SPATIAL AND TEMPORAL TRENDS OF SOIL MOISTURE IN THE GREAT LAKES
REGION OF THE USA, 1900-2008
By
Aaron Micah Pollyea

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Geography – Master of Science
2013

ABSTRACT
SPATIAL AND TEMPORAL TRENDS OF SOIL MOISTURE IN THE GREAT LAKES
REGION OF THE USA, 1900-2008
By
Aaron Micah Pollyea
Soil moisture is a key integrative variable in natural and managed ecosystems,
dependent on a location’s climate, vegetative cover, soil type, topography, and other
factors. A study of spatial and temporal variations in soil moisture in the Great Lakes
region of the USA was undertaken to better understand the impacts of varying land cover
and soil types as well as variations in climate. Soil moisture was simulated with the Soil
Water and Assessment Tool (SWAT) model for 30 different combinations of land cover
and soil types at 29 climate observation sites across the region for 1900-2008.
Significant increases in growing season soil moisture levels were found across much
of the region during the study period associated with concurrent increases in
precipitation. Some distinct changes in seasonality were also noted. The average date
during the spring season on which soil moisture fell below field capacity is gradually
occurring up to 18 days earlier in the year over the study period. Similarly, the dates at
which the minimum soil moisture occurred each warm season and the date of recharge of
soil moisture to field capacity late in the calendar year occurred up to 19 days earlier.
Collectively, the results suggest that soil moisture and plant available water across the
region have increased with time with a corresponding decrease in the risk of waterrelated stress, which has important implications for plant-based agriculture and the
management of natural resources.

To my wife Darlene, who has put up with the trials and tribulations of a lifetime
student. It must be difficult listening to a husband who talks about things you have no
knowledge of. I’m glad you had the patience and continued to listen.
Also, to my parents, Randall and Nancy Pollyea, who helped start my journey into
higher education and have continued to support me today by spurring me forward,
knowing I can be capable of anything I put my mind to. Who would have thought that a
son who wanted to be an Astronomer his entire life, with his head amongst the stars,
would fall back to Earth and into the clouds to study Climatology?

iii

ACKNOWLEDGMENTS

I would like to thank the continual, and much needed support from Dr. Jeffery
Andresen. Both his leadership and knowledge has allowed me to take the time and put
forth the effort to pursue this research and this degree. I would also like to thank the
entire Climate Office and the Geography department at Michigan State University for all
the encouragement and resources that they have provided me throughout my time
working with them.

iv

TABLE OF CONTENTS

LIST OF TABLES

vi

LIST OF FIGURES

viii

INTRODUCTION

1

METHODS

12

RESULTS AND DISCUSSION

26

SUMMARY AND CONCLUSIONS

53

APPENDICES

59

BIBLIOGRAPHY

157

v

LIST OF TABLES

Table 1:

Soil Moisture Variables (mm) averaged across the study region and across soil
types, 1979-2008
60

Table 2:

Soil Moisture Variables (mm) averaged across study region and across land cover
types, 1979-2008
60

Table 3:

Comparison of Precipitation Data from observed precipitation and the associated
grid box
60

Table 4:

Differences in Simulated and Observed Plant Available Soil Moisture in mm of
Plant Available Water
60

Table 5:

Composition of Soil Textures used for plant available water comparisons

61

Table 6:

Precipitation Breakpoints as determined from a SNHT

61

Table 7:

Relative Standard Deviations of soil moisture variables. Soil or Land Cover
groupings. ET=Evapotranspiration, PERC=Percolation, REVAP=Return Water
from Shallow Aquifer, SURQ=Surface Runoff
61

Table 8:

Average Annual Simulated Evapotranspiration (mm), 1979-2008 by land cover 62

Table 9:

Average Annual Simulated Evapotranspiration (mm), 1979-2008, by soil
texture

63

Table 10:

Average Annual Simulated Drainage (mm), 1979-2008, by land cover

64

Table 11:

Average Annual Simulated Drainage (mm), 1979-2008, by soil texture

65

Table 12:

Average Annual Simulated Runoff (mm), 1979-2008, by land cover

66

Table 13:

Average Annual Simulated Runoff (mm), 1979-2008, by soil texture

67

Table 14:

Rate of change (mm/year) of yearly soil attributes, selected climate stations,
comparison of soil textures

68

Change in annual precipitation trends (mm/year) for selected stations, 19002008

69

Table 15:

vi

Table 16:

30-year average (1979-2008) annual hydrology variables (mm) and associated
rate of change (mm/year) by land cover for selected stations
69

Table 17a:

30-year average (1979-2008) inter-annual hydrology variables and associated rate
of change by land cover for selected stations
70

Table 17b:

30-year average (1979-2008) inter-annual hydrology variables and associated rate
of change by land cover for selected stations
71

Table 18:

Slopes of annual evapotranspiration trends, by land covers of selected stations
72
(mm/year)

Table 19:

Slopes of annual percolation trends by land covers of selected stations (mm/year)
73

Table 20:

Slopes of annual water return trends, by land covers of selected stations
(mm/year)

Table 21:

74

Slopes of annual runoff trends, by land covers of selected stations (mm/year) 75

Table 22:
Intercepts of regression trends of inter-annual soil moisture variables, by land
covers of selected stations. Average of SL, L, and SiL soil type
76
Table 23:

Slopes of trends of inter-annual soil moisture variables, by land covers of selected
stations, Average of SL, L, and SiL soil types
77

Table 24:

Temporal scale, in months, of plant available soil moisture for selected stations.
All soils and land covers
78

Table 25:

Rate of Change of Yearly Average Temperatures ( C / year). Spearmans rs level
of significance
79

Table 26:

Rate of Change of Yearly Precipitation (mm/ year). Spearmans rs level of
significance

o

vii

81

LIST OF FIGURES

Figure 1:

An idealized annual soil moisture curve showing relative levels of soil moisture,
Evapotranspiration (ET), and Potential Plant Moisture Stress (PMS). Taken from
Roadside Revegetation: An Integrated Approach to Establishing Native Planets.
Chapter 5, Figure 5.22.
82

Figure 2:

Average Annual precipitation (mm) from 1979-2008 across the study region.

83

Figure 3:

Average Annual Temperatures (F) from 1979-2008 across the study region.

84

Figure 4:

Water Holding Capabilities of Soils by texture. (Figure 5.9. As seen in Foth,
1990)

85

Figure 5:

Average Annual Evapotranspiration (mm) for a grassland land cover on a loamy
86
soil, 1979-2008.

Figure 6:

Average Annual Runoff (mm) for a grassland land cover on a loamy soil, 19792008.
87

Figure 7:

Average Annual Drainage (mm) for a grassland land cover on a loamy soil, 19792008.
88

Figure 8:

Example of an annual soil moisture curve, 1979-2008, South Haven, MI. Corn
land cover and Loam soil type.
89

Figure 9:

Average Date of Soil Moisture Discharge averaged across soil textures, Pasture
Land Cover, 1979-2008.
90

Figure 10a:

Average Date of Minimum Soil Moisture, averaged across soil textures, Corn
land cover, 1979-2008.
91

Figure 10b:

Average Date of Minimum Soil Moisture, averaged across soil textures, Wheat
land cover, 1979-2008.
92

Figure 11:

Average Minimum Soil Moisture amount, averaged across soil textures, Wheat
land cover, 1979-2008.
93

Figure 12:

Average Date of Soil Moisture Recharge, averaged across soil textures, Pasture
land cover, 1979-2008.
94

viii

Figure 13:

Average Annual Soil Moisture Days, averaged across soils, Pasture land cover,
1979-2008.
95

Figure 14:

Trends in Temperature (C/year) across the study area, 1900-2008.

96

Figure 15:

Great Lakes Drainage Basin.

97

Figure 16:

Center of Grids from Midwestern Regional Climate Center. For Interpretation of
the references to color in this and all other figures, the reader is referred to the
electronic version of this thesis.
98

Figure 17:

Map of Historical Climate Network Observation Sites used for this study.

Figure 18a:

Comparison of SWAT plant available soil moisture to Observed plant available
soil moisture. Bondville, IL.
100

Figure 18b:

Comparison of SWAT plant available soil moisture to Observed plant available
soil moisture. Champaign, IL.
101

Figure 18c:

Comparison of SWAT plant available soil moisture to Observed plant available
soil moisture. Topeka, IL.
102

Figure 19:

Rate of Change: Precipitation (mm/year) 1900-2008. Contours are at 0.5 mm/year
intervals.
103

Figure 20:

Rate of change of Evapotranspiration. Average of all soils, Corn land cover
(mm/year), 1900-2008. Contours are 0.15 mm/year.

99

104

Figure 21:

Rate of change of Evapotranspiration. Average of all soils, Forest land cover
(mm/year). Contours are 0.15 mm/year.
105

Figure 22:

Rate of change of Evapotranspiration. Average of all soils, Pasture land cover
(mm/year). Contours are 0.15 mm/year.
106

Figure 23:

Rate of change of Evapotranspiration. Average of all soils, Urban land cover
(mm/year). Contours are 0.15 mm/year.
107

Figure 24:

Rate of change of Evapotranspiration. Average of all soils, Wheat land cover
(mm/year). Contours are 0.15 mm/year.
108

Figure 25:

Rate of change of Percolation. Average of all soils, Corn land cover (mm/year).
Contours are 0.5 mm/year.
109

ix

Figure 26:

Rate of change of Percolation. Average of all soils, Forest land cover (mm/year).
Contours are 0.5 mm/year.
110

Figure 27:

Rate of change of Percolation. Average of all soils, Pasture land cover (mm/year).
Contours are 0.5 mm/year.
111

Figure 28:

Rate of change of Percolation. Average of all soils, Urban land cover (mm/year).
Contours are 0.5 mm/year.
112

Figure 29:

Rate of change of Percolation. Average of all soils, Wheat land cover (mm/year).
Contours are 0.5 mm/year.
113

Figure 30:

Rate of change, return of soil moisture from shallow aquifer. Average of all soils,
Corn land cover (mm/year). Contours are 0.01 mm/year.
114

Figure 31:

Rate of change, return of soil moisture from shallow aquifer. Average of all soils,
Forest land cover (mm/year). Contours are 0.01 mm/year.
115

Figure 32:

Rate of change, return of soil moisture from shallow aquifer. Average of all soils,
Pasture land cover (mm/year). Contours are 0.01 mm/year.
116

Figure 33:

Rate of change, return of soil moisture from shallow aquifer. Average of all soils,
Urban land cover (mm/year). Contours are 0.01 mm/year.
117

Figure 34:

Rate of change, return of soil moisture from shallow aquifer. Average of all soils,
Wheat land cover (mm/year). Contours are 0.01 mm/year.
118

Figure 35:

Rate of change, Runoff. Average of all soils, Corn land cover (mm/year).
Contours are 0.25 mm/year.

119

Rate of change, Runoff. Average of all soils, Pasture land cover (mm/year).
Contours are 0.25 mm/year.

120

Rate of change, Runoff. Average of all soils, Urban land cover (mm/year).
Contours are 0.25 mm/year.

121

Rate of change, Runoff. Average of all soils, Wheat land cover (mm/year).
Contours are 0.25 mm/year.

122

Figure 36:

Figure 37:

Figure 38:

Figure 39:

Rate of change, Day of Year of the beginning of soil water discharge. Average of
SL, L, and SiL. Corn land cover (days/year). Contours are 0.05 days/year.
123

Figure 40:

Rate of change, Day of Year of the beginning of soil water discharge. Average of
SL, L, and SiL. Forest land cover (days/year). Contours are 0.05 days/year. 124
x

Figure 41:

Rate of change, Day of Year of the beginning of soil water discharge. Average of
SL, L, and SiL. Pasture land cover (days/year). Contours are 0.05 days/year. 125

Figure 42:

Rate of change, Day of Year of the beginning of soil water discharge. Average of
SL, L, and SiL. Urban land cover (days/year). Contours are 0.05 days/year. 126

Figure 43:

Rate of change, Day of Year of the beginning of soil water discharge. Average of
SL, L, and SiL. Wheat land cover (days/year). Contours are 0.05 days/year. 127

Figure 44:

Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Corn
land cover (µm/(days*year)). Contours are µm /(days/year).
128

Figure 45:

Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Forest
land cover (µm/(days*year)). Contours are µm /(days/year).
129

Figure 46:

Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Pasture
land cover (µm/(days*year)). Contours are µm /(days/year).
130

Figure 47:

Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Urban
land cover (µm/(days*year)). Contours are µm /(days/year).
131

Figure 48:

Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Wheat
land cover (µm/(days*year)). Contours are µm /(days/year).
132

Figure 49:

Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and
SiL. Corn land cover, days/year. Contours are 0.05 days/year.
133

Figure 50:

Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and
SiL. Forest land cover, days/year. Contours are 0.05 days/year.
134

Figure 51:

Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and
SiL. Pasture land cover, days/year. Contours are 0.05 days/year.
135

Figure 52:

Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and
SiL. Urban land cover, days/year. Contours are 0.05 days/year.
136

Figure 53:

Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and
SiL. Wheat land cover, days/year. Contours are 0.05 days/year.
137

Figure 54:

Rate of change, Minimum Yearly Soil Moisture value. Average of SL, L, and SiL,
Corn land cover, mm/year. Contours are 0.1 mm/year.
138

xi

Figure 55:

Rate of change, Minimum Yearly Soil Moisture value. Average of SL, L, and SiL,
Forest land cover, mm/year. Contours are 0.1 mm/year.
139

Figure 56:

Rate of change, Minimum Yearly Soil Moisture value. Average of SL, L, and SiL,
Wheat land cover, mm/year. Contours are 0.1 mm/year.
140

Figure 57:

Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Corn land
cover (µm/(days*year)). Contours are µm /(days/year).
141

Figure 58:

Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Forest
land cover (µm/(days*year)). Contours are µm /(days/year).
142

Figure 59:

Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Pasture
land cover (µm/(days*year)). Contours are µm /(days/year).
143

Figure 60:

Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Urban
land cover (µm/(days*year)). Contours are µm /(days/year).
144

Figure 61:

Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Wheat
land cover (µm/(days*year)). Contours are µm /(days/year).
145

Figure 62:

Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and
SiL. Corn land cover, days/year. Contours are 0.1 days/year.
146

Figure 63:

Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and
SiL. Forest land cover, days/year. Contours are 0.1 days/year.
147

Figure 64:

Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and
SiL. Pasture land cover, days/year. Contours are 0.1 days/year.
148

Figure 65:

Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and
SiL. Urban land cover, days/year. Contours are 0.1 days/year.
149

Figure 66:

Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and
SiL. Wheat land cover, days/year. Contours are 0.1 days/year.
150

Figure 67:

Rate of change, Soil Moisture Days. Average of SL, L, and SiL. Urban land
cover, (soil moisture days / days) / year. Contours are 0.1 (soil moisture days /
days) / year.
151

Figure 68:

Rate of change, Soil Moisture Days. Average of SL, L, and SiL. Wheat land
cover, (soil moisture days / days) / year. Contours are 0.1 (soil moisture days /
days) / year.
152

xii

Figure 69:

Rate of change: Winter Precipitation (mm/year) 1900-2008. Contours are at 0.25
mm/year intervals.
153

Figure 70:

Rate of change: Spring Precipitation (mm/year) 1900-2008. Contours are at 0.25
mm/year intervals.
154

Figure 71:

Rate of change: Summer Precipitation (mm/year) 1900-2008. Contours are at 0.25
mm/year intervals.
155

Figure 72:

Rate of change: Fall Precipitation (mm/year) 1900-2008. Contours are at 0.25
mm/year intervals.
156

xiii

INTRODUCTION

The Great Lakes region is an area dominated by large open bodies of water, large
agricultural areas due to its fertile soils, a wide range of both man-made land covers such
as field crops and urban environments, as well as more natural land covers such as
heavily forested areas and open plains. This region was originally defined as the states of
Illinois, Indiana, Iowa, Michigan, Minnesota, Ohio, Pennsylvania, and Wisconsin by the
United States Great Lakes Commission in 1955, though the Canadian provinces of
Ontario and Quebec also sit on this commission as they are directly tied to the Great
Lakes and associated waterways. The Great Lakes drainage basin is limited by the
Mississippi Drainage basin to the west and the Ohio drainage basin to its south. This
basin contains the entire state of Michigan and the areas of the Great Lakes states closest
to the Lakes (Figure 28).
The climate of the region is influenced by long waves in the upper air flow and the
warm season polar jet stream. These meteorological features are caused by temperature
gradients in the west to east movement of the upper air (Rossby 1939). Variability in the
Great Lake regions climate can be attributed to the position and strength of these long
waves and the jet stream. These same features also affect which sources of air move into
the region, be it warm moist air from the Gulf of Mexico, cool moist air from the
Canadian Rocky Mountains, or cool dry air from Hudson Bay to the north or the Pacific
Northwest. As this region is far away from the Gulf of Mexico, which is the source of
much of the atmospheric moisture east of the Rocky Mountains, many of the states have

