return

127

/******************************************************************

&routine zero_out
/******************************************************************

CALC FOREST = 0

CALC OAKWOODLAND = 0
CALC MESQUITE = 0
CALC GRASSLAND = 0
CALC DESERTSCRUB = 0
CALC RIPARIAN = O
CALC AGRICULTURE = 0
CALC BARREN = 0

CALC WATER = 0

CALC URBAN = 0

&return

128

REFERENCES

Aguilar, RD. and SR. Loftin. 1991. Sewage sludge application in semiarid
grasslands: Effects on runoff and surface water quality. N. Mex. Water
Resour. Res. Inst. Rep. 265.

Aldon, ER 1964. Ground-cover changes in relation to runoff and erosion in west-
central New Mexico. U.S. Dep. Agric. For. Serv. Note RM-34.

Anderson, D., L.P. Hamilton, H.G. Reynolds, and RR. Humphrey. 1963. Reseeding
desert grassland ranges in southern Arizona. Ariz. Agr. Exp. Stn. Bull. 249.

Anderson, J .R., E. Hardy, J. Roach, and R. Witmer. 1976. A land-use and land-cover
classification system for use with remote sensor data. US. Geological Survey
Profession Paper 964.

Arizona Department of Water Resources (ADWR). 1990. Preliminary hydrographic
survey report for the San Pedro watershed. Adjudication Division.

Arizona Department of Water Resources (ADWR). 1999. Hydrographic survey report
for the San Pedro watershed. Adjudication Division.

Bahre, CJ. 1977. Legacy of Change: Historic Human Impact on Vegetation in the
Arizona Borderlands. University of Arizona Press: Tucson.

Bahre, C]. and ML. Shelton. 1993. Historic vegetation change, mesquite increases,
and climate in southeastern Arizona. Journal of Biogeography. 20: 489-504.

Balbach, H.E. Military Land Use. [Online] Available http://www.agronomv.org/diV's/a2.
October 16, 2000.

Bartlett, JR. 1854. Personal Narrative of Explorations and Incidents in Texas, New
Mexico, California, Sonora, and Chihuahua Connected with the United
States and Mexico Boundary Commission during the Years 1850, ’51, and
’53. 2 vols. D. Appleton & Company: New York.

Bock, J.H. and CE. Bock. 1992. Short-term reductions in plant densities following
prescribed fire in an ungrazed semidesert shrub-grassland. Southwest
Naturalist. 37: 49-53.

Boserup, E. 1965. The Conditions of Agricultural Growth. George Allen & Unwin
LTD: London.

129

Branson, FA. and IR. Owen. 1970. Plant cover, runoff, and sediment yield
relationships. Water Resour. Res. 6: 783-790.

Branson, F.A., G.F. Gifford, K.G. Kenard, and RF. Hadley. 1981. Rangeland
hydrology. Range Science Series No.1. Society for Range Management:
Denver, CO.

Bromley, D.W. 1988. Economic importance of grazing on interindustry analysis.
Arizona Ag. Exp. Sta. Technical Paper 2440.

Brown, DE. 1978. Grazing, grassland, cover, and gamebirds. Trans. N. Am. Wild].
Nat. Resour. Conf. 43: 477-485.

Brown, DE. and CH. Lowe. 1980. Biotic communities of the Southwest. US. Dept.
Of Agriculture General Technical Report RM- 78.

Brown, D.E., N.B. Cannony, and RM. Turner. 1981. Drainage map of Arizona
showing perennial streams and some important wetlands. Arizona Game and
Fish Dept., U.S. Geol. Survey map.

Caton, DD. 1985. Western livestock ranching and rangelands. ERS, USDA and
USDI.

Chavez, P.S. Jr. 1988. An improved dark-object subtraction technique for
atmospheric scattering correction of multispectral data. Remote Sensing of
Environment. 24: 459-479.

Chen, L. and L. Lee. 1992. Progressive generation of control frameworks for image
registration. Photogrammetric Engineering & Remote Sensing. Vol. 58, 9:
1321-1328.

Chisholm, M. 1962. Rural settlement and land use. Aldine Publications: Chicago, IL.