1

some of the lowest amounts of annual precipitation in the United States, with Michigan
40th, Wisconsin 43rd, and Minnesota 47th out of 50. Short time-frame synoptic weather
events such as cyclones (low pressure systems), anti-cyclones (high pressure systems)
and short-waves (a feature that moves along a long-wave) also impact the climate of the
region. Cyclones are very important to the annual precipitation totals in the study region
as they can contribute over half of the annual precipitation totals. Mescoscale convective
systems also contribute a significant portion of precipitation to the region, about 10% of
the warm season totals. These events are large scale clusters of thunderstorms that can
last for several hours or more. The largest portion of the annual precipitation totals occur
during the mid to late warm season, thus these events have a large impact on the
precipitation – evapotranspiration deficit. (Andresen and Winkler 2009)
The climate of the region is also directly influenced by the lakes. The lakes moderate
temperatures in both the warm and cool seasons providing a more mild climate with less
extremes downwind of the lakes when compared to areas further west such as the
northern Great Plains and other continental climates. This moderation is due to the delay
that the Lakes generally reach their maximum and minimum temperatures, generally 4 to
6 weeks after when the atmosphere reaches those values (Eichenlaub et al. 1990). The
lakes also produce ‘Lake Effect’ snow during the cold season when cool air moves across
the lakes and is warmed by the lake surface where it picks up more moisture. This
moisture is then deposited when the land surface cools the air once again, mainly in the
regions downwind of Lake Michigan and Lake Superior, with an increase in snowfall of
between 25-30% (Changnon and Jones 1972). The lakes also produce changes in

2

atmospheric drag in the air moving across them. When the air moves across the lake and
then impacts on the downwind shoreline there is a reduction of the wind speed and could
possibly produce vertical motion that goes on to produce lake effect clouds, most of them
occurring in the autumn and cold season when the region is experiencing northwesterly
winds. All of these modifications can occur together and produced enhanced results.
(Andresen and Winkler 2009)
The land covers of the region can be described as mainly agricultural, with the land
surface area for corn, soybeans, and pasture land accounting for nearly 20%, 15%, and
10% respectively, adding up to slightly over 40% of the land surface area when including
all agricultural products. This is followed by forested areas, both evergreen and
deciduous forests, that account for over a quarter of the surface, and then by open/pasture
land with over 10% and finally urban areas with close to 10%. The soils are just as
diverse. When looking at the soils that make up more than half of the land area in the
Great Lakes we see that over 20% of this area is covered by Sand, Sandy Loam, and Silty
Loam each. Loam soils make up slightly more than 12%, Clay Loams about 4%, and
Silty Clay Loams make up only slightly more than 2%.
The Great Lakes region is an important component of our nations agricultural output.
In 2007, the states of Indiana, Illinois, Iowa, Michigan, Minnesota, New York, Ohio,
Pennsylvania, and Wisconsin combined produced over 63% of the US production of
grain corn, 35% of the nation’s acreage of apple orchards, and over 87% of the nation’s
acreage of tart cherry orchards to give some examples (USDA/NASS Census of
Agriculture 2007). Since evapotranspiration is greater than the rate of precipitation during

3

the warm season, all of these agricultural products depend on plant available water
located in the soil. Georgakakos et al (1995) correctly stated that soil moisture found in
the upper 2 meters of the soil column is the “primary hydrologic variable that controls
and is controlled by land surface processes.”
Lack of this soil moisture can lead to lower crop yields or total crop loss in severe
cases (Mishra and Cherkauer 2010). Drought, or the deficit of soil moisture caused by
evapotranspiration outweighing precipitation on a predefined timescale, severity, and/or
spatial scale, can be one of the most devastating natural disasters in economic terms.
Sawyer (1964) said that almost half of the incoming solar radiation during the summer in
mid-latitude regions goes directly into evapotranspiration. The estimated loss for the
1988 drought alone was $40 billion (American Meteorological Society, 1997). With the
Great Lakes region being heavily dependent on agriculture, the idea of drought and lack
of adequate soil moisture is one of the most important issues facing agricultural
production and the economy of the Great Lakes.
Shown below is an idealized soil moisture curve that would be typical for many
Northern Hemisphere locations including the Midwest of the United States (Figure 1).
This simplified graph shows the reduction of moisture in the soil column during the
spring due to increasing temperatures and increasing water use by a vegetative land cover
(ET, i.e. evapotranspiration). It also shows the recharge of moisture in the soil column
during the early fall to winter due to decreasing vegetative water use and cooling
temperatures leading to lower ET. Finally, this image also shows a potential plant

4

moisture stress that is caused by a lack of water that is easily accessible to a generic plant
at the surface. The basic concepts of this image will be revisited later in the paper.
Previous studies of Drought and Pluvial
First we must define plant available soil moisture as it is defined differently than soil
moisture. Soil moisture is the total amount of water in the soil column. Rodriguez-Iturbe
and Porporato (2004) correctly state that soil moisture is “the center of the hydrologic
cycle”. This is in contrast to plant available soil moisture which measures the moisture
that a plant can draw out of the soil column, which means that plant available soil
moisture is a lower total then the values given by soil moisture. What this also means that
while there can be water in the soil column, a plant may not be able to utilize any of it
and will become water stressed due to a decrease in transpiration.
th

During the 20 Century there have been many definitions of drought for many
different purposes and all using different techniques in an attempt to quantify the deficit
of precipitation compared to evapotranspiration. Landsberg (1982) summarized it well
when he said, “No universal definition of drought exists…one can say that drought is a
term generally restricted to land areas where some form of agriculture, horticulture or
silviculture is possible.” But still this doesn’t cover many of the ways the term drought
has been used. Heim (2002) better summarizes drought categories into four types:
meteorological, hydrological, agricultural, and socio-economic. What all of these drought
indices try and bring together is the idea that water entering the soil column, from above
as precipitation or from below as water returning from a shallow aquifer, has an impact

5

on the quality of the land or the quality of land use. The most direct way of describing
these impacts is through understanding or simulating plant available water.
Rodriguez-Iturbe (2000) summed up soil moisture well when he said that one could
provide a “description of soil moisture at a point as a function of climate, soil, and
vegetation.” The attempt to describe drought or the reduction of plant available soil
moisture due to differences between precipitation and evapotranspiration over a large
surface area and a large period of time is difficult as that none of these indices mentioned
previously can perform an adequate job of comparing one site to another. Nor can an
adequate job attempting to describe an abundance of moisture in the soil column.
Beven and Kirby (1979) were able to show that topography had a significant impact
on the scale at which soil moisture varies spatially, both in surface roughness and the
overall slope of the landscape. Topography was then grouped with other variables such as
vegetation, soil texture and other soil characteristics. Everything else was considered
atmospheric forcing, i.e. meteorology including precipitation or melt water, temperature,
wind speed and relative humidity, evapotranspiration, and finally incoming solar
radiation. These two groupings had different temporal and spatial scales.
Rodriguez-Iturbe and Porporato (2004) said that “The interaction between water
balance and plants is responsible for some of the fundamental differences among various
biomes and for the developments of their space-time patterns.” Differences in soil
characteristics, even inside the same soil class, and even inside the same field, can be
enough to impact soil moisture calculations. Differences in land cover, such as tree

6

density, tree albedo, and differing slopes also can lead to differences in soil moisture and
associated indices in the same stand of trees.
Plant available soil moisture, and soil moisture overall, have a direct impact on many
climate variables including precipitation, temperature, and relative humidity, and vice
versa. Rand (1982) determined that soil moisture was important in forecasting summer
time temperatures and precipitation. A negative deviation from normal resulted in higher
temperatures in the early summer months and decreased the amounts of precipitation.
Rowntree and Bolton (1983) also concluded in their study of atmospheric responses to
soil moisture in Europe that soil moisture can have a significant impact on precipitation,
relative humidity and temperature in the area of the region in question. Delworth and
Manabe (1988) came to similar conclusions when they found that soil moisture is
important in determining the variability of summer time air temperatures. They also
showed that changes in soil moisture can directly lead to air temperature variation by
changing the proportion of energy transfer between latent and sensible fluxes. Findell and
Eltahir (1997) found in their study of soil moisture feedback with precipitation that there
was a significant linear correlation between soil saturation conditions and later
precipitation during the summer months. Pan et al (1995) also came to a similar
conclusion in a two year study. Georgakakos et al (1995) and Cayan and Georgakakos
(1995) found that there was a non-linear and complex relationship between soil moisture
and both temperature and precipitation, i.e. evapotranspiration.
Lettenmaier et al (1994) showed that when studying the Historical Climate Network
observation sites from 1948 to 1988 that there was increasing temperatures west of Lake

7

Michigan and increasing precipitation west of Lake Erie. They also found that the
increases in precipitation took place in the non-winter months. Karl and Knight (1998)
came to the conclusion in their study stating, “Increases in total precipitation are strongly
affected by increases in both frequency and intensity of heavy and extreme precipitation
events.” They also found that the frequency of precipitation has increased by over 6 days
per century. Groisman et al (2004) found that the trend in precipitation showed an
increase of between 7% and 15% nationwide as well as a 14% to 20% increase in Heavy
th

th

(95 percentile) and Very Heavy (99 percentile) precipitation events. The increase in
these events can increase the amount of surface run-off and change the water balance of a
soil column. This same paper showed that an earlier spring thaw, of two to three weeks,
allowed snow cover and spring to have an earlier onset. Giorgi (2000) and Groisman et al
(2004) found that there were significant warming trends over North America and a
positive trend in precipitation.
These trends in precipitation can have significant effects on soil moisture not just due
to the increase in incoming water to the top of the column, but also reducing the amount
of evapotranspiration that can occur due to lower incoming solar radiation. There is also a
non-linear relationship between precipitation and surface runoff that is dependent on the
saturation of the soil column (Georgakakos et al 1995). A large amount of rainfall can
produce very little flooding or surface run off if the soil is well below field capacity, but
even a moderate amount of rainfall can have the opposite outcome if the soil is near
saturation.

8

Plant available soil moisture is vitally important to agriculture, Hollinger and Isard
(1994) stated this fact. But accurate measurements of soil moisture are difficult at both
the spatial and temporal scale to simulate due to microclimate, highly variable soils (all
with different water holding characteristics), and highly variable land covers. These
highly variable soils and land covers are the primary difficulty in any comparison of soil
moisture and how it is changing over time.
Previous studies with soil moisture and soil moisture simulation.
Of particular interest to this study is the paper by DeLiberty and Legates (2003) in
which they determined if soil moisture could be simulated with readily available data and
to determine soil moisture climatology in Oklahoma using SWAT (Soil Water
Assessment Tool). This is similar to this study, but with a more limited geographical
range and different region than the Great Lakes (the state of Oklahoma) and a more
limited time scale (1960 to 1991). They also used specific soil and land cover types at
each of their climate observation sites. In order to geographically represent their findings,
they used an inverse distance weighting method to interpolate their simulated data. The
results of their study found that there was a significant impact on a year’s average soil
moisture when there was a drier autumn that had a lower amount of recharge.
Along with Entin et al (2000), they examined the temporal scale at which soil
moisture anomalies occur. DeLiberty and Legates found a scale of 1.8 months in
Oklahoma and Entin et al found a scale of approximately 1.5 to 2.0 months in Iowa and
2.0 to 2.3 months for Illinois. Narasimhan et al (2005) used SWAT in their paper to
provide a simulation for six drainage basins in Texas. They found that SWAT worked

9

well for simulating stream flow and its simulated soil moisture was well correlated with
NDVI, a satellite measured vegetation index used to determine plant health.
In a paper on soil moisture in root-zone soil moisture, Jayawickreme et al (2008)
studied the interaction of land cover over different seasons in East Lansing, MI. They
found that during the winter soil moisture moved downwards in the soil column. During
the spring thaw soil conditions were stable or increasing in soil moisture content due to
snow melt, and in the grassland land cover, were at or near field capacity. Using the VIC
(Variable Infiltration Capacity) model to generate soil moisture, Sheffield and Wood
(2008) found that the central United States had increasing soil moisture on the order of
0.05% to 0.2% volume per year, and that these were correlated significantly with
increasing trends in precipitation.
Objective questions to be answered
In this study, using individual climate observation site data, there are multiple
subjects explored. The study will explore how precipitation is changing over the time
frame of 1900-2008 as well as spatially in the Great Lakes region. This study will then
examine how plant available soil moisture is changing both temporally and spatially over
the same study region using multiple generalized soils and differing land covers. Then the
question of how the overall wetness of a year is changing over time along with if dry
conditions over the Great Lakes area are getting longer and drier or shorter and wetter
will be answered. Finally, the temporal scale of soil moisture anomalies will also be
examined.

10

This knowledge is of great importance as knowing the overall trend of how soil
moisture is changing could affect the dates in which agricultural crops are planted and
possible water stress on those plants. Knowing water stress on plants can also affect the
amount of irrigation that producers may need to use. This is a vital piece of information
for the economics of agriculture due to increased or decreased water stress on the crop,
the frequency and amount of irrigation needed, and the ongoing debate of water rights.

11

METHODS
In general, comprehensive observations of soil moisture are not available over
extended periods of time in the Great Lakes region study area. For this study, a processbased simulation model, the Soil Water Assessment Tool (SWAT) was used to estimate
soil moisture in the top 150 cm of the soil profile across a wide variety of representative
climates, soils, and land cover types across the project domain. While the use of
simulation models may provide good estimates of reality under a wide range of
hypothetical conditions, such efforts are complicated by both the very large number of
processes and phenomena involved in the system dynamics and = extremely large
variability in time and space.(Rodriguez-Iturbe and Porporato, 2004)
Use of SWAT (Soil Water Assessment Tool)
The Soil Water Assessment Tool (SWAT) is a process-based simulation model
originally developed by the USDA that is used to simulate soil moisture with soils and
land covers over large watersheds and over long continuous periods of time. The original
version of SWAT is a successor of the USDA’s CREAMS model (Chemicals, Runoff,
and Erosion from Agricultural Management Systems (Knisel et al. 1980) as well as other
precursor models including GLEAMS, EPIC, and SWRRB (Arnold et al. 1998).
The primary physical strategy behind the SWAT model is the standard water balance
t

state equation: SWt = SWt-1 + ∑ t-1 (Ri – Qi – ETi – Pi – QRi).
Where SW is plant available soil water, R is precipitation, Q is runoff, ET is
evapotranspiration, P is percolation and QR is return flow. SWAT can simultaneously
simulate of to eight large segments of a watershed of a sub-basin, referred to as HRUs, or
12

Hydrological Response Units. The properties of the land cover and soils are either
predefined in the models data base or they can be user defined. Further details describing
the SWAT model can be found in Arnold et al (1998).
SWAT has been used in a wide variety of applications since its development ranging
from soil erosion to estimation of soil moisture to river basin hydrology to other waterrelated applications (Gassman et al 2007). Several previous studies have used SWAT to
simulate soil moisture, with the model performing well under a number of applications.
For example, Narasimhan and Sirnivasan (2005) used SWAT to simulate soil moisture in
Texas in order to estimate the Soil Moisture Deficit Index (SMDI) and the
Evapotranspiration Deficit Index (ETDI). They concluded that SWAT preformed well as
determined by the correlation with NDVI (Normalized Difference Vegetation Index) for
agriculture and pasture type land covers. DeLiberty and Legates (2003) used SWAT in a
study of soil moisture variability and soil moisture climatology to estimate the temporal
scale of soil moisture variation in Oklahoma for both historical and projected future time
frames. They found that SWAT estimated soil moisture magnitude and variability
accurately over a wide range of conditions. SWAT has been commonly used in Total
Maximum Daily Load (TMDL) analyses, assessing the effectiveness of USDA soil
conservation efforts, and to perform large scale assessments of base flow and snow melt
for very large river basins (Gassman et al 2007). However, the model does require a
significant amount of input parameterization and other information in order to operate
properly. In addition, the developers of the model (Arnold et al 1998) stated that while
SWAT uses easily available inputs for large areas, is computationally simple to run on

13

large basins, and simulate for long periods of time, it also requires proper calibration. A
summary of previous studies using SWAT can be found in Gassman et al (2007).
In this study, the SWAT model was set up to simulate soil moisture across the Great
Lakes region for a number of climates, soils, and land cover types. Thirty HRUs were
utilized for the simulations. The time period of 1900-2008 was chosen for the study time
frame, which includes the great historic droughts of the 1930’s across the region
(Andresen et al., 2001) as well as recent major droughts in 1988 and 2005 (Mishra and
Cherkauer, 2010) and a relatively wet period during the past 1-2 decades (Andresen,
2012). The start of the study at the beginning of the 20th century also allows examination
of trends that began before the 1930’s time frame back to the beginning of the period of
large numbers of reliable instrumental records across the region (Andresen, 2012).
Climate Data
Climate variables needed by SWAT as input are daily maximum and minimum
temperatures, precipitation, total solar radiation on a horizontal surface, average relative
humidity and average wind speed. The primary data needed for simulation in SWAT,
daily maximum and minimum temperatures and total precipitation, were obtained from
the Daily Global Historical Climate Network (GHCN), a high quality, long term record
of daily climate data across the USA (Menne et al 2011). For the study time frame, 29
climate observation sites were chosen for their geographical distribution and climate
record completeness over the Great Lakes region and surrounding areas (Figure 17). The
geographical range of the climate observation sites spans an area from 40.6 to 47.0
degrees north latitude and 75.9 to 94.1 degrees west longitude. At least one station site

14

was chosen for each state in the Great Lakes region as well as bordering states if they
were available for the time period. Both Clinton and Fayette, Iowa as well as Pine River,
Minnesota (all to the west of the actual Great Lakes basin itself) were included in this
study to better account for the east to west precipitation gradient that exists across the
region. Besides the spatial distribution of the observation sites, the other major criterion
utilized in the selection of study sites was the amount of missing data in the site records
as determined by data analysis in a database program by the author. Other input climate
variables, solar radiation, relative humidity, and wind speed, were generated
synthetically. Solar radiation was simulated with the WGEN stochastic weather generator
(Richardson and Wright 1984) following the method of Andresen et al (2001). Daily
average wind speed and relative humidity, needed to estimate potential
evapotranspiration (ET) in the model using the Penman-Monteith estimation method,
were generated synthetically using the WXGEN climate generator given information
from a statistical database for the USA (Sharply and Williams, 1990). Missing climate
data estimates were obtained with data from representative neighboring observing based
on distance to the original observation site and similarities in terrain.
Observed climate variables from climate stations that are a part of the Historical
Climate Network were used rather than upscaled gridded data.
Two spatial scales of climate data were initially considered for use in this study:
gridded and point level data (i.e. data from individual climate observation sites). Many
studies of drought and soil moisture content have utilized data from individual locations
(i.e. climate observing sites), and then spatially averaged over grid cells of varying scales.