Christiansen, L.D. 1988. The extinction of wild cattle in southern Arizona. Journal of
Arizona History. 29: 89-100.

Clark, J .S. 1990. Landscape interactions among nitrogen, species composition, and
long-term fire frequency. Biogeochemistry. 2: 1-22.

Clark, R. City of Sierra Vista, AZ. [Online] Available http://www.ci.sierra—
vista.az.us. October 17, 2000.

Clary, WP. and BF. Webster. 1989. Managing grazing of riparian areas in the Inter-
mountain Region. USDA For. Serv. Gen. Tech. Rep. INT -263 .

130

Cook, CW. 1966. Factors affecting utilization of mountain slopes by cattle. Journal
of Range Management. 19: 200-204.

Cox, J.R., H.L. Morton, J.T. Labaume, and K.G. Renard. 1983. Reviving Arizona’s
rangelands. Journal of Soil Water Conservation. 38: 342-346.

Dicken, P. and Lloyd, PE. 1990. Location in space: theoretical perspectives in
economic geography. Harper Collins Publishers: New York.

Douglas, I. 1994. “Human Settlements.” In Changes in Land Use and Land-Cover:
A Global Perspective, ed. W.B. Meyer and BL. Turner.
Cambridge University Press: Cambridge.

Dregne, HE. 1983. Desertification of arid lands: Advances in deserts and arid land.
Harwood Academic Publications: New York.

Flory, EL. 1936. Comparison of the environment and some physiological responses
of prairie vegetation. Ecology. 17: 67-103.

Food and Agriculture Organization (F A0). 1995. Production Yearbook: United
Nations FAO Statistics Series, No. 48.

Food and Agriculture Organization (FAO). 1997. AFRICOVER land cover
classification. Georgetown University Press: Washington, DC.

Freethey, G.W. 1982. Hydrologic analysis of the upper San Pedro River basin from
the Mexico-US. Boundary to Fairbanks, Arizona. US. Geol. Surv. Open-file
Rpt 82- 752.

Gary, H.L., S.R. Johnson, and S.L. Ponce. 1983. Cattle grazing impact on surface
water quality. J. Soil Water Conserv. 38: 124-128.

Gifford, GP. and RH. Hawkins. 1978. Hydrologic impact of grazing on infiltration:
A criticial review. Water Resour. Res. 14: 305-313.

Glendening, GE. and HA. Paulsen, Jr. 1955. Reproduction and establishment of
velvet mesquite as related to invasion of semidesert grasslands. US.
Department of Agriculture Technical Bulletin 1 12 7.

Greenfell, W.B., B.M. Browning, and W.B. Stienecker. 1980. Food habits of
California Upland gamebirds. Calif Dep. Fish Gam. Rep. 80-1.

Grover, H.D. and HE. Musick. 1990. Shrubland encroachment in southern New

Mexico, USA: an analysis of desertification process in the American
Southwest. Climatic Change. 17: 305-330.

131

Hanson, C.L., A.R. Kuhlman, C.J. Erickson, and J .K. Lewis. 1970. Grazing effects on
runoff and vegetation. Journal of Range Management. 23: 418-420.

Hastings, J.R. and RM. Turner. 1965. The Changing Mile. University of Arizona
Press: Tucson, AZ.

Hecht, ME, and R.W. Reeves. 1981. Atlas of Arizona. Office of Arid Lands
Studies, University of Arizona: Tucson.

Herbel, CH. and AB. Nelson. 1969. Grazing management on semi-desert ranges in
southern New Mexico. USDA Agr. Res. Serv. Jornado Exp. Range Rep. 1.

Hibbert, AR. 1981. Increases in streamflow after converting Chaparral to grass.
Water Resources Res. 7: 71-80.

Hobbs, N.T., D.S. Schimel, C.E. Owensby, D.S. Ojima. 1991. Fire and grazing in the
tallgrass prairie: Contingent effects on nitrogen budgets. Ecology. 72: 1374-
1382.

Hodgson, H.J. 1976. The importance of forage in livestock production in the United
States. Annual meeting, American Society of Agronomers: Washington, DC.

Holechek, J.L., R.D. Pieper, and CH. Herbel. 1998. Range Management. Princeton
Hall: Upper Saddle River, NJ.

Hough, H. 1993. Satellite applications blossoming down on the farm. Earth Obs.
Mag. 5: 36-38.

Hughes, LE. 1998. Thirty years of rotation grazing in the Mojave Desert.
Rangelands. 20: 6-9.

Hyder, D.N. 1973. Arid shrublands. Proceedings of the third workshop of the United
States and Australia rangelands panel. Society of Range Management:
Denver, CO.

Jensen, J.R., D.C. Cowen, J. Halls, S. Narumalani, N.J. Schmidt, B.A. Davis, and B.
Burgess. 1994. Improved urban infrastructure mapping and forecasting for
Bell-South using remote sensing and GIS technology. Photogrammetric
Engineering and Remote Sensing. 60: 339-346.

Jensen, J .R. 2000. Remote sensing of the environment: an earth resource perspective.
Prentice-Hall: Upper Saddle River, NJ.

Kemp, D. 1990. Global environmental issues: a climatological approach. Routledge:
New York.

132

Kepner, W.G., D.T.Heggem, R.P. Czaja, E.J. Evanson, and OM. Edmonds. 2000.
The San Pedro River Spatial Data Archive; A Database Browser for

Community-based Environmental Protection. Presented at the Community
Assessment Workshop, Washington, DC. February 9-11, 2000.

Kimmins, J .P. 1987. Forest ecology. Macmillan: New York.

Kruse, W.H. and R. Jemison. 2000. Livestock management in the American
southwest: ecology, society, and economics. Elsevier Science BV:
Amsterdam, Netherlands.

Lacher, L.J. 1994. Hydrologic and legal issues of the Upper San Pedro River Basin,
Arizona. Presented at semi-arid land surface atmosphere program meeting,
Phoenix, AZ.

Long, J. and M. Harwell. 1993. USMAB Hurnan-Dominated Systems Directorate:
Societal Issues Workshop. Proceedings of the Second Workshop of the
USMAB Human-Dominated Systems Directorate Core Project, Easton, MD.

Lusby, GO 1970. Hydrologic and biotic effects of grazing vs. non-grazing. Journal
of Range Management. 23: 256-260.

Mackie, R.J. 1978. Impact of livestock grazing on wild ungulates. Trans. N. Am.
Wildl. Nat. Resourc. Conf 43: 462-476.

Marsett, R. 2000. Personal interview.

Martin, SC. 1975. Ecology and management of southwestern semidesert grass-shrub
ranges: the status of our knowledge. US. Department of Agriculture, Forest
Service General Technical Report RM-I 5 6.

Mauget, SA. and DR. Upchurch. ENSO-related climate and agricultural impacts
over the southwestern United States. [Online] Available
http://vawconfex2.com/ams/99annual/abstracts/892.htm. October 17, 2000.

McCalla, G.R. II, W.H. Blackburn, and LB. Merrill. 1984. Effects of livestock
grazing on infiltration rates. Journal of Range Management. 37: 265-269.

McCarty, H.H. and J .B. Lindberg. 1966. A preface to economic geography. Prentice-
Hall: Englewood Cliffs, NJ.

McCarty, K. 1976. Desert documentary: the Spanish years, 1776-1821. Historical
Monograph no.4. Arizona Historical Society: Tucson.

McClaran, Mitchel P. 1995. The desert grassland. The University of Arizona Press:
Tucson, AZ.

133

McCully, RD. 1968. 1967 Forest Research. USDA For. Serv. Pacific SW For. and
Range Exp. Sta. Phoenix, AZ.

McNeely, J .A. 1990. How conservation strategies contribute to sustainable
development. Environmental Conservation. 17: 9-13.

Meyer, W.B. and BL. Turner. 1994. Changes in land use and land cover: A global
perspective. University Press: Cambridge, Great Britain.

Moran, M.S., Y. Inoue, and EM. Barnes.1997. Opportunities and limitations for
image-based remote sensing in precision crop management. Remote Sensing
of Environment. 61 : 319-346.