15

th

In their study of 20 century drought in the US, Andreadis et. al (2005) used 2489
weather stations for daily precipitation and 1904 stations for daily temperature data that
were then averaged over a 0.5 degree resolution grid. DeLiberty and Legates (2003), in
their study on variability of soil moisture, used an inverse-weighted method for
interpolating data from individual climate observation sites and concluded that it
overcame several significant limitations of traditional weighted-average approaches,
accounted for the sphericity of the Earth, and was one of the most reliable techniques for
estimating daily precipitation. On the other hand, gridded data offer the potential
advantage of spatially and temporally complete series records, even though the base data
on which they are based may vary over time.
In a preliminary comparison study of point-based vs. gridded data to determine which
data set would be more representative, Upscaled, gridded climate data was obtained from
Midwestern Regional Climate Center (http://mcc.sws.uiuc.edu/) for the time period of
1900-2008. Data was provided with a scale of 2/3

rds

of a degree latitude and 1 degree of

longitude (Figure 16 for map of the locations of the center of the grids). This data was
then analyzed for any inconsistencies, errors and incongruities which then showed several
things that are worth noting.
The interpolation of precipitation records from observation sites (data obtained from
Historical Climate Network) inside the grid boxes (Kunkel et. al. 2005) caused the ratio
of wet days with recorded precipitation to dry days that were much higher than expected
(Table 3) for the time period of 1900-2009 and across a selection of 10 stations across the
study region. Of the grid boxes checked the highest ratio seen was 2.08 times greater than
16

observed at Rogers City, MI and the lowest ratio was 1.57 times greater at both Alpena
and Ann Arbor, MI. In contrast, it was seen that the precipitation totals were fairly close
to the measured amounts with values ranging from 0.94 (Grayling, MI) to 1.16 (Wilkes
Barre, PA) times the normal observed amounts.
Based on the results of the preliminary study, the individual, point-based data were
chosen for this study due to the relative increase in precipitation frequency and reduction
in precipitation amounts associated with gridded data, which could lead to a number of
changes in the simulation including reduced runoff for the gridded data and interception
of precipitation by the plant canopy, changes in antecedent soil moisture and infiltration,
and potential changes in the amount of Potential Evapotranspiration (PET). It was also
observed that in some cases the gridded minimum daily temperature exceeded the
maximum daily temperature (due to observation time differences in station series used to
compute the spatial averages). This problem was found to be relatively more common
early in the study period of record when there were lower numbers of climate observation
sites available inside the grid box.
One of the most difficult challenges in simulating soil moisture is accounting for
physical soil characteristics that may vary considerably over only short distances
(Cambardella et al 1994). Over a large study area such as the Great Lakes region,
SWAT’s computationally heavy processes could be a liability (Arnold et. al. 1998). To
account for the potentially large variations in soils across the study domain, SWAT was
run at each individual site for 6 soils representing a textural range from a coarse-textured
sand to a fine-textured clay loam, which are representative of the wide variety of soil

17

types found in the Great Lakes basin. Each soil was defined in the simulation to be 1500
mm in depth and with homogeneous textural characteristics throughout the column. Even
though a homogeneous soil column is unlikely to be observed, the model estimations are
then more directly comparable between study sites. The physical characteristics of these
soils can be found in Table 5, with a total range of 107 mm (sand) – 309 mm (silt clay
loam) of plant extractable water for a 1500 mm profile. The soils used will often be
referred to in the text and tables of this paper by abbreviations in order to be succinct.
Sand, S; Sandy Loam, SL; Loam, L; Silt Loam, SiL; Silt Clay Loam, SiCL; Clay Loam,
CL. For all simulations, a single small, relatively flat basin with less than 1 degree
average slope was assumed at each study site.
SWAT also takes land surface vegetative cover type into account as an input variable.
Five land cover types were chosen represent typical land covers found across the Great
Lakes region. As noted previously, the Great Lakes region is home to a significant
portion of the nation’s agriculture. To simulate soil moisture on agricultural landscapes,
three land covers were selected: corn (for grain), grass-based pasture, and winter wheat.
Winter wheat was specifically chosen as it is a fall-planted crop unlike the others which
are planted in the spring season. The two other land covers used were mixed forest
(which includes oak and pine) and urban (medium-density residential). These land covers
were used to represent a range of common surface types in the region, but also to
examine the hydrologic differences between differing surface types (e.g. runoff from
forested surfaces tends to be relatively lower than runoff from paved urban areas. For all
cover types except corn, the physical parameterization provided in the SWAT modeling

18

software was used. For corn more modern values typical of current productions systems
2

were used. The biomass to energy ratio was increased from 39 to 45 (kg/ha)/(MJ/m )
(Loomis and Amthor 1999), the maximum leaf area was increased from 3.00 to 5.35, and
maximum canopy height and maximum rooting depth were changed to 2.75m and 1.5m
respectively in order for the model to simulate realistic biomass totals (Grant and Hesketh
1992 for canopy height and LAI, and Canadell et al 1996 for rooting depth). Applied
amounts of nitrogen and phosphorus were each assumed to be unlimited in this study, and
the model was automatically set to apply fertilizer anytime nutrient stress was detected.
This procedure was applied to isolate the effects of climate, soil type, and land cover
type.
Prior to the application of SWAT across the study domain, the model was calibrated
and validated with a set of observed soil moisture data obtained from the Illinois Climate
Network (Hollinger et. al 1994), which maintains 17 observation sites across the state of
Illinois. Observation frequency at the network sites varies from monthly or bi-weekly
observations to daily. Three soil moisture observation series were chosen from the dataset
based on their observation frequencies, periods of record, and soil types: Bondville,
Champaign, and Topeka. Soil series at these observation sites are Elburn silt loam,
Drummer silty clay loam, and Uldolpho Fine Sandy Loam, respectively. Soil moisture
and climate data was collected from each site for the period of record of 1989-2009.
Climate data was then entered into the SWAT model along with soil parameters of the
observation sites and a pasture land cover. The observed amount of soil moisture in the

19

soil column was then compared to the same variable from the SWAT output on a daily
basis for the period of record studied.
The initial results were not encouraging. The simulated plant available soil moisture
agreed with field capacity and near saturation levels of plant available soil water, but
SWAT estimates tended to over-estimate the soil moisture during observed drier periods
by more than 15% of total plant available soil moisture when averaged over the entire
year and ~20% absolute difference.
Internal variables, such as ESCO (soil evaporation compensation coefficient) in the
model itself had to be adjusted in order to allow SWAT to draw water from deeper in the
water column to meet evaporative demand. The result of these changes was a very close
agreement between the simulated and observed values (Table 4 and Figures 18a, 18b,
18c) with differences of 5 mm of plant available soil moisture or less on average over the
study period, less than 3% of total plant available soil moisture in the soil column and ~
12% absolute difference, a great improvement over the initial model output. It was seen
that these results were similar to both the yearly variation in plant available soil moisture
and the magnitude of the change in soil moisture. This meant that SWAT was a logical
choice to continue to use.
But SWAT was found to have limitations that make its use difficult in some
circumstances. Rainfall that occurs over midnight is broken into the two days in which it
occurs; this causes a lower amount of precipitation to be recorded for both days. This
limitation also can be a part of the observed data set. Large sub-basins make simulating
run-off difficult due to varying soil types and saturated conductivity as well as differing

20

land covers and management styles of those land covers. Finally, soil composition and
other data aren’t often available at the spatial scales for large basins to be able to describe
infiltration of water over a large basin (Arnold et al, 1998).
Modifications to the SWAT base code, as shown in Baumgart (2005), were needed
due to the model not allowing crops to properly uptake Nitrogen and Phosphorous
fertilizer when there was nutrient stress, as well as the previously mentioned problems.
These elements were intentionally set as unlimited in the model in order to isolate the
effects of climate and land cover on soil moisture without the influence of soil fertility.
Allowing unlimited fertility allows the Corn and Wheat crop land covers to fully develop
and achieve yields on par with observed amounts with results of between 5.0 to 8.5 t/ha
for corn, and 3.7 – 6.9 t/ha for winter wheat (United States Department of Agriculture,
2009).
Another modification, from the same source, allowed accurate use of crop growth
models inside of SWAT when air temperature was below the base temperature of the
crop. This change allowed crops to grow if the temperature dropped below this minimum
value after planting, rather than having a crop failure in that year. There were many cases
seen where the model would suffer from this, more so in climatologically cool years.
Another change was needed to help the model function properly when dealing with soil
temperatures. Soil temperatures in the base model were under estimated when compared
to observed readings. This also caused the soil to stay frozen longer which also led into
the crop growth issues previously mentioned above. Finally, a code issue was also found
o

in which SWAT would replace observed temperatures of 0 C with a simulated
21

temperature using the internal WXGEN weather generation model. This replacement was
undesirable in that it could affect snow melt by increasing this value above or below the
freezing point.
In order to examine spatial patterns of the model output across the study region, a
type of spatial interpolation was necessary. In a previous study, DeLiberty and Legates
(2003) used an inverse-weighted method, as shown in Shepard (1968), for interpolating
data after concluding that it overcame several significant limitations of traditional
weighted-average approaches, accounted for the sphericity of the Earth, and is one of the
most reliable techniques for estimating daily precipitation. One very common objective
analysis technique is the spline method. In a study by Hartkamp et al (1999), splining was
described as “… a deterministic, locally stochastic interpolation technique that represents
two dimensional curves on three dimensional surfaces (for other studies see: Eckstein
1989; Hutchinson and Gessler 1994). Splining may be thought of as the mathematical
equivalent of fitting a long flexible ruler to a series of data points. Like its physical
counterpart, the mathematical spline function is constrained at defined points.”
The authors recommend splining because inverse distance weighting has no error
surface and kriging, an interpolative technique to determine the value of a variable at an
unknown point, requires a normal distribution. Price et al (2000) determined that the use
of ANUSPLIN, a set of FORTRAN programs that uses splines in analysis of monthly
mean precipitation, and monthly mean maximum and mean minimum temperatures
across multiple climate stations, in Canada and the Great Lakes region, was appropriate.
Price et al (2000) concluded that ‘thin-plate smoothing splines such as ANUSPLIN

22

enable better prediction in such regions because they calibrate a spatially varying
dependence on elevation that uses all available data points.’ For these reasons splining
was chosen as a way to display spatial patterns.
In order to determine the temporal trends of the simulated output from SWAT, a type
of regression analysis was needed to analyze the relationship of those variables over time.
There are quite a few different techniques to achieve this goal, many of them described in
detail in Lanzante (1996). One technique was used often in papers dealing with climate
change, described in Sen 1968 ( Hoaglin et al 1983, Gaffen and Ross 1999, Gahrs and
Lanzante 2000, Gaffen and Wang 2001, etc), and called the ‘median of pairwise slopes
regression’. This is computed by finding the slope between all possible pairs of points
and then taking the median of those values to determine the final slope. This technique
was used in order to determine temporal trends in the modeled soil moisture data
generated from SWAT.
Time Trend Analysis, Breakpoints.
A Standard Normal Homogeneity Test (SNHT) test was applied to the Precipitation
data for each station to determine a break point where precipitation trends may have
changed. This process was described in Khaliq and Ouarda (2007) as well as
Alexandersson and Moberg (1997). Differences in the slope of the rate of change of
precipitation could skew the results of any long term trend seen in the model output. For
example, if yearly precipitation amounts were fairly static from 1900-1950 and then
began to quickly rise. The rate of change for the entire time period would only show a
mild overall change.

23

The results of this comparison showed that while there were breakpoints at the 95%
and the 99% confidence level, they were not consistent across the study area, nor did
every station show break points. 18 climate stations showed a break point at the 95%
confidence interval and 12 showed the same at the 99% confidence interval. The average
year in which these breaks occurred was 1965 and 1967 with a standard deviation of 28.2
and 23.8 years respectively (Table 6).
As it is seen, only 62% of the climate stations showed a breakpoint at even the 95%
confidence interval. Even with looking at only these stations showed a large standard
deviation. Any breakpoint would have to be applied across all climate stations and this
would either not catch the breakpoint or place an artificial one where none existed. To
move forward in determining any long term trends across spatial and temporal
dimensions it was decided to not include a breakpoint.
Comparing results between soils or land covers
One of the first issues that comes up when attempting to view spatial data across
multiple output dimensions, in this case land cover and soil types, is a way to best display
it on a map for ease of understanding. In order to accomplish this goal, this study wished
to see if there was a larger difference between comparisons of the model output of land
covers and the differences in soil types.
Yearly results of the slope of variables generated by the model were combined under
the same land covers (i.e. all six soils that were in each land cover) and under soils (i.e.
all five land covers that were in each soil). A relative standard deviation test was
performed on both groups. The results of this test showed that combining all soils in a

24

specific land cover had a smaller relative standard deviation of the resultant slopes than
combining all land covers under a specific soil (Table 7). The results are striking in that
some variables show a whole order of magnitude difference in the same variable when
comparing between groups of soils and groups of land covers.
Furthermore, differences between soils within each land cover type were explored.
While all soils behave differently based on texture, it was seen that the rate at which a
soil moisture variable changes over time is quite similar between soils. It is also seen that
different textured soils behave in a very similar manner for both daily and yearly model
time-steps (see the final column of Table 14 for examples).
The three moderately textured soils (SL, L, and SiL) behave closely together while
the coarse and very fine textured soils (S, SiCL, and CL) tend to have a larger difference
in their change over time then the overall average. It is seen that displaying the results of
the average of the SL, L, SiL together would be a good way of geographically displaying
the temporal trends simulated. Their yearly soil attributes closely resemble each other and
generally fall in the center of the model results for these attributes.
For all of these yearly soil moisture variables we must keep in mind the soil water
t

balance equation: SWt = SWt-1 + ∑ t-1 (Ri – Qi – ETi – Pi – QRi).

25

RESULTS AND DISCUSSION
In order to properly understand the changes in soil moisture across the Great Lakes
region, one must first look at a baseline of climate variables and derived hydrologic
variables to better gauge the relative scale of the changes. Annual temperature and
precipitation totals for each of the study sites was averaged over a 1979 to 2008 time
frame. Further, precipitation and its trends drive each aspect of the hydrologic variables
of evapotranspiration, runoff and drainage.
Precipitation and Temperature
As was written earlier, Rodriguez-Iturbe and Porporato (2004) have stated that soil
moisture is “the center of the hydrologic cycle”. The driving factor behind changes in soil
moisture is primarily precipitation, and a lesser extent temperature through potential
evapotranspiration (Delworth and Manabe 1988).
Precipitation is a more complex variable spatially across the Great Lakes region. The
far eastern sections of the study region (New York, Pennsylvania, and eastern Ohio) have
the greatest amounts of average annual precipitation. Further west precipitation totals are
greater as you move from north to south. The differences can be attributed to both
differences in synoptic weather patterns and relative nearness of the Gulf of Mexico as a
source of moisture. (Andresen 2007)
The highest total average annual precipitation in this study is found at Warren, PA
(1195 mm) and the lowest is found at Pine River, MN (724 mm), with a mean of all
stations of 900 mm. (Figure 2)

26

Temperature is one of the primary driving variables behind the calculation of
evapotranspiration, and one expects to see a north-south gradient. Temperature across the
Great Lakes region in the 1979 to 2008 time period is fairly well behaved spatially. As is
expected, there is a strong warming trend to average annual temperatures as you move
further south, and a suggestion of a slight east to west warming trend that can be
attributed to moving deeper into the continent (See Figure 3).
The warmest average annual (1979-2008) temperature in the study area is Delphi, IN
with 11.2C, and the coolest average annual temperature can be found at Iron Wood, MI
with 4.3C, and a mean of all stations of 7.9C.
Annual Spatial-Temporal Trends in Simulated Soil Statistics
Land cover can have a significant impact on soil moisture. Evapotranspiration rate,
leaf interception, and infiltration rate can all be directly affected by the type of surface
cover (Rodriguez-Iturbe et al, 1999). Averaging the soil moisture variables across soil
textures for a given land cover shows the greatest amount of evapotranspiration occurring
on the pasture land cover (593.3 mm) and the lowest on the winter wheat land cover
(476.3 mm). Pastures long growing season and its peak growth in the warmest period of
the year allows for this high ET and the lower temperatures surrounding the winter wheat
help keep the ET value low for that land cover. The highest value for drainage is seen in
the winter wheat land cover (354.9 mm) and lowest with the urban land cover (80.6 mm).
The low water use of winter wheat in the warm and wet season helps increase the
drainage value, and the overall impermeability of the urban land cover makes it difficult
for surplus water to drain into the shallow aquifer. This same impermeability of the urban