Moran, M.S., R. Bryant, K. Thome, W. Ni, Y. Nouvellon, M.P., Gonzalez-Dugo, J.
Qi, and TR. Clarke. 2000. A refined empirical line approach for reflectance
factor retrieval from Landsat-5 TM and Landsat-7 ETM+. Remote Sensing of
Environment (submitted).

Mouat, DA. and CF. Hutchinson. 1995. Desertification in developed countries.
Kluwer Academics Publishers: London.

National Research Council. 1992. Restoration of Aquatic Ecosystems 168.

National Research Council. 1998. People and pixels: linking remote sensing and
social science. Diana Liverrnan, et al., eds. National Academy Press:
Washington, DC.

Officer, J. E. 1987. Hispanic Arizona, 1536-1856. University of Arizona Press:
Tucson.

Osborn, B. 1956. Cover requirements for the protection of range site and biota.
Journal of Range Management. 9: 75-80.

Phinn, SR. 1998. A framework for selecting appropriate remotely sensed data
dimensions for environmental monitoring and management International
Journal of Remote Sensing. 19: 3457-3463.

Phinn, S.R., D. Ward, T. Rowland, and M. Stanford. 1998. Remotely sensed solutions
for monitoring and managing rapidly developing coastal environments.
Proceedings of 12‘h Annual Australian ESRI and Erdas Users Conference.
Gold Coast, Australia.

134

Pluhar, J .J ., R.W. Knight, and R.K. Heitschmidt. 1987. Infiltration rates and sediment
production as influenced by grazing systems. Journal of Range Management.
40: 240-243.

Putman, F., K. Mitchell, and G. Bushner. 1988. Water resources of the Upper San
Pedro Basin. Rpt. For the Spec. Studies Sec, Hydrology Div., ADWR.

Qi, J ., R.C. Marsett, M.S. Moran, D.C. Goodrich, P. Heilman, Y.H. Kerr, G. Dedieu,
A. Chehbouni, and X.X. Zhang. 1999. Spatial and temporal dynamics of
vegetation in the San Pedro River basin area Remote Sensing of Environment.
this issue.

Qi, J ., R.C. Marsett, and P. Heilman. 2000. Forage estimation from remotely sensed
data. 2nd International Conference on Geospatial Information in Agriculture
and Forestry, Lake Buena Vista, FL.

Ratliff, RD. 1986. Cattle responses to continuous and seasonal grazing of California
annual grasslands. Journal of Range Management. 39: 482-485.

Reynolds, H.G. and SC. Martin. 1968. Managing grass-shrub cattle ranges in the
southwest. US. Dept. of Agr. Agr. Handb. 162.

Ridd, M. 1995. Exploring a V-I-S model for urban ecosystem analysis through
remote sensing: comparative anatomy of cities. International Journal of
Remote Sensing. 16: 2165-2185.

Roath, LR. and WC. Krueger. 1982. Cattle grazing influence on a mountain riparian
zone. Journal of Range Management. 35: 100-104.

Roeske, RH. and W.H. Werrell. 1973. Hydrologic Conditions in the San Pedro River
Valley. Ariz. Water Commission Bull. 4.

Rummell, RS. 1987. Range ecosystem research: The challenge of change. USDA
Agricultural Information Bulletin 346.

Sauer, CO. 1944. A geographic sketch of early man in America. Geographical
Review. 34: 529-573.

Savory, A. and SD. Parsons. 1980. The Savory grazing method. Rangelands. 2: 234-
237.

Schwartzrnan, P.W. 1990. A hydrogeologic resource assessment of the lower
Babocomari watershed. M.S. thesis. Department of Hydrology and Water
Resources. University of Arizona, Tucson.

135

Sharp, A.L., J .J . Bond, J .W. Nueberger, A.R. Kuhlman, and J .K. Lewis. 1964. Runoff
as affected by intensity of grazing on rangeland. Journal of Soil Water
Conservation. 19: 103-106.

Sinclair, R. 1967. Von Thunen and urban sprawl. Annals of the Association of
American Geographers. 57: 72-87.

Skole, D.L., C.O. Justice, J.R.G. Townshend, and AC. Janetos. 1997. A land cover
change monitoring program: strategy for an international effort. Mitigation
and Adaptation Strategies for Global Change. 2: 157-175.