27

land cover makes it have the highest average value of surface runoff (313.1 mm)
compared to the almost negligible amount of runoff seen for the forest (mixed northern
hardwood in this case) land cover (3.2 mm) where the relatively high water use of the
land cover keeps the upper soil layers dry and able to absorb precipitation and the trees
themselves acting to block water from easily running off. (Table 1) The duration and
timing of a crops growing season can change when the greatest water use can be
expected. Winter Wheat, for example, can be expected to draw water from the soil
column much earlier (and later) in the year than a typical Corn land cover. Another
example is the interception of a land cover in determining the amount of water entering
into the soil column during a precipitation event. An Urban land cover may behave in a
similar manner to a Forested land cover for light precipitation events as the interception
of leaves in the forest may be similar to the runoff generated by the impermeable surfaces
of the urban terrain, but behave much differently under extreme precipitation events.
As can be seen in Figure 4 (Foth, 1990), soil type and texture can have a large
influence on soil moisture in a given annual cycle. Soils that have a silt loam texture
generally have the largest water holding capability of all soil textures, with the more
extreme texture classes of sand and clay holding far less. The differences between sand
and clay, however, are that sandy soils tend to hold water at higher water potentials than
more moderately textured soils, and clay textured soils at lower water potentials. Given
higher overall hydraulic conductivities, this also allows sandy soils to replenish water
drawn out of the soil column by roots relatively quickly; while water removed from the
soil column by roots on clay textured soil is more slowly replenished. Averaging the soil

28

moisture variables across land cover types for a given soil texture shows that ET is highly
dependent on the water holding capacity of the soil itself. The highest values can be seen
with the Silty Clay Loam textured soil (570.1 mm) and the lowest on the Sandy textured
soil (471.6 mm). Drainage is inversely related to plant available water holding capacity as
larger pore sizes allow less water holding and more drainage with the highest values on
sandy soil (343.9 mm) and the lowest values on silty clay loam textured soil (222.5 mm).
Finally, when averaged across land covers, runoff is affected by the pore space of the soil
as well with the highest amount found on clay loam textured soils (113.6 mm) and the
smallest on Sandy Loam soils (82.8 mm) (Table 2).
When analyzing the results of the SWAT model, the average of Sandy Loam, Loam,
and Silty Loam soils are used as they are representative of all six soils. This holds true
throughout the discussion, unless otherwise noted.
Evapotranspiration
Mean annual total evapotranspiration averaged across soils and seen in all land cover
types is similar in pattern to average annual temperatures with a general increase from
north to south and a slight east to west increase as well. This trend is likely associated
with the trend in mean annual temperatures as well as northeast to southwest increases in
mean growing season solar radiation. Near the lakes, moderation of spring time
temperatures keeps local conditions cooler than locations of the same latitude that are far
from the lakes, allowing less ET to accumulate. And while the opposite is true in the fall,
when temperatures near the lakes are slightly higher compared to other sites, most plant
water use has trailed off as the plants goes into senescence. On a loamy texture soil with a

29

grassland cover, the highest average annual ET value for the 1979 to 2008 period was at
Delphi, IN (684 mm), the lowest was at Sault Ste Marie, MI (498 mm), and a mean
across all stations of 602 mm. (Figure 5)
Runoff
Runoff shows in interesting spatial pattern when looked at across the 1979 to 2008
time period. Many different factors impact this simulated variable including amount of
average annual precipitation (where higher values of precipitation occur, higher amounts
of runoff are expected), and severity of precipitation events (heavy precipitation events
would result in more runoff occurring then if the same amount of rainfall fell over a
number of days). When comparing a grassland land cover on a loamy textured soil the
eastern portion of the study area shows high run off (Watertown, NY having an average
of 25.9 mm per year), and as you move west the amount of runoff decreases until a
minimum value is reached at Wauseon, OH (5.7 mm). A significant ‘ridge’ of high runoff
values can be found stretching from north-central Illinois (Aurora, IL with 20.3 mm)
north to the western Upper Peninsula of Michigan (Ironwood, MI with 23.0 mm) (see
Figure 6). Much of the pattern of runoff can be explained through the amount of
precipitation that occurs in the fall season (September, October, and November and a
correlation coefficient of 0.55 when comparing fall precipitation and runoff) when there
is less evapotranspiration occurring, and the soil may be recharged.
Drainage
The average annual drainage, or the amount of soil water that flows downward and
out from the bottom of the soil column in a given year, is very similar spatially to the

30

pattern of average annual precipitation. In fact, when doing a simple correlation between
average annual precipitation and drainage across all stations, there is a 0.92 correlation.
When comparing a grassland land cover on a loamy textured soil, the highest values are
found in the eastern portion of the study area with a maximum average annual value of
553.8 mm at Warren, PA. There is a strong east to west downward pattern with a
minimum value of 163.8 mm at Pine River, MN. The mean value for the region is 286
mm. (Figure 7)
Time Averaged Inter-Annual Variables across time, 1979-2008.
A typical annual pattern for the seasonal drawdown and recharge of soil moisture at
South Haven, MI Data were plotted in a ‘box and whiskers’ graph format for modelsimulated values of daily soil moisture in the top 150cm of the soil profile during the year
(Figure 8 above). The solid blue line is a 9-day moving average of the median value of
th

th

soil moisture for a given day. The darker interior areas are the 25 to 75 percentiles of
th

th

all the daily values, and the lighter areas extend out to the 10 and 90 percentiles. The
beginning of soil moisture drawdown each year begins around day 50 (late February)
with the total moisture reaching a seasonal minimum about day 190 (mid July).
Replenishment/recharge begins at that point and continues until field capacity is reached
once again near day 300 (late October).
One additional variable that will be discussed is ‘Field Capacity Days’. This is the
total amount of water in the soil column over a year expressed as the sum of the total
amount of soil moisture for each day (mm * days). It is effectively the area of the soil
moisture graph below the soil moisture curve, and describes how ‘wet’ any given year is.
31

Annual variables such as evapotranspiration, runoff, and drainage help explain water
usage on a large temporal scale. But the inter-annual variables such as when the plant
available soil moisture falls below field capacity in the spring, when the minimum
amount of soil moisture occurs, when the date is that the soil is recharged to field
capacity, and the number of field capacity days (discussed above) each year are
extremely important when attempting to describe smaller temporal scale changes.
Day of Discharge
The day at which soil moisture in the top 150cm of the soil profile begins to drop
below field capacity in the spring is referred to in this study as the ‘Day of Discharge’.
The variable has a strong association with spring temperatures as the warmer the spring,
the faster snow will melt and the earlier ET can begin to accumulate. Thus there is a
strong south to north trend with the latest Day of Discharges occurring in the northern
sections of the study area. Averaged across all soil textures, the earliest Day of Discharge
th

can be found at Norwalk, OH (Day 37.4, or February 6 ), and the latest can be found in
th

Cheboygan, MI (Day 67.9, or March 9 ). The mean Day of Discharge is Day 53.6, or
rd

February 23 . (Figure 9)
Day of Minimum Soil Moisture
The timing of the minimum soil moisture value occurs each year is dependent on
many variables: annual ET, seasonal temperatures, and seasonal precipitation to name
just a few. The strongest correlation is between this variable and average summer
temperatures across all study sites, land covers and soil types. There is a north-east to

32

south-west decreasing (earlier) trend apparent in the 1979-2008 averages when looking at
this variable geographically. The earliest value in the study area, when looking at a
th

pasture land cover and averaging across soils is at Delphi, IN (Day 210.8, or July 30 )
th

and the latest value is at Pine River, MI (Day 241.1, or August 29 ) with a mean of Day
th

224.9 (August 13 ).
Figure 10a and b shows two different land cover types, corn and wheat. As can be
seen in the images the different land cover types affect the date of minimum soil moisture
differently. As corn is a late season grower, it uses the largest quantities of water later in
the summer whereas wheat uses it much earlier in the year. Corn has an average value of
th

th

Day 195.1 (July 14 ), a maximum of 223.4 (Aug 11 ) and minimum of 176.7 (June
th

th

26 ), while wheat has an average day of 166.9 (June 16 ) a maximum of 198.3 (July
th

st

18 ) and a minimum of 150.5 (May 31 ). This is nearly a month difference for the
average, maximum, and minimum dates between the two crops.
Minimum Amount of Soil Moisture
The minimum amount of soil moisture each year, like the day of occurrence, is
dependent on many variables including ET, seasonal temperatures, seasonal precipitation,
etc. There is a general north to south decrease in this variable, which makes sense as
shorter growing seasons in the north translate into less seasonal ET accumulation and
plant water use. The highest values of average annual minimum soil moisture were found
in northern sites, with 80.7mm at Ironwood, MI and 80.5 mm at Two Harbors, MN. The

33

lowest value was at Owosso, MI of 28.5 mm, and a geographical mean value of 43.1 mm.
(Figure 11)
Day of Recharge
The day of recharge is the day at which the soil column returns to field capacity, and
is dependent on seasonal precipitation, length of growing season, how dry the summer
was, etc. There is much less of a spatial trend apparent in the 1979-2008 average of this
variable. The earliest in the year that this occurs is found at Warren, PA on day 299.6
th

(October 27 ), and the latest that it occurs is found at Cheboygan, MI on day 354.7
th

th

(December 20 ) with a geographical mean of day 331.2 (November 27 ). (Figure 12)
Soil Moisture Days
As was described previously, Soil Moisture Days is a way of looking at how wet a
soil is over a given year. As it is the summation of daily soil moisture values, it is
impacted by each variable that also impacts soil moisture. When examining the 19792008 averages spatially, there is a pattern that can be described. The lowest values (driest
overall soils) can be found in the central part of the study region, Michigan, Indiana,
Illinois, and eastern Wisconsin. The lowest value in the region can be found at Oconto,
WI with 273.3 SMD. The highest values can be seen in the eastern and western areas of
the study region with the highest value found at Ironwood, MI with 319.9 SMD. The
spatial average of the study region is 290.4 SMD. (Figure 13)
Differences between Land Cover and Soil Types
Varying soil type can have a large influence on soil moisture in a given annual cycle,
and thus, over long periods of time. Soils that have a silt loam texture generally have the
34

largest water holding capability of all soil textures, with the more extreme texture classes
of sand and clay holding far less. The differences between sand and clay, however, are
that sandy soils tend to hold water at higher water potentials than more moderately
textured soils, and clay textured soils at lower water potentials. This also allows sandy
soils to replace water drawn out of the soil column by roots to be rapidly replaced, and
water removed from the soil column by roots on a clay textured soil would only slowly
be replaced (Foth, 1990).
Temporal Changes in Temperature and Precipitation
Spatial-Temporal Trends in Temperature
Across the 1900-2008 study period, temporal trends in average annual temperature
across the region are spatially heterogeneous (Figure 14) ranging from -0.007 C/year to
+0.017 C/year with an mean of 0.002 C/year. Overall, a larger number of observation
sites showed an increase in temperature (14) compared to stations showing a decrease in
temperature (12). Three stations showed no trend at all, and none of these were
significant. This is not completely surprising, and agrees with previous research that has
identified a number of shorter term trends within the 109-year study period, including a
th

period of steady temperatures from the beginning of the 20 century to the 1930’s, a
slow drop in temperatures from the 1930’s to the early 1980’s, and finally a rapid
warming from the 1980’s onwards (Andresen 2007).
Spatial-Temporal Trends in Precipitation
Precipitation is the dominant influence on water for the soil moisture variables in the
SWAT model. Overall, with only a few exceptions, annual precipitation was found to
35

increase across the Great Lakes region during the period studied. (Table 26, Figure 19).
The average rate of change for the entire study area was 0.733 mm/year. The majority, 24
of the 29 climate stations, had a positive slope with an average rate of change of 0.938
mm/year. 5 of the 29 stations had a negative slope with an average rate of change of 0.253 mm/year. 3 of the 5 that had negative precipitation slopes were found in
Wisconsin. This shows that the vast majority of the study area became significantly
wetter through the period of study. For example, the slope of 1.682 mm/year at Ann
Arbor, MI, suggests an increase of 181.7 mm in annual precipitation over the study
period.
The pattern of precipitation change over the study period when compared to the
average annual precipitation amounts for the 1981-2010 is worth noting. The area with
the greatest average annual precipitation in the study area is north-western Pennsylvania
was associated with near level or slightly decreasing annual precipitation values trends.
In contrast, show moderate to strong increases in annual precipitation were found in the
driest regions in the study area, (the central Upper Peninsula, eastern sections of the
Lower Peninsula of Michigan, east-central Minnesota, and north-western Wisconsin).
Evapotranspiration
Evapotranspiration (ET) is a combination of soil evaporation and transpiration from
plants on a land surface, and as such, is dependent on both precipitation and temperature.
As precipitation increases there is more moisture entering into the soil column for
transpiration by plants. Increasing temperature increases the rate of evaporation from the
surface of the soil.

36

The rate of ET is very different dependent on land cover type. In general, the most
positive rates of change for ET were found on Urban land covers across all soil types,
with a minimum value of -0.130 mm/year, a maximum of 0.811 mm/year and a mean of
0.248 mm/year, and the most negative rates of change were found on a forested land
cover when looking across all soils, with a minimum value of -0.451 mm/year, a
maximum of 0.579 mm/year and a mean of 0.064 mm/year (Table 18 and Figures 20 to
24).
Drainage
Drainage is defined as water which passes downward from the rooting zone It
depends on both rainfall and antecedent soil moisture content, and as such is a stochastic,
state-dependent component whose magnitude and temporal occurrence are controlled by
the entire soil moisture dynamics (Rodriguez-Iturbe and Porporato, 2004) and is a good
representation of how saturated the soils are through the annual cycle.
Temporal changes of drainage are similar to both the precipitation trends and the ET
trends (Table 19 and Figures 25 to 29). Drainage increased across much of the study area,
similar to precipitation. Decreases were observed in Central Wisconsin and eastern Iowa.
The largest positive and negative trends across all soils were observed over the forest
land cover (maximum of 1.82 mm/year, minimum of -0.41 mm/year, and an average of
0.69 mm/year).
Water Return
REVAP, as it is labeled in the SWAT model, is water which returns upward to the
soil column from the shallow aquifer. This variable is a complex combination of the

37

amount of precipitation occurring, when it occurs, and the depth of the rooting zone of
the land cover. Rooting depth is an issue here as 2 of the land covers, corn and forest,
have a rooting zone equal to or deeper than the simulated soil column.
Interestingly enough, each of the before mentioned land covers still closely resemble
the other three land covers in spatial pattern of the rate of change and the scale of the rate
of change. A mix of increasing and decreasing trends of water return was found across
the region. Decreasing areas are the Upper Peninsula of Michigan and northern
Wisconsin. Areas of increase are, depending on the land cover, the western and southern
areas of the study region (Table 20 and Figures 30 to 34).
Across all soils the highest positive rates of change in water return are unsurprisingly
seen in a forested land cover (maximum of 0.033 mm/year, minimum of -0.010 mm/year,
and an average of 0.006 mm/year), as the trees draw water from the deepest portion of the
soil column allowing moisture. The highest negative rates of change were found on a
pasture surface (maximum of 0.03 mm/year, minimum of -0.018 mm/year, and an
average of 0.005 mm/year).
Runoff
Runoff is the amount of water that leaves or enters the system by moving laterally
across the surface of the land cover rather than infiltrating into the soil column. This is
due to the impermeability of the surface, high antecedent soil moisture which prevents
infiltration, or a combination of both. In general, yearly runoff trends closely resembled
the yearly precipitation trends across the region. Almost no change in annual runoff was
observed across the region over the period, while the urban land cover had the expected

38

highest rate of change of annual runoff due to impermeable surfaces that followed the
rate of change for precipitation.
Spatially, runoff increased across the entire study area with the highest amounts in
northern Illinois and along the Lake Erie coast. Decreases were observed in central
Wisconsin and eastern Iowa. (Table 21 and Figures 35 to 38). Both the maximum
positive and negative trends in runoff were associated with the urban land cover across all
soils (maximum of 1.045 mm/year, minimum of -0.325 mm/year, and an average of
0.351 mm/year).
Temporal Changes in Hydrologic Variables and Derived Variables.
Inter-annual Daily Trends – Summary
The change of soil moisture levels within a given year is complex. In order to detect
any long term changes then each year a derived statistic such as the day at which
discharge occurs must be examined and compared to previous years. Slowly a pattern
begins to emerge that can describe the behavior of these statistics.
In modeling soil moisture, it is important to know how much soil moisture variation
is caused by small temporal scale events and long temporal scale events in order to
understand how accurate a model will be in reproducing expected soil moisture results
and what scale the spatial and temporal resolution should be for the simulation (Entin et
al 2000). Following the research of Entin et al (2000) and Deliberty and Legates (2003),
the temporal scale was calculated using the procedures set forth in their studies. In the
Entin et al study, they found a temporal scale of between 1.3 to 2.1 months for the upper
1m of soil in Iowa and Illinois. The Deliberty and Legates study found, for Oklahoma, a

39

1.8 month temporal scale. In this study the results correspond with the previous studies
results fairly closely. Between all of the soils the average temporal scale across all the
sites ranged from 1.35 months (41 days) for a corn land cover to 1.93 months (61 days)
for an urban land cover, with an average temporal scale across land covers and soils of
1.65 months (50 days). See Table 24 for more detailed results.
Additional seasonal soil moisture variables examined in the study for changes over
time were: the day of year when the soil first begins to fall from field capacity, the rate of
discharge of soil moisture from the beginning of use until the minimum value of the year,
the day of year when the minimum soil water value is reached, the rate of recharge of soil
moisture from the day of that minimum value back to field capacity, and the day of year
when the soils field capacity is reached. Finally there is the summed number of days at or
above field capacity. Examples from selected stations can be found on Tables 22 and 23
in the Appendix for all of these variables.
Day of Discharge and Discharge Rate Trends and Patterns
This study examined the day of year when the soil water began to decrease from field
capacity. If the soil moisture fell below field capacity and then returned to field capacity
before falling again, the day of year was then taken as the latter value. After these dates
were obtained, a pair-wise step regression was performed as described above to calculate
trends and rates of change for the study period.
The spatial patterns for the rate of change for the date at which this start of discharge
occurs were found to be dependent on the land cover used. Positive rates of change,
meaning that the date at which discharge begins is increasingly later in the year, were

40

observed in southern Wisconsin across land cover types. This same pattern can be seen in
the Saginaw Bay area of Michigan except for forest and urban land covers. Negative
trends were observed across the rest of the study, with discharge beginning earlier in the
year. The most negative trends are in south-east Michigan and the Pine River Dam area
of Minnesota (Figures 39 to 43).
The observed changes are noteworthy. For example, at Ann Arbor, MI, the rate of
change of -0.0991 days / year for a corn land cover translates into a 10.7 day shift earlier
in the year during the study period. With the same land cover in Viroqua, WI the shift
was 1.1 days later in the year. Across soil types, the largest positive trend is 0.104
days/year found on a pasture land cover at Viroqua, WI, and the largest negative trend is
also on the pasture land cover with a trend of -0.269 days/year found at Pine River, MN.
The rate of discharge is obtained as the total amount of soil moisture lost each year
between the day of year where soil moisture first begins to fall in the soil column and
through the day of year where the minimum amount of soil moisture is reached. While
each soil texture class has differing field capacities and minimum plant available water
values, the magnitude of the rate of change for this calculated variable is very similar.
The trends for this variable are quite small, on the order of µm/(day*year).
Spatially the patterns are dissimilar, as can be expected, over the differing land
covers. The greatest trend change can be seen in a wheat land cover as it begins its water
use earlier in the warm season, thus drawing the soil moisture down earlier. The smallest
trend changes over the study period can be found with the urban land cover. The rate of
change is the highest in southern Wisconsin, the central Upper Peninsula and Saginaw

41

Bay area of Michigan (Figures 44 to 48). Across all soils, the largest positive change in
the rate of discharge is 2.77 µm/(day*year) found at Pine River, MN on forested land
cover, and the largest negative trend is -1.55 µm/(day*year) found at Fond du Lac, WI on
the Wheat land cover.
When temperatures begin to increase seasonally in the late winter / early spring,
snowmelt and relatively low ET ensure that the soil column is saturated. Overall, the start
of soil moisture discharge across most of the region has become earlier by 1-2 weeks or
more during the study period. As to the physical causality behind these changes, one
major suspect is temperature trends across the observation locations (Table 25). As can
be seen in Figures 52-56, there is a strong correlation between higher rates of change for
average yearly temperature and the date of the beginning of discharge. Higher
temperatures cause ET to begin earlier in the year; it can also cause snow melt to occur
earlier, allowing the spring recharge to end earlier. The same pattern can be seen when
looking at the rate of discharge against both the maximum and minimum temperature
rates of change, which are strongly correlated. None of the other variables were
correlated with changes in temperatures at any level of significance except for the Day of
Year where soil moisture reaches field capacity against maximum temperatures.
Comparison of these variables against the rate of change of yearly precipitation
(Table 26) results in correlations that is much higher than temperature as a whole. This
stands to reason as precipitation is the major driving force behind changes in soil
moisture. But it is interesting to see that there is a higher correlation with minimum
temperature when compared to the rate of change of the rate of discharge, and the rate of

42

change of the day of minimum soil moisture when compared to any of the temperature
slopes. These are both variables that occur during the first half of the year. This could be
caused by lower plant ET in the early part of the year making temperature the driving
force rather than the gain or loss of moisture from precipitation.
Day of Minimum Soil Moisture Trends and Patterns
The timing of the minimum amount of soil moisture each season is an important to
know for water management and irrigation scheduling. It is also the date at which the soil
begins to recharge due to decreased ET or increasing precipitation.
Changes in the timing of the date of minimum soil moisture were found to differ
widely across the various land covers, sometimes significantly. This is due to the either
differences in the plants growing season and the changes in precipitation during those
times (i.e. early in the spring for winter wheat, or later in the summer for corn, or the long
growing season of the pasture and forest land covers). Trends towards earlier dates were
observed across the land covers (towards over the Upper Peninsula of Michigan, and
sections of northern Wisconsin and Minnesota (Figures 49 to 53). There were also
differences between land covers such as a trend towards later dates on a pasture land
cover in NE Michigan but negative trends on all other land covers and the Illinois stations
showing near zero trends except for a winter wheat land cover. Across all soils, the
largest positive rate of change is 0.11 days/year found on the urban land cover at
Farmington, MN, and the largest negative rate of change is -0.17 days/year found at Big
Rapids, MI on the pasture land cover. The differences in this variable appear to be
associated with the timing of peak water usage during the growing season. Corn tends to

43

use water earlier in the year due to rapid growth and canopy development when ET is
increasing from increasing temperatures and still high values of plant available water.
This is why the large portion of the study area has negative trends. The positive trends
(getting later in the year) are associated with the same areas that are having a reduction of
ET through steady temperatures, increased cloudiness and precipitation, etc.
For the forest land cover the date has generally come earlier in the year except for the
stations near the Lake Erie and the northern Lake Michigan area. The earlier dates are
especially associated with areas that are also experiencing some of the largest increases in
ET. The more south-western areas that show this can also be explained by a general
decreasing trend in precipitation during the autumn season or a lower rate of change then
the surrounding areas (seasonal precipitation rates of change can be seen in Figures 69 to
72).
As pasture has a long growing season, many different variables impact the timing of
its day of minimum soil moisture. The strongest trend towards an earlier day of year can
be found in the western Lower Peninsula of Michigan, and the strongest trend towards a
later day of year can be found in the eastern Lower Peninsula. A linear regression (with
the rate of change of minimum soil moisture on a pasture land cover as the dependent
variable) was performed with many rates of change of variables discussed in this paper in
an attempt to describe the spatial behavior of this land cover and rate of change such as
rate of change of ET, rate of change of runoff, rate of change in seasonal precipitation,
2

etc. A model was developed that didn’t explain a significant portion of the variability (r
of 0.56), but explained a high amount of that variability (p-value of 4e-4) using only
44

yearly precipitation, ET, percolation and water being drawn back into the soil column.
Each of these rates of change had at least a 95% confidence rating with ET and
percolation above a 99% confidence rating with no co-linearity found between these
variables. This suggests that as the rates of change of precipitation and water re-entering
the soil column increase, and ET and percolation decrease that the date of minimum soil
moisture gets later in the year.
Looking at the spatial distribution of this variable on an urban land cover there were
also patterns that couldn’t easily be understood. This spatial distribution required another
linear regression (with date of minimum soil moisture with an urban land cover as the
dependent variable) to fully understand. A model was developed that didn’t explain a
2

significant portion of the variability (r of 0.34), but explained a high amount of that
variability (p-value of 4.4e-3) using only the inverse rate of change for ET (99.8%
confidence), and the rate of change for percolation (96.7% confidence) with no colinearity found between these variables. This means that as average yearly ET decreases
and percolation increases, the day of the year where the minimum soil moisture trends
later in the year. This stands to reason as a higher ET would mean less water in the
system throughout the year and more percolation means there is more often a greater
amount of water then the soil column can hold.
The wheat land cover for this variable has very little spatial continuity across the
study region. This complex variable surface also required a linear regression model to
attempt to better explain the behavior. The linear regression showed that the rate of
change of summer time precipitation had the greatest effect on the rate of change of the
45

date of minimum soil water (96.4% confidence), but still explained very little of the
2

actual variability (r of 0.15). As summer precipitation increases, the date of minimum
soil water gets later in the year. This makes sense when you consider that wheat has a
growing season outside of the summer season.
Minimum Amount of Soil Moisture Trends and Patterns
As soil moisture decreases during the growing season, it can stress the plant life
above it. But if the level of soil moisture at its minimum was getting higher or lower, the
impact could be significant. The rate of change for the minimum amount of soil moisture
was highly dependent on both soil type and land cover. Within a land cover category, the
rate of change is consistently in the same direction, but the scale can be quite different.
Because of the diversity of the results, we will confine our discussion to corn, forest, and
wheat, which are representative of the range of the results (Figures 54 to 56 respectively).
In general, there was a widespread increase in the minimum yearly value of soil moisture
across the study area. The general exceptions are the Wisconsin observation sites and
Upper Peninsula of Michigan, the southern observation sites, and in New York.
These are fairly striking results. For Ann Arbor, MI, the rate of change for corn on a
loamy soil is 0.1301 mm/year and a beginning soil moisture value of 68.1 mm. This
starting value increases to 82.2 mm at the end of the study period, of, an increase in the
minimum plant available soil moisture of over 20%.
It is important to note that each of the combinations of land cover and soil types had a
different outcome for the rate of change of the yearly minimum soil moisture. For corn
there is a very distinct geographical distribution and steep slopes. As discussed earlier,
46

using the rates of change of minimum soil moisture amount as the dependent variable in a
linear regression against multiple independent variables was used to better explain the
geographical distribution. We can see that this variable has a direct relationship with ET
rates of change (>99.9% confidence rating) and annual precipitation. It does stand to
reason that higher precipitation contributes to an overall higher ET. Combined with
increasing temperatures this geographical distribution makes sense. The combination of
temperature and precipitation trends is arguably the most important aspect of changes in
minimum soil moisture trends in the study region. As precipitation increases, so does the
amount of soil moisture. While the increasing temperatures help lengthen the growing
season and increase ET, the lowering amount of water stress put on plants due to the
increase in soil moisture is important to see.
The statistics for trends on forest were different. A longer growing season and a
deeper root system made the water being drawn back into the soil column from the
shallow aquifer more important to the rate of change (99.1% confidence). Combined with
precipitation, to a lesser extent due to the higher interception by the leaf canopy, the rate
of change for this land cover and soil is much more modest than corn on loam.
The third land cover looked at in this paper was wheat. Wheat is a more complex land
cover in that the ET decreases significantly during the summer following maturity of the
crop. The largest contributor to the spatial distribution of the rate of change was the water
reentering the soil from the shallow aquifer (>99.9% confidence rating). To a lesser
extent both the spring season and summer season precipitation contributes to the
distribution with the summer precipitation inversely related. This inverse relationship is

47

explained by understanding that higher amounts of rainfall in the summer contribute to a
larger ET. Keeping ET low allows the shallow aquifer to better recharge the lower
portions of the soil column, and having a great chance to increase the minimum soil water
value.
Averaged across all soils the maximum positive rate of change for this variable is
0.207 mm/year found on the forest land cover at Fayette, IA, the maximum negative
trend is -0.176 mm/year found on the corn land cover at Viroqua, WI. The average
overall trend across all land covers and soils is 0.046 mm/year.
Day of Recharge and Rate of Recharge Trends and Patterns
As the growing season comes to a close the level of soil moisture begins to increase.
The amount and frequency of precipitation also adds to the increase. The rate of change
of seasonal recharge, the increase of the soil moisture from its minimum value back to
field capacity was examined was found to vary across the region. Negative rates of
change (recharge happening less quickly with time) were observed in the southern areas
of the study region with the wheat land cover. The highest positive rates of change
(recharge happening more quickly) are found in northern Wisconsin, the Upper Peninsula
of Michigan, and northern Minnesota for the corn and winter wheat land covers. One
commonality across all land covers was a positive rate of change in many areas of
Michigan (Figures 57 to 61).
The day of recharge describes when the soil has finally recharged to its field capacity.
For this region, this typically occurs during the early fall to the early winter. A relatively
later recharge date during the next year’s spring was uncommon. Out of all the climate

48

observation stations, combinations of land covers and soils, and over 108 years of
simulation, in only a tiny fraction of the seasons (<0.1%) was the soil not fully recharged
by the beginning of the next start of discharge.
Regionally, positive trends toward a later day of recharge across land covers (with the
highest rate of change for the pasture land cover) were found over large portions of
Wisconsin. The largest changes in a negative direction (full recharge occurring earlier in
the year) were found on the pasture land cover in the far northwest of the study area.
Urban and wheat land covers had the lowest rates of change in both a positive and
negative direction.
In Watertown, NY, for example, the rate of change is -0.1610 days per year with a
th

Forest land cover with a starting intercept of day 313 (November 9 ). By the end of the
rd

study period this has changed to day 296 (October 23 ), a relative change of over 2
weeks earlier in the year. In contrast, Viroqua, WI, has a positive rate of change on a
th

pasture land cover of +0.1184 with an intercept of day 321 (November 17 ). By the end
th

of the study period this has changed to day 333 (November 29 ). The greatest positive
rate of change in the study area for the rate of recharge is found on the pasture land cover
at Pine River, MN (4.32 µm/(day*year)), the greatest negative rate of change for the rate
of recharge is -7.85 µm/(day*year) found on the winter wheat land cover at Delphi, IN
and an average across the study region of +0.41 µm/(day*year). As for the day of
recharge, the greatest positive rate of change in the study region was found at Fond du

49

Lac, WI with a rate of +0.137 days/year on a corn land cover, the greatest negative rate of
charge was at Pine River, MN on a pasture land cover with a rate of -0.403 days/year.
Other commonalities across land covers include a trend towards earlier recharge across
much of Michigan and a lower rate of change on the Lake Erie coastline in Ohio as
compared to the surrounding areas (Figures 62 to 66).
There is a strong geographical distribution within each land cover, though each one
shows a different distribution and scales of the rates of change (Figures 70- 79). For the
corn land cover, the rate of recharge during the study period is decreasing. In contrast, the
rate of recharge is increasing across Central Michigan and the Upper Peninsula of
Michigan along with eastern New York. The relationships between the other rates of
change are complex. Percolation and runoff are significantly related to this rate of
change, at a 99.98% and 99% confidence rating and a negative relationship with runoff.
The change in minimum temperatures also plays a part with a negative relationship of
97.2% confidence rating.
As stated previously, the forest land cover has a long growing season with extended
roots able to draw moisture from deep in the soil column. The geographical distributions
of the rates of change for this land covers recharge rate are somewhat similar to corn.
Positive trends were found across Central Michigan, the Upper Peninsula, and
southeastern sections of the region. A large portion of this spatial variability comes from
annual precipitation, over a 99% confidence rating. This is from both amount of yearly
precipitation change and the amount per event of precipitation change.

50

For pasture, there was a large decrease in the rate of recharge across the western
portion of the study area that generally follows the decrease of precipitation rate. Still, for
areas where there was an increase in precipitation such as south-west Michigan, the
eastern Upper Peninsula, and the Lake Erie coastline there was still a slight decrease or
only a weak increase in the rate of recharge. This is perhaps related to increasing
temperatures throughout the year, greater ET rates and increasing plant water usage
through the recharge period.
Changes in the rate of recharge for the urban land cover were minor compared to the
other land covers. This is mainly due to the impermeability of the surface type making
increases in precipitation increase runoff rather than ground recharge. The strongest
upwards trends can be seen in northern Wisconsin and the western Upper Peninsula. The
strongest downwards trends occur in the southern areas of the study region. For the wheat
land cover, there was a pronounced slowing of the recharge rate across almost the entire
study region. The only exceptions are the far eastern areas of the region and north-west
Michigan.
Soil Moisture Days Trends and Patterns.
One of the more interesting aspects of this study is looking for trends in soil moisture
days, which is essentially a summation of the time and magnitude of soil moisture at or
above the field capacity threshold.
The rates of change for this variable closely resemble the rates of change for
precipitation across the study area (Figure 19). Negative rates of change are seen in
Wisconsin, suggesting an overall drying of the soil column, while positive rates of

51

change were observed almost everywhere else. The greatest positive rate of change in
across the study region was at Pine River, MN with a pasture land cover (+0.406
SMD/year), while the greatest negative rate of change was found at Viroqua, WI with an
urban land cover (-0.148 SMD/year). The average trend change across all land covers and
soil types was +0.116 SMD/year. Spatial trends for urban and wheat land covers, which
are representative of other land covers are given in Figures 67 and 68.
These changes for the urban surface is somewhat more distinct due to the greater rates
of runoff for that cover type which increases the overall sensitivity of soil moisture
changes associated with changes in precipitation.
For the wheat land cover the rates of change are more moderate due to increasing ET
during the summer. The spatial patterns are similar to the precipitation rates of change
with lower amounts in Wisconsin, but still close to 0, with the highest rates over
Michigan and the western portions of the study area.
Summary of Yearly Spatial-Temporal Trends in Soil Statistics
In general, the results from this study are similar to those of Karl and Knight (1998),
who found large scale increases in precipitation along with an increase in extreme
precipitation events in the Great Lakes region. One exception was the state of Wisconsin,
were precipitation was observed to decrease. Figure 83 shows a spatial distribution of
locations with precipitation trends 95% or greater confidence. In general, it suggests a
significant increase in precipitation over the study period. Ranges for these increases are
from 0.7 mm/year to 1.7 mm/year.