Slater, P.N., Doyle, F.J., Fritz, N.L., and R. Welch. 1983. Photographic systems of
remote sensing. American Society of Photogrammetry Second Edition of
Manual of Remote Sensing. I: 23 1-291 .

Society for Range Management. 1989. A glossary of terms used in range
management. 3lrd ed. Society for Range Management: Denver, CO.

Society for Range Management. 1998. A glossary of terms used in range
management. 4th ed. Society for Range Management: Denver, CO.

Sprague, HE. 1974. Grasslands. Iowa State University Press: Ames, IA.

Stewart, O.C. 1951. Burning and natural vegetation in the United States.
Geographical Review. 41 :317-320.

Stuth, J .W. and B.G. Lyons. 1993. Decision support systems for the management of
grazing lands. Parthenon Publishing: New York.

Stuth, J.W., W.T. Hamilton, J.C. Conner, and DP. Sheehy. 1993. Decision support
systems in the transfer of grassland technology. Proceedings of XVII
International Grassland Congress. pp. 234-242.

Thomthwaite, C.W., J .R. Mather, and DB. Carter. 1978. Three water balance maps
of western North America. Johns Hopkins Press: Baltimore, MD.

Thomwaite, CW. 1979. An evaluation of cannery waste disposal by overland flow
spray irrigation, Campbell Soup Company, Paris (Texas) plant. Climatology.
22(2): 1-74.

Thrall, G.I. 1987. Land use and urban form: The consumption theory of land rent.
Methuen: New York.

Thurow, T.L., W.H. Blackburn, and CA. Taylor, Jr. 1986. Hydrologic characteristics

of vegetation types as affected by livestock grazing systems. Journal of Range
Management. 40: 322-329.

136

 

 

Tiedemann, A.R., D.A. Higgins, T.M. Quigley, H.R. Sanderson, and DB. Marx.
1987. Responses of fecal coliform in streamwater to four grazing strategies.
Journal of Range Management. 41: 322-329.

Tueller, RT. 1989. Remote sensing technology for rangeland management
applications. Journal of Range Management. 43: 442-453.

Turner, B.L., D. Skole, S. Sanderson, G. Fischer, L. Fresco, and R. Leemans. 1995.
Land-use and land-cover change. The International Geosphere-Biosphere
Program: Geneva, Switzerland.

United States Department of Agriculture (USDA). 1988. Agricultural statistics.
United States Department of Agriculture (USDA). 1997. Agricultural statistics.

Vionnet, LB. and T. Maddock. 1992. Modeling of groundwater flow and
surface/groundwater interaction for the San Pedro River basin: Part I.
University of Arizona Dept. of Hydrology and Water Resources report 92-010.

Voigt, W., Jr. 1976. Public Grazing Lands: Use and Misuse by Industry and
Government. Rutgers University Press: New Brunswick, NJ.

Wadleigh, CH. 1971. Wastes in relation to agriculture and forestry. USDA Misc.
Publ. 1065.

Wagner, T.W., and RM. Ryznar. 1999. Toward understanding urban processes with
remotely sensed data. Proceedings of Geoinformatics '99. Ann Arbor, MI. pp.
1-11.

Warren, S.D., W.H. Blackburn, and CA. Taylor, Jr. 1986. Effects of season and stage
of rotation cycle on hydrologic condition of rangeland under intensive rotation
grazing. Journal of Range Management. 39: 486-491.

Wood, J .E. 1969. Rodent populations and their impact on desert grasslands. N. Mex.
Agric. Exp. Stn. Bull. 555: Las Cruces, NM.

Wood, J.B., T.S. Bickle, W. Evans, J .C. Germany, and V.W. Howard, Jr. 1970. The
Fort Stanton mule deer herd. N Mex. Agric. Exp. Stn. Bull. 567.

Wood, MK. and W.H. Blackburn. 1981. Sediment production as influenced by
livestock grazing. Journal of Range Management. 34: 228-231.

Wright, HA, and AW. Bailey. 1982. Fire ecology: United States and southern
Canada. John Wiley & Sons: New York.

137

Y

II4i@lliljljlljiflflijljlflijlI1