52

SUMMARY AND CONCLUSION
Consequences for water usage in the Great Lakes region
The main conclusions that can be drawn from this study can be summarized in the
following way:
1. For the thirty-year time frame (1979-2008), precipitation is greatest in the eastern
portion (Pennsylvania and New York) of the study area and tends to decrease as
you move north-west, with the lowest amounts in Minnesota, the central Upper
Peninsula of Michigan, and the north-east portion of the Lower Peninsula of
Michigan. Precipitation is increasing across the majority of the Great Lakes
region over the time period studied and this has a direct impact on plant available
soil moisture across all land covers and soils. The largest trends in precipitation
can be found in the state of Michigan with high rates of change in eastern
Minnesota, north-eastern Iowa, and south-central Wisconsin. Seasonally, autumn
shows the widest range of rates of change with western most areas showing
decreasing precipitation and increasing everywhere else in the study region. The
summer months show the largest spatial area of increasing precipitation with a
reducing amount in the Upper Peninsula of Michigan.
2. For the thirty-year time frame (1979-2008), evapotranspiration has a strong south
(highest values) to north (lowest values) spatial trend with higher values moving
further northward in the western portion of the study area. Evapotranspiration
trends across soils and land covers are generally increasing across the time period
studied mainly due to increasing temperatures and precipitation across the Great

53

Lakes region. The greatest positive trends can be seen on pasture land in
Michigan, the greatest negative trends can be seen in forested areas in western
Wisconsin and the Upper Peninsula of Michigan, and the smallest rates of change
can be seen in the winter wheat land cover. These trends can be attributed to
growing season timing and length.
3. For the thirty-year time frame (1979-2008), percolation was strongly tied to the
pattern seen for precipitation. Areas of higher precipitation showed rates of higher
percolation and areas of lower precipitation showed lower amounts of percolation.
Percolation had similar trends no matter the land cover across the study region
with positive trends showing an increasing amount of water leaving the soil
column. The pattern is closely tied to the rates of change for precipitation.
4. For the thirty-year time frame (1979-2008), runoff was spatially varied greatly
with the highest values seen for the pasture land cover on a loamy soil in a northsouth line stretching from northern Illinois to the Upper Peninsula of Michigan as
well as a gradual increase in an easterly direction. The lowest values for the 30year period were seen in eastern Michigan and Minnesota. Surface runoff is both
closely tied to land cover and precipitation trends. Forest land cover shows no
change in runoff and urban sees the largest changes. This rate of change follows
closely with precipitation changes as well as the land covers ability to prevent
water movement on the surface. This finding could be seen as a predictor of the
ability of urban water drainage systems ability to handle storm run-off. It also
directly affects soil erosion which can impact bluff and beach erosion (tourism),

54

loss of top soil (agriculture), and the introduction of fertilizer and other man-made
chemicals into the Great Lakes waterways (fisheries).
5. The year to year rates of change of the daily variables shows significant trends
almost everywhere. Each of these traits directly impacts the areas of agriculture,
forestry and urban water management.
a. For the 1979-2008 time frame there is a strong south to north trend with
the latest Day of Discharges occurring in the northern sections of the study
area. The trends for the entire time frame show the day of year when the
soil begins to discharge is generally getting earlier in the year from earlier
snowmelt from increased temperatures. This occurs across soil and land
cover types. The date at which the soil recharges is also moving earlier in
the year over much of the study area. Notable exceptions are where the
trend for precipitation amounts is negative. When looking at the pasture
and urban land covers due to the near constant water use during the warm
seasons, one can see this pattern very clearly.
b. The rate of discharge is based on two variables, both of which are
changing. While the beginning of discharge is occurring earlier in the
year, the day of minimum plant available soil moisture is also getting
earlier across much of the study area. This leads to an increase in the rate
of discharge. Areas that have a slower rate of discharge are correlated with
areas with a positive precipitation rate of change. The rate of recharge also
varied significantly across the study region. Land covers that have a

55

growing season that extends into the period of where temperatures begin
to cool and PET begins to decrease are slowing down the rate of recharge.
These patterns can be seen in the corn, pasture, and wheat land covers
where a long growing season that is stretching later in the year is and is
reducing the rate of recharge.
c. In the 1979-2008 time period, the day of minimum soil moisture shows a
similar spatial pattern across all land covers, with a north-east to southwest decreasing (earlier) trend. The day of year where the soil moisture is
lowest is also changing across the region as it is affected by both the land
cover and the amount and timing of precipitation. It is getting earlier in the
year over much of the study area with notable differences between land
covers, with the strongest movement towards the beginning of the year
often found in Michigan (with a movement of almost 20 days earlier in
one case) and the strongest movement towards the end of the year in the
south western, western, and Lake Erie areas.
d. The average of the 1979-2008 time period shows a drier soil conditions in
the northern areas of the study region and wetter areas in the south for the
corn and wheat land cover, while the forest and urban land covers show
spatial homogeneity. The minimum amount of soil moisture is increasing
in all areas except where there is a negative trend in precipitation. This is
an important find as it directly affects the amount of water stress

56

agricultural products will experience without irrigation or the amount of
irrigation needed to prevent water stress in the same crop cover.
e. The 1979-2008 time frame shows much less of a spatial trend across the
study region for the date at which field capacity was reached in the fall
with only a slight trend towards later recharge dates occurring in the
northern areas than in the southern areas. The trends for the 1900-2008
show positive trends toward a later day of recharge across land covers
(with the highest rate of change for the pasture land cover) were found
over large portions of Wisconsin.
f. When examining the 1979-2008 averages of Field Capacity Days, or the
integration of the total amount of daily soil moisture in the soil column,
shows that the lowest values (driest overall soils) can be found in the
central part of the study region, Michigan, Indiana, Illinois, and eastern
Wisconsin. Field Capacity Days also show an upwards trend across the
study region, only with the notable exceptions of Wisconsin and parts of
the western Upper Peninsula of Michigan, but then only slight negative
trends. This finding shows that over the course of an entire year during the
time frame of this study that soil is getting more wet overall.
All of these soil moisture trends show a different Great Lakes region in the present
then even 100 years ago. It shows a Great Lakes region that is slowly getting wetter and
warmer. It shows a region where agricultural producers and urban planners have needed
to change their policies and ideas about water entering their fields or drainage systems.

57

Finally it shows that these trends will continue into the future as climate change continues
to affect the Great Lakes region.

58

APPENDICES

59

Corn
Forest Pasture Urban Wheat
ET
499.9
570.6
593.3
506.4
476.3
Drainage
311.2
326.4
292.7
80.6
354.9
Runoff
88.9
3.2
14.1
313.1
68.6
Table 1: Soil Moisture variables (mm) averaged across study region and across soil types, 19792008.
S
SL
L
SiL
SiCL
CL
ET
471.6
528.9
540.7
526.6
570.1
537.9
Drainage
343.9
288.6
268.3
266.8
222.5
248.8
Runoff
83.2
82.8
91.5
106.8
107.5
113.6
Table 2: Soil Moisture variables (mm) averaged across study region and across land cover types,
1979-2008.

Grid / Observed
Annual Precipitation
Totals
Alpena, MI
1.09
Ann Arbor, MI
1.08
Chicago, IL
1.01
Geneva, NY
1.05
Grayling, MI
0.94
Indiana, PA
0.95
Rogers City, MI
1.02
Sheboygan, WI
1.04
Traverse City, MI
1.02
Wilkes Barre, PA
1.16

Grid / Observed
Annual Wet Days

Precipitation
amount Ratio

1.57
1.57
1.87
1.79
1.31
1.73
2.08
2.06
1.76
1.82

Station Name

0.69
0.69
0.54
0.59
0.72
0.55
0.49
0.50
0.58
0.64

Table 3: Comparison of Precipitation Data from observed precipitation and the associated grid
box.
Observation Station SWAT – Observed PAW Absolute Value of SWAT – Observed
Bondville, IL
0.65
22.98
Champaign, IL
3.27
26.53
Topeka, IL
5.06
27.24
Table 4: Differences in Simulated and Observed Plant Available Soil Moisture in mm of
Plant Available Water.
60

Soil Name
Clay % Silt % Sand %
Sand
5
5
90
Sandy Loam
10
30
60
Loam
20
40
40
Silty Loam
10
60
30
Silty Clay Loam
30
60
10
Clay Loam
35
35
30
Table 5: Composition of Soil Textures used for plant available water comparisons.

Yearly
Timeframe
Full Year
Dec, Jan, Feb
Mar, Apr,
May
Jun, Jul, Aug
Sep, Oct, Nov

Average Year Average Year STDEV Year STDEV Year
95%
99%
95%
99%
confidence
confidence
confidence
confidence
1965
1961

1967
1973

28.25
39.85

23.82
39.57

1957

1934

31.76

N/A

# of
# of
stations
stations
99%
95%
confidence confidence
18
12
10
6
3

1958
N/A
42.33
N/A
6
1965
1969
28.18
6.18
9
Table 6: Precipitation Breakpoints as determined from a SNHT.

1
0
4

Land Cover / Soil
ET
PERC
REVAP
SURQ
Grouping
Sand
174.84
110.45
251.56
120.11
Sandy Loam
274.83
68.32
172.68
119.87
Loam
246.00
99.16
154.37
135.85
Silty Loam
197.47
102.09
192.89
233.92
Silty Clay Loam
148.487
96.23
234.87
283.45
Clay Loam
278.88
101.75
384.43
199.21
Corn
38.08
5.42
6.26
2.20
Forest
27.17
6.59
26.74
5.79
Pasture
92.90
8.60
7.43
5.03
Urban
100.12
70.44
102.01
3.69
Wheat
10.33
4.02
5.24
2.61
Table 7: Relative Standard Deviations of soil moisture variables. Soil or Land Cover groupings.
ET=Evapotranspiration, PERC=Percolation, REVAP=Return Water from Shallow Aquifer,
SURQ=Surface Runoff.

61

Station
Corn
Forest Pasture Urban Wheat
Ann Arbor, MI
505.0
589.9
601.4
523.4
479.7
Angelica, NY
493.2
543.0
601.8
505.1
463.0
Aurora, IL
520.3
581.7
630.8
528.1
489.4
Big Rapids, MI
487.2
557.9
586.9
499.8
464.5
Cheboygan, MI
452.9
508.8
531.1
450.4
434.6
Clinton, IA
529.2
608.2
644.8
525.1
498.2
Coldwater, MI
518.7
610.9
609.4
533.4
492.5
Delphi, IN
543.4
637.7
672.9
544.1
505.5
Dixon, IL
526.1
614.9
637.9
532.5
497.2
Fond du Lac, WI
490.5
587.2
579.0
486.9
476.7
Farmington, MN
512.9
590.5
604.6
502.0
492.1
Fayette, IA
538.2
597.7
635.6
541.2
506.8
Iron Mountain, MI
431.8
520.7
569.5
471.3
454.9
Ironwood, MI
471.4
529.3
533.7
478.4
453.1
Norwalk, OH
521.4
600.3
624.2
536.3
489.9
Oberlin, OH
516.1
604.6
614.3
532.3
481.6
Oconto, WI
479.0
538.7
584.0
482.4
460.9
Owosso, MI
481.5
523.9
562.1
493.0
465.1
Pine River, MN
483.7
543.6
544.4
474.2
472.6
Portage, WI
521.6
580.0
618.9
525.4
491.6
South Haven, MI
481.4
545.3
566.0
497.6
462.5
Spooner, WI
501.4
553.9
598.6
493.2
479.3
Sault Ste Marie, MI 442.8
498.0
491.6
436.4
415.4
Two Harbors, MN
483.7
557.3
542.9
466.8
469.5
Viroqua, WI
510.1
588.0
600.3
507.9
484.7
Wooster, OH
536.6
621.0
632.1
553.9
497.1
Warren, PA
529.4
586.6
622.4
548.9
493.1
Watertown, NU
478.3
531.5
557.5
495.3
457.3
Wauseon, OH
509.7
597.6
608.3
519.1
483.5
Table 8: Average Annual Simulated Evapotranspiration (mm), 1979-2008, by land cover.

62

Station
S
SL
L
SiL
SiCL
CL
Ann Arbor, MI
477.6
536.8
550.7
535.6
588.5
550.1
Angelica, NY
472.7
520.5
529.7
520.4
554.5
529.6
Aurora, IL
489.1
549.5
561.6
547.0
593.6
559.5
Big Rapids, MI
457.2
516.0
530.1
515.6
567.7
529.0
Cheboygan, MI
416.9
473.3
486.7
471.4
520.5
484.5
Clinton, IA
500.9
562.4
574.6
557.6
602.5
568.6
Coldwater, MI
491.8
551.7
564.4
550.0
597.1
562.9
Delphi, IN
517.0
581.7
593.6
578.0
624.1
589.9
Dixon, IL
501.5
562.1
574.3
559.6
602.5
570.3
Fond du Lac, WI
463.7
523.9
536.8
520.5
566.9
532.6
Farmington, MN
485.5
543.6
553.2
538.7
574.8
546.9
Fayette, IA
511.0
567.1
576.1
563.8
594.7
570.8
Iron Mountain, MI
436.1
488.3
500.9
486.6
529.0
496.9
Ironwood, MI
440.1
492.3
502.9
491.8
529.9
502.1
Norwalk, OH
494.6
553.3
565.3
551.7
597.5
564.0
Oberlin, OH
488.1
547.8
561.1
546.2
595.6
559.7
Oconto, WI
448.5
508.2
521.3
506.2
551.5
518.4
Owosso, MI
444.8
502.3
516.1
501.1
551.8
514.4
Pine River, MN
452.8
506.5
515.2
501.3
537.1
509.4
Portage, WI
490.6
548.2
560.1
545.8
584.6
555.6
South Haven, MI
446.7
506.2
521.2
504.4
565.2
519.7
Spooner, WI
472.5
527.9
537.9
523.6
558.4
531.4
Sault Ste Marie, MI 400.9
452.5
465.6
452.3
503.7
465.9
Two Harbors, MN
450.3
506.3
515.7
503.4
536.6
512.0
Viroqua, WI
483.8
540.0
550.6
536.9
572.3
545.5
Wooster, OH
508.2
568.1
579.6
566.6
608.0
578.2
Warren, PA
506.6
557.0
565.2
555.6
588.3
563.8
Watertown, NU
446.0
500.8
513.7
500.4
549.6
513.3
Wauseon, OH
482.2
542.5
555.5
540.6
587.6
553.3
Table 9: Average Annual Simulated Evapotranspiration (mm), 1979-2008, by soil texture.

63

Station
Corn
Forest Pasture Urban Wheat
Ann Arbor, MI
369.5
353.9
335.1
116.1
412.2
Angelica, NY
390.2
430.7
359.0
127.8
442.8
Aurora, IL
324.2
371.5
306.5
76.0
376.5
Big Rapids, MI
339.3
358.8
317.8
95.7
381.2
Cheboygan, MI
253.7
265.6
233.0
69.9
288.6
Clinton, IA
254.9
260.7
214.4
47.8
304.7
Coldwater, MI
348.0
343.2
333.0
91.6
395.0
Delphi, IN
338.6
341.4
295.6
82.9
398.0
Dixon, IL
310.3
334.5
295.8
60.8
364.6
Fond du Lac, WI
208.2
180.2
180.1
35.2
237.9
Farmington, MN
203.6
201.9
180.3
28.3
240.0
Fayette, IA
295.7
350.7
300.2
52.7
351.7
Iron Mountain, MI 246.3
227.8
173.1
39.4
242.8
Ironwood, MI
311.8
358.0
333.0
82.0
355.7
Norwalk, OH
359.6
360.6
329.4
99.9
410.9
Oberlin, OH
363.9
352.5
333.6
105.5
416.3
Oconto, WI
260.1
286.3
229.5
56.0
297.4
Owosso, MI
280.7
298.0
254.3
69.5
311.4
Pine River, MN
174.4
179.8
169.8
20.4
203.4
Portage, WI
281.6
326.2
270.0
52.9
330.4
South Haven, MI
354.3
370.1
338.0
107.2
392.1
Spooner, WI
206.2
226.1
177.0
28.2
245.0
Sault Ste Marie, MI 330.4
345.9
340.6
123.3
376.4
Two Harbors, MN
230.2
225.3
230.3
48.5
263.2
Viroqua, WI
258.0
275.8
252.1
44.6
303.5
Wooster, OH
390.3
393.0
374.6
110.1
451.6
Warren, PA
553.2
608.3
557.5
211.7
619.4
Watertown, NU
481.8
549.4
501.2
182.8
531.3
Wauseon, OH
305.1
289.8
272.9
71.8
348.7
Table 10: Average Annual Simulated Drainage (mm), 1979-2008, by land cover.

64

Station
S
SL
L
SiL
SiCL
CL
Ann Arbor, MI
394.1
336.1
313.9
312.9
255.8
291.2
Angelica, NY
414.7
367.0
347.7
342.6
305.8
322.9
Aurora, IL
366.9
310.2
286.2
282.5
235.1
264.9
Big Rapids, MI
373.4
315.2
294.1
293.2
243.4
272.1
Cheboygan, MI
289.3
232.8
215.8
219.5
175.4
200.1
Clinton, IA
288.0
230.0
209.0
209.9
167.0
195.1
Coldwater, MI
378.7
320.7
297.7
295.8
244.8
275.2
Delphi, IN
371.4
309.9
286.1
282.4
233.7
264.2
Dixon, IL
348.1
290.4
266.8
264.1
222.3
247.6
Fond du Lac, WI
235.3
178.4
160.3
163.5
122.0
150.3
Farmington, MN
233.6
179.5
162.9
163.6
131.8
153.6
Fayette, IA
338.8
285.6
264.2
259.1
226.3
247.2
Iron Mountain, MI
245.4
195.4
178.9
182.4
145.8
167.3
Ironwood, MI
354.9
301.4
285.3
282.5
245.0
259.4
Norwalk, OH
386.5
329.9
308.2
305.2
256.1
286.5
Oberlin, OH
391.0
332.8
309.9
309.3
255.6
287.7
Oconto, WI
295.0
239.2
219.1
220.6
178.8
202.5
Owosso, MI
314.2
257.6
237.7
239.0
187.6
220.5
Pine River, MN
206.5
155.8
142.3
145.2
113.8
133.7
Portage, WI
322.0
267.7
245.9
243.0
205.7
228.8
South Haven, MI
393.4
337.2
311.1
308.9
239.4
284.0
Spooner, WI
236.4
186.0
168.1
169.3
139.6
159.5
Sault Ste Marie, MI 374.2
320.8
303.1
301.9
244.2
275.6
Two Harbors, MN
263.2
210.3
192.7
191.6
159.7
179.5
Viroqua, WI
293.6
241.1
220.2
218.5
182.0
205.4
Wooster, OH
421.0
363.4
340.4
334.9
287.8
315.9
Warren, PA
584.8
531.6
511.7
499.8
454.8
477.2
Watertown, NU
527.3
468.4
448.7
444.9
391.7
414.6
Wauseon, OH
332.2
275.1
252.1
251.2
202.0
233.4
Table 11: Average Annual Simulated Drainage (mm), 1979-2008, by soil type.

65

Station
Corn
Forest
Pasture
Urban
Wheat
Ann Arbor, MI
70.5
1.2
8.2
305.2
53.1
Angelica, NY
92.6
1.9
14.7
342.2
70.1
Aurora, IL
115.1
6.0
22.0
355.6
93.6
Big Rapids, MI
94.6
4.4
16.4
326.1
75.3
Cheboygan, MI
68.6
1.7
11.4
255.1
51.9
Clinton, IA
86.7
2.5
11.0
299.2
67.9
Coldwater, MI
90.7
3.2
14.7
332.2
69.9
Delphi, IN
99.4
2.2
12.9
354.9
77.9
Dixon, IL
120.3
4.6
20.9
363.2
94.5
Fond du Lac, WI
66.5
1.0
8.2
245.0
50.4
Farmington, MN
75.8
1.9
8.5
262.3
59.3
Fayette, IA
121.5
5.7
19.5
362.4
96.6
Iron Mountain, MI
71.5
2.2
10.0
240.7
51.4
Ironwood, MI
111.0
6.1
26.8
332.3
85.2
Norwalk, OH
81.6
1.6
8.9
326.0
61.8
Oberlin, OH
78.3
1.1
9.0
318.8
60.3
Oconto, WI
86.7
2.6
13.1
289.0
66.9
Owosso, MI
61.9
1.0
7.4
261.2
47.7
Pine River, MN
65.8
1.6
9.0
227.4
47.7
Portage, WI
107.6
6.5
21.8
334.5
88.6
South Haven, MI
86.0
6.1
17.6
317.1
67.1
Spooner, WI
74.1
1.9
8.4
260.7
56.9
Sault Ste Marie, MI
75.7
4.0
15.9
290.2
57.0
Two Harbors, MN
70.1
1.2
10.6
267.9
51.2
Viroqua, WI
101.1
6.9
18.4
317.0
80.8
Wooster, OH
88.8
1.7
9.1
351.8
67.0
Warren, PA
114.5
2.9
17.6
436.2
84.6
Watertown, NU
127.6
7.2
29.4
409.1
99.1
Wauseon, OH
73.8
0.5
6.9
297.4
56.4
Table 12: Average Annual Simulated Runoff (mm), 1979-2008, by land cover.

66

Station
S
SL
L
SiL
SiCL
CL
Ann Arbor, MI
72.2
72.3
80.6
96.7
100.2
103.9
Angelica, NY
86.9
88.3
98.8
113.6
114.5
123.7
Aurora, IL
102.7
100.2
112.2
130.1
130.5
135.1
Big Rapids, MI
89.7
90.3
97.4
112.7
110.1
120.1
Cheboygan, MI
68.2
69.5
73.2
84.6
79.9
91.0
Clinton, IA
81.1
79.0
87.9
103.8
101.5
107.5
Coldwater, MI
85.8
85.3
95.5
111.8
114.9
119.5
Delphi, IN
91.7
90.0
102.0
121.4
123.8
127.7
Dixon, IL
104.9
102.9
114.4
132.1
131.5
138.2
Fond du Lac, WI
65.7
64.5
70.0
82.7
78.4
84.2
Farmington, MN
71.8
69.6
77.0
90.4
88.0
92.6
Fayette, IA
104.2
102.9
115.4
132.7
134.2
137.5
Iron Mountain, MI
66.8
67.0
71.4
81.5
77.4
86.8
Ironwood, MI
97.7
100.0
105.5
119.4
118.9
132.2
Norwalk, OH
80.0
79.4
89.2
105.9
109.0
112.3
Oberlin, OH
77.9
77.5
86.8
102.7
105.7
110.4
Oconto, WI
81.0
79.4
86.5
99.9
97.4
105.9
Owosso, MI
64.3
64.4
70.3
84.1
83.0
88.9
Pine River, MN
63.4
61.2
66.2
77.1
73.2
80.7
Portage, WI
96.8
95.6
106.1
122.7
121.9
127.5
South Haven, MI
81.0
78.9
90.1
108.5
116.3
117.8
Spooner, WI
71.6
68.4
76.4
89.2
85.3
91.3
Sault Ste Marie, MI
72.2
74.7
80.7
94.9
101.0
107.6
Two Harbors, MN
69.3
67.2
75.5
88.9
87.7
92.5
Viroqua, WI
90.6
88.8
99.5
114.6
116.2
119.4
Wooster, OH
85.0
84.3
96.0
114.7
120.1
122.1
Warren, PA
104.4
108.9
120.7
142.2
154.2
156.5
Watertown, NU
113.3
119.0
125.9
142.8
146.0
159.8
Wauseon, OH
73.2
71.1
81.0
96.9
98.0
101.7
Table 13: Average Annual Simulated Runoff (mm), 1979-2008, by soil texture.

67

Watertown,
NY

Viroqua, WI

Two Harbors,
MN

Clinton, IA

Ann Arbor,
MI

Climate
Soil
All
Station Attribute Soils

S

SL

ET

0.34 0.30

1

1

PERC

3

3

L

8.96

3.18

1.12

1.19

3.48

2.82

0.01

0.01

0.01

36.89

27.54

1

0.22

0.16

28.57

17.76

ET

-0.06 -0.05 -0.06 -0.07 -0.07 -0.01 -0.08 -0.07

44.71

2.72

PERC

-0.06 -0.09 -0.05 -0.06 -0.06 -0.06 -0.06 -0.06

23.43

7.95

233.48

196.16

7.57

7.45

REVAP 0.01 0.00
SURQ

0.18 0.12

REVAP 0.00 0.00
SURQ
ET

3

0.32

3

0.34

1
3

1.22

1.21

1.16

0.00

0.01

0.01

0.14

0.00

0.16

0.00

0.39
1.15

0.19 0.25

0.00

2

CL

3

0.32

1

SiCL

1

1.18 1.21

1

SiL

3

0.00

1

Relative
SL, L, Relative
STDEV
and STDEV
SL, L,
SiL All Soils
SiL

0.36

3

0.00

0.33

0.00

-0.12 -0.10 -0.12 -0.11 -0.12 -0.12 -0.12 -0.12
1

1

1

2

2

1

9.25

1.15

3

7.22

6.53

1

13.40

4.38

2

20.09

18.39

-0.19 -0.22 -0.24 -0.19 -0.20 -0.10 -0.17 -0.21

25.68

12.36

0.01 -0.01

0.04

0.02

0.03 -0.02 0.02

0.03

169.57

38.95

REVAP 0.00 -0.01

0.00

0.00

0.00

0.00

279.64

28.93

-0.02 -0.01 -0.03 -0.01 -0.02 -0.01 -0.02 -0.02

53.78

44.95

0.02 0.01

0.26

0.21 0.26

1

3

1

1

2

PERC

3

1

0.92 1.03

REVAP 0.02 0.02
SURQ 0.24 0.18
ET
PERC
SURQ
ET
PERC

3

0.91 0.93

3

REVAP 0.00 0.00

3

0.98

1

0.02

2

0.18

0.26

3

0.92

1

0.02

2

0.23

0.27
0.86
0.02
0.26

3
1
2

0.26
0.89
0.02
0.29

3
2
2

0.00

0.27

3

0.87

1

0.02

2

0.28

0.00

0.26
0.92
0.02
0.22

0.02

0.01

0.02

0.05

0.01

0.02

89.37

20.65

3

3

3

3

3

0.88

3

0.94

5.48

5.44

0.00

0.00

53.44

30.00

0.99

0.93

0.89

0.00

0.00

0.00

0.85

0.00
1

SURQ 0.15 0.12 0.11 0.15 0.15 0.20 0.17 0.14
22.80
16.26
Table 14: Rate of change (mm/year) of yearly soil attributes, selected climate stations,
1
2
3
comparison of soil textures. 0.95 confidence, 0.99 confidence, 0.999 confidence

68

Climate Station Change in Precipitation (mm / year)
3

Ann Arbor, MI

1.68
-0.34

Clinton, IA

3

Two Harbors, MN

1.50
-0.20

Viroqua, WI

2

Watertown, NY
1.22
Table 15: Change in annual Precipitation trends (mm/year) for selected climate stations:
1
2
3
1900-2008. 0.95 confidence, 0.99 confidence, 0.999 confidence.
Land
Drainage
Runoff
ET
ET Slope Drainage
Runoff
Cover
Slope
Slope
Corn
512.90
0.31
361.53
1.24
70.90
0.14
Forest
586.53
0.25
358.00
1.27
0.77
0.00
Pasture
598.84
0.50
338.78
1.22
7.70
0.01
Urban
528.04
0.38
133.32
0.81
283.51
0.56
Wheat
478.78
0.19
413.32
1.42
53.22
0.12
Corn
537.80
-0.01
245.71
-0.07
87.64
-0.17
Forest
605.35
-0.17
263.77
0.08
2.19
0.00
Pasture
643.53
-0.09
216.55
-0.26
10.85
-0.01
Urban
540.55
0.05
49.37
-0.01
282.57
-0.27
Wheat
497.06
-0.11
306.07
-0.02
67.97
-0.13
Corn
492.43
0.24
221.12
0.97
70.60
0.25
Forest
558.32
0.27
224.65
1.14
0.91
0.00
Pasture
543.61
0.21
230.19
1.26
10.17
0.04
Urban
478.58
0.40
51.75
0.22
252.90
0.65
Wheat
469.38
0.20
263.29
1.00
51.47
0.18
Corn
519.21
-0.10
249.32
-0.08
100.88
0.00
Forest
587.40
-0.38
277.74
0.23
5.74
0.00
Pasture
599.30
-0.40
254.59
0.15
17.00
0.00
Urban
522.49
-0.05
46.04
-0.04
301.31
-0.09
Wheat
484.03
-0.12
305.42
-0.09
79.77
-0.01
Corn
485.54
0.05
473.33
0.96
129.11
0.14
Forest
530.37
0.05
553.04
1.07
5.25
0.00
Pasture
554.04
-0.10
506.45
1.10
27.97
0.05
Urban
498.06
0.06
205.53
0.61
384.29
0.38
Wheat
456.69
0.02
531.84
0.94
99.56
0.11
Table 16: 30-year average (1979-2008) annual hydrology variables (mm) and associated rate of
change (mm/year) by land cover for selected stations.
Watertown,
NY

Viroqua,
WI

Two
Harbor, MN

Clinton, IA

Ann Arbor,
MI

Station
Name

69

DOY Start
DOY Min DoY Min Min SW Min SW
of
SoD Slope
SW
SW Slope
(mm)
Slope
Discharge
Corn
43.3
-0.10
188.2
-0.11
72.0
0.09
Forest
43.9
-0.18
210.0
0.00
35.8
0.03
Pasture
45.3
-0.10
218.6
-0.11
33.7
0.03
Urban
44.6
-0.18
242.5
0.01
45.9
0.04
Wheat
45.4
-0.06
180.3
0.02
129.4
0.17
Corn
45.9
-0.05
142.3
-0.04
80.1
-0.04
Forest
54.5
-0.03
155.4
0.00
41.1
-0.02
Pasture
50.6
-0.08
170.1
0.04
40.4
0.02
Urban
59.7
-0.04
182.5
-0.01
39.5
-0.01
Wheat
46.1
-0.08
139.7
-0.04
124.2
0.05
Corn
52.9
-0.10
181.8
0.00
54.4
0.04
Forest
59.9
-0.05
184.9
-0.04
43.9
0.03
Pasture
52.3
-0.14
184.4
-0.04
80.4
0.20
Urban
64.4
-0.12
182.6
-0.02
62.8
0.04
Wheat
50.5
-0.03
132.6
0.00
88.7
0.16
Corn
54.9
0.01
157.1
0.00
72.3
-0.18
Forest
65.9
0.07
167.8
-0.08
47.8
0.00
Pasture
61.9
0.08
179.1
-0.01
44.9
0.00
Urban
74.6
0.10
190.4
0.04
58.0
0.03
Wheat
55.2
0.02
132.9
0.01
121.3
0.03
Corn
48.0
-0.04
159.2
0.00
63.7
-0.06
Forest
46.3
-0.07
166.4
0.00
37.9
-0.02
Pasture
46.9
-0.07
177.5
-0.02
35.6
-0.01
Urban
46.2
-0.06
179.7
-0.07
57.5
-0.01
Wheat
45.6
-0.08
134.6
0.01
117.3
-0.02
Table 17a: 30-year average (1979-2008) inter-annual hydrology variables and associated rate of
change by land cover for selected stations.
Land
Cover

Watertown,
NY

Viroqua,
WI

Two
Harbor, MN

Clinton, IA

Ann Arbor,
MI

Station
Name

70

Field
Field
End of
EoD Slope Capacity Capacity
Recharge
Days
Days Slope
Corn
305.7
-0.16
318.8
0.18
Forest
430.6
-0.19
279.6
0.27
Pasture
458.7
-0.18
280.7
0.17
Urban
412.1
-0.08
265.3
0.33
Wheat
490.2
-0.12
340.1
0.13
Corn
309.0
0.03
319.3
0.01
Forest
334.8
0.11
281.4
-0.03
Pasture
343.1
0.08
277.6
-0.03
Urban
396.0
0.00
248.3
-0.02
Wheat
294.7
0.00
334.3
0.00
Corn
322.7
-0.09
319.7
0.14
Forest
340.0
-0.15
295.4
0.19
Pasture
324.9
-0.08
317.5
0.33
Urban
383.0
-0.01
273.5
0.36
Wheat
288.4
-0.10
329.7
0.17
Corn
295.9
0.01
323.9
-0.03
Forest
314.9
0.01
296.5
0.03
Pasture
328.9
0.12
294.5
-0.05
Urban
394.3
0.00
254.6
-0.15
Wheat
280.1
0.00
337.3
0.01
Corn
298.3
-0.05
317.8
0.02
Forest
298.0
-0.16
297.0
0.11
Pasture
309.0
-0.15
292.1
0.12
Urban
341.8
-0.11
285.6
0.11
Wheat
280.2
0.01
337.7
0.02
Table 17b: 30-year average (1979-2008) inter-annual hydrology variables and associated rate of
change by land cover for selected stations.
Land
Cover

Watertown,
NY

Viroqua,
WI

Two
Harbor, MN

Clinton, IA

Ann Arbor,
MI

Station
Name

71

Two Harbor, MN

Clinton, IA

Ann Arbor, MI

Climate
Station

Land Cover

ET trends average ET trends for average
of all soils
of SL, L, and SiL
(mm/year)
(mm/year)
1

Corn
Forest

0.30
0.25

Pasture

0.47

Urban

0.48
0.19
0.01
-0.15
-0.08
0.05
-0.10

1
2

Wheat
Corn
Forest
Pasture
Urban
Wheat

2

Corn
Forest
Pasture

0.22
0.24
0.21

Urban

0.41

3
1

Wheat

1

0.31
0.25
0.50

1
1

0.38
0.19
-0.01
-0.17
-0.09
0.05
-0.11

2

0.24
0.27
0.21
0.40

3
1

Watertown,
NY

Viroqua, WI

0.20
0.20
Corn
-0.07
-0.10
Forest
-0.37
-0.38
Pasture
-0.35
-0.40
Urban
-0.05
-0.05
Wheat
-0.10
-0.12
Corn
0.04
0.05
Forest
0.05
0.05
Pasture
-0.08
-0.10
Urban
0.07
0.06
Wheat
0.02
0.02
Table 18: Slopes of annual evapotranspiration trends, different land covers of selected stations
1
2
3
(mm/year). 0.95 confidence, 0.99 confidence, 0.999 confidence

72

Percolation trends Percolation trends for
for all soils
SL, L, and SiL
(mm/year)
(mm/year)
3

1.27

Forest

1.33

Pasture

1.25

Urban

0.60

Wheat
Corn
Forest
Pasture
Urban
Wheat

1.43
-0.08
0.07
-0.25
-0.01
-0.03

Corn

0.99

Forest

1.15

Pasture

1.25

Urban

0.21

Wheat
Corn
Forest
Pasture
Urban
Wheat

1.01
-0.10
0.20
0.12
-0.04
-0.11

Corn

0.93

Forest

1.08

Pasture

1.08

Urban

0.54

Watertown, NY

Clinton, IA

Ann Arbor, MI

Corn

Two Harbor, MN

Land Cover

Viroqua, WI

Climate
Station

3
3
3
3

3
3
3
3
3

3
2
3
3
3

1.24
1.27
1.22
0.81

3
3
3
3
3

1.42
-0.07
0.08
-0.26
-0.01
-0.02
0.97
1.14
1.26
0.22

3
3
3
3
3

1.00
-0.08
0.23
0.15
-0.04
-0.09
0.96
1.07
1.10
0.61

3
2
3
3
3

Wheat
0.92
0.94
Table 19: Slopes of annual percolation trends, different land covers of selected stations
1
2
3
(mm/year). 0.95 confidence, 0.99 confidence, 0.999 confidence

73

Two Harbor, MN

Clinton, IA

Ann Arbor, MI

Climate
Station

Land Cover
Corn
Forest
Pasture

Water return
Water return trends
trends for all soils for SL, L, and SiL
(mm/year)
(mm/year)
0.002
0.003
0.001
-0.003
-0.002
-0.004
2

Urban
Wheat
Corn
Forest
Pasture
Urban
Wheat

0.008
0.021
-0.002
-0.007
-0.000
0.002
0.004

Corn

0.016

Forest

0.025

Pasture

0.023

1
1
1
2

Urban

1

0.009
0.021
-0.004
-0.008
0.002
0.003
0.004

1

0.016

1

0.023

1

0.023

1

Watertown,
NY

Viroqua, WI

0.010
0.011
Wheat
0.021
0.020
Corn
0.001
0.002
Forest
0.006
0.008
Pasture
0.005
0.006
Urban
0.001
0.001
Wheat
-0.007
-0.006
Corn
-0.001
-0.000
Forest
-0.004
-0.005
Pasture
-0.002
-0.002
Urban
-0.001
-0.002
Wheat
-0.005
-0.005
Table 20: Slopes of annual water return trends, different land covers of selected stations.
1
2
3
(mm/year). 0.95 confidence, 0.99 confidence, 0.999 confidence

74

Watertown, NY

Viroqua, WI

Two Harbor, MN

Clinton, IA

Ann Arbor, MI

Climate
Station

Land Cover
Corn
Forest
Pasture
Urban

All soils
(mm/year)
0.14
0.00
0.01
1

0.63
0.12
-0.16
0.00
-0.01
-0.28
-0.12

Wheat
Corn
Forest
Pasture
Urban
Wheat

2

Corn

SL, L, and SiL
(mm/year)
0.14
0.00
0.01
1

0.56
0.12
-0.17
0.00
-0.01
-0.27
-0.13

1

Forest

0.25
0.00

Pasture

0.04

Urban

0.70

Wheat

0.19
0.01

0.18
0.00

1

0.00

1
2
1

Corn
Forest

0.25
0.00
0.04
0.65

1
2
1

0.00
0.00
0.10
0.01
0.15
0.00
0.05

Urban

0.00
-0.09
-0.01
0.14
0.00
0.05

1

Pasture
Urban
Wheat
Corn
Forest
Pasture

1

0.42
0.38
Wheat
0.12
0.11
Table 21: Slopes of Annual Runoff trends, different land covers of selected stations (mm/year).
1
2
3
0.95 confidence, 0.99 confidence, 0.999 confidence

75

Field
Capacity
Days

Day of
1
Recharge

Recharge
Rate
µm/day

Minimum
Soil Water
Amount
mm

Day of
Minimum
Soil Water

Discharge
Rate
µm/day

Day of
Discharge

Land
Cover

Climate
Station
Watertown,
Two
Ann Arbor,
Viroqua, WI
Clinton, IA
NY
Harbors, MN
MI

Corn
52.83
-0.95
194.65
52.96
1.07
325.42
298.60
Forest
60.59
-1.06
201.33
26.33
1.09
360.29
247.09
Pasture
56.08
-0.97
224.65
25.74
1.14
356.03
258.54
Urban
64.04
-0.86
246.47
32.91
0.98
409.77
225.93
Wheat
60.59
-1.06
201.33
26.33
1.09
360.29
326.58
Corn
48.69
-0.79
182.36
80.77
1.22
288.90
321.05
Forest
54.33
-1.00
192.19
30.92
1.24
314.92
286.73
Pasture
56.58
-1.04
208.46
27.05
1.32
330.31
283.20
Urban
66.61
-0.83
241.40
35.10
0.99
404.67
249.20
Wheat
50.90
-0.54
168.34
121.23
1.03
268.84
337.25
Corn
68.56
-0.83
222.67
50.12
1.63
321.34
310.39
Forest
71.10
-0.95
234.45
31.38
1.25
353.21
279.34
Pasture
72.54
-0.770
238.43
67.59
1.08
328.09
292.45
Urban
84.68
-0.84
240.80
56.51
0.83
412.57
237.64
Wheat
62.79
-1.17
161.79
77.17
1.07
285.32
316.11
Corn
53.96
-0.74
194.33
86.20
1.58
294.12
326.97
Forest
54.63
-0.90
209.16
32.97
1.58
313.50
292.76
Pasture
54.61
-0.90
225.85
32.25
1.78
321.08
292.87
Urban
71.58
-0.81
241.40
47.01
0.87
412.67
261.88
Wheat
53.75
-0.73
147.56
121.99
0.84
271.83
338.34
Corn
51.26
-0.86
196.00
58.80
1.40
297.68
314.76
Forest
52.11
-0.98
206.16
30.79
1.55
313.08
286.56
Pasture
52.35
-0.92
226.31
29.89
1.73
321.85
277.85
Urban
52.37
-0.72
243.79
47.19
1.28
347.29
272.32
Wheat
53.20
-0.65
153.93
123.98
0.71
267.01
338.19
Table 22: Intercepts of regression trends of Inter-annual soil moisture variables, different land covers of selected stations. Average of
1
SL, L, SiL soil type. When this value is above 365, it means that the recharge date is in the next year.
76

-0.10

0.50

-0.11

Forest

2

0.94

Pasture
Urban

-0.18
-0.10

1

-0.18

Wheat

-0.06

Corn
Forest
Pasture
Urban
Wheat
Corn
Forest

-0.05
-0.03
-0.08
-0.04
-0.08
-0.10
-0.05

Pasture

1

0.09

2.10

-0.16

0.18

0.00

0.03

3.54

-0.19

0.27

0.19

-0.11

0.03

2.85

-0.18

0.17

1.02

0.01

0.04

0.94

-0.08

0.33

-0.03

-0.12

0.13

-0.45
-0.21
0.37
0.13
-6.47
0.72
-1.42

0.03
0.11
0.08
0.00
0.00
-0.09
-0.15

0.01
-0.03
-0.03
-0.02
0.00
0.14
0.19

1.42

-0.08

0.33

-0.01
-0.10
0.01
0.01

1

0.36

1

0.02

2

1.95
-0.25
-0.32
0.77
-0.12
0.45
0.59
0.41

-0.04
0.00
0.04
-0.01
-0.04
0.00
-0.04

0.17
-0.04
-0.02
0.02
-0.01
0.05
0.04
0.03

-0.14

1.28

-0.04

0.20

Urban

-0.12

0.86

-0.02

0.04

0.18

Wheat
Corn
Forest

-0.03
0.01
0.07

2.28
-1.26
-1.86

0.00
0.00
-0.08

0.16
-0.18
0.00

0.39
-0.58
-1.24

Pasture

1

1

Field
Capacity
Days
days/year

Date of
Discharge
days/year

Discharge
Rate
µm/(days*
year)
Date of
Minimum
Soil Water
days/year
Minimum
Soil Water
Amount
(mm/year)
Recharge
Rate
µm/(days*y
ear)
Date of
Recharge
days/year

Land Cover

Climate
Station
Ann Arbor, MI
Clinton, IA
Two Harbors,
MN
Viroqua, WI

Corn

1

0.17
-0.03
0.03

Watertown,
NY

-0.46
-0.01
0.00
-4.64
0.12
-0.05
0.08
Urban
0.10
-0.59
0.04
0.03
0.20
0.00
-0.15
Wheat
0.02
0.38
0.01
0.03
-2.04
0.00
0.01
Corn
-0.04
-0.25
0.00
-0.06
2.04
-0.05
0.02
Forest
-0.07
0.32
0.00
-0.02
2.93
-0.16
0.11
Pasture
-0.07
0.50
-0.02
-0.01
2.73
-0.15
0.12
Urban
-0.06
0.06
-0.07
-0.01
1.34
-0.11
0.11
Wheat
-0.08
0.47
0.01
-0.02
1.54
0.01
0.02
Table 23: Slopes of trends of Inter-annual soil moisture variables, different land covers of
1
2
3
selected stations. Average of SL, L, SiL soil type. 0.95 confidence, 0.99 confidence, 0.999
confidence

77

Average

Clay
Loam

Silty Clay
Loam

Silty
Loam

Loam

Sandy
Loam

Sand

Land
Cover

Climate
Station
Watertown,
Two
Ann Arbor,
Viroqua, WI
Clinton, IA
NY
Harbors, MN
MI

Corn
1.28
1.13
1.11
1.18
1.31
1.15
1.19
Forest
1.23
1.38
1.38
1.36
1.72
1.41
1.41
Pasture
1.26
1.28
1.37
1.31
1.88
1.41
1.42
Urban
1.22
1.56
1.90
1.93
2.51
2.21
1.89
Wheat
1.72
1.65
1.59
1.60
1.61
1.57
1.62
Corn
1.36
1.31
1.32
1.31
1.35
1.33
1.33
Forest
1.33
1.57
1.59
1.56
1.93
1.61
1.60
Pasture
1.49
1.53
1.66
1.51
2.00
1.60
1.63
Urban
1.39
1.74
2.00
2.21
3.00
2.47
2.13
Wheat
1.85
1.80
1.79
1.82
1.70
1.80
1.79
Corn
1.51
1.60
1.58
1.57
1.74
1.59
1.60
Forest
1.78
1.79
1.87
1.81
2.45
1.91
1.93
Pasture
1.82
2.06
2.07
2.03
2.36
2.15
2.08
Urban
1.84
2.32
2.38
2.54
3.03
2.86
2.50
Wheat
1.59
1.47
1.43
1.48
1.49
1.47
1.49
Corn
1.40
1.38
1.33
1.36
1.46
1.37
1.38
Forest
1.51
1.70
1.80
1.73
2.09
1.79
1.77
Pasture
1.46
1.50
1.51
1.50
1.78
1.57
1.55
Urban
1.64
2.04
2.42
2.51
2.88
2.62
2.35
Wheat
1.63
1.66
1.56
1.61
1.70
1.65
1.64
Corn
1.42
1.27
1.16
1.26
1.18
1.19
1.25
Forest
1.63
1.35
1.32
1.36
1.27
1.31
1.37
Pasture
1.57
1.37
1.23
1.29
1.18
1.25
1.31
Urban
1.38
1.26
1.17
1.23
1.96
1.37
1.40
Wheat
1.86
1.64
1.53
1.62
1.55
1.6
1.63
Average 1.53
1.57
1.60
1.63
1.88
1.69
*
Table 24: Temporal Scale, in months, of plant available soil moisture. Selected stations. All soils and land covers.

78

Station
Ann Arbor, MI
Angelica, NY
Aurora, IL
Big Rapids, MI
Cheboygan, MI
Clinton, IA
Coldwater, MI
Delphi, IN
Dixon, IL
Fond du Lac, WI
Farmington, MN
Fayette, IA
Iron Mountain, MI
Ironwood, MI
Norwalk, OH
Oberlin, OH
Oconto, WI
Owosso, MI
Pine River, MN
Portage, WI
South Haven, MI
Spooner, WI
Sault Ste Marie, MI
Two Harbors, MN
Viroqua, WI
Wooster, OH
Warren, PA
Watertown, NU
Wauseon, OH
Spearmans
Spearmans
Spearmans
Spearmans
Spearmans
Spearmans
Spearmans

1
2
3

Average
Temperature
0.013
-0.005
0.006
0.002
0.001
0.006
-0.007
0.004
-0.002
0.001
0.011
-0.001
0.001
-0.004
-0.004
-0.003
-0.004
-0.005
0.015
-0.004
0.015
0.013
0.004
0.007
-0.005
-0.004
0.004
0.000
0.002
0.02

Maximum
Temperature
0.012
-0.008
0.004
0.007
-0.008
-0.001
-0.014
-0.004
-0.007
-0.011
0.005
-0.002
0.005
-0.002
-0.014
-0.006
-0.002
-0.011
0.013
-0.001
0.008
0.012
0.011
-0.004
0.000
-0.014
0.000
-0.002
0.000
0.04

Minimum
Temperature
0.014
-0.003
0.007
-0.003
0.009
0.013
-0.001
0.014
0.004
0.014
0.017
-0.002
-0.002
-0.008
0.007
0.003
-0.005
0.002
0.016
-0.004
0.022
0.014
-0.002
0.020
-0.009
0.007
0.009
0.003
0.005
0.08

0.31

2.0e-5

6.7e-7

1.4e-3

5.8e-4

0.03

4

0.39

0.75

0.16

5

0.93

0.39

0.76

6

0.18

0.06

0.44

7

0.73

0.98

0.25
o

Table 25: Rate of Change of Yearly Average Temperatures ( C / year). Spearmans rs level of
1

significance Rate of Change of the Beginning Date of Discharge. All observation stations.
79

2

3

4

Rate of Change of Discharge Rate. Rate of Change of Day of Minimum Soil Moisture. Rate
5

6

of Change of Minimum Soil Moisture. Rate of Change of Recharge Rate. Rate of Change of
7

Day of End of Recharge. Rate of Change of Field Capacity Days.

80

Station

Precipitation

Ann Arbor, MI

1.68

Spearmans

Angelica, NY

0.82

Spearmans

Aurora, IL

1.14

Spearmans

Big Rapids, MI

1.12

Spearmans

Cheboygan, MI

0.68

Spearmans

Clinton, IA

-0.34

Spearmans

1
2
3

Level of
Significance
0.037
5.2e-5
0.410

4

2.5e-4

5

0.033

6

1.5e-3

7

1.8e-4
Coldwater, MI
1.15
Spearmans
Delphi, IN
0.28
Dixon, IL
1.20
Fond du Lac, WI
-0.28
Farmington, MN
0.88
Fayette, IA
1.00
Iron Mountain, MI
0.13
Ironwood, MI
0.58
Norwalk, OH
0.70
Oberlin, OH
0.53
Oconto, WI
-0.25
Owosso, MI
0.75
Pine River, MN
0.97
Portage, WI
1.19
South Haven, MI
0.71
Spooner, WI
0.37
Sault Ste Marie, MI
1.74
Two Harbors, MN
1.50
Viroqua, WI
-0.20
Wooster, OH
1.31
Warren, PA
1.22
Watertown, NY
0.87
Wauseon, OH
-0.20
Table 26: Rate of Change of Yearly Precipitation (mm/ year). Spearmans rs level of significance
1

2

Rate of Change of the Beginning Date of Discharge. All observation stations. Rate of Change
3

4

of Discharge Rate. Rate of Change of Day of Minimum Soil Moisture. Rate of Change of
5

6

Minimum Soil Moisture. Rate of Change of Recharge Rate. Rate of Change of Day of End of
7

Recharge. Rate of Change of Field Capacity Days

81

Figure 1: An idealized annual soil moisture curve showing relative levels of soil moisture, Evapotranspiration (ET), and Potential
Plant Moisture Stress (PMS). Taken from Roadside Revegetation: An Integrated Approach to Establishing Native Planets. Chapter 5,
Figure 5.22. For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of
this thesis.
82

Figure 2: Average Annual precipitation (mm) from 1979-2008 across the study region.

83

Figure 3: Average Annual Temperatures (F) from 1979-2008 across the study region.

84

Figure 4: Water Holding Capabilities of Soils by texture. (Figure 5.9. As seen in Foth, 1990)

85

Figure 5: Average Annual Evapotranspiration (mm) for a grassland land cover on a loamy soil, 1979-2008.

86

Figure 6: Average Annual Runoff (mm) for a grassland land cover on a loamy soil, 1979-2008.

87

Figure 7: Average Annual Drainage (mm) for a grassland land cover on a loamy soil, 1979-2008.

88

End of Fall Recharge

Beginning of
Spring Discharge

Discharge
Recharge

Date of Minimum
Soil Moisture

Figure 8: Example of an annual soil moisture curve, 1979-2008, South Haven, MI. Corn land cover and Loam soil type.

89

Figure 9: Average Date of Soil Moisture Discharge averaged across soil textures, Pasture Land Cover, 1979-2008.

90

Figure 10a: Average Date of Minimum Soil Moisture, averaged across soil textures, Corn land cover. 1979-2008.

91

Figure 10b: Average Date of Minimum Soil Moisture, averaged across soil textures, Wheat land cover. 1979-2008.

92

Figure 11: Average Minimum Soil Moisture amount, averaged across soil textures, Wheat land cover, 1979-2008.

93

Figure 12: Average Date of Soil Moisture Recharge, averaged across soil textures, Pasture land cover, 1979-2008.

94

Figure 13: Average Annual Soil Moisture Days, averaged across soils, Pasture land cover, 1979-2008.

95

Figure 14: Trends in Temperature (C/year) across the study area, 1900-2008.

96

Figure 15: Great Lakes Drainage Basin

97

Figure 16: Center of Grids from Midwestern Regional Climate Center.

98

Figure 17: Map of Historical Climate Network Observation Sites used for this study.

99

Figure 18a: Comparison of SWAT plant available soil moisture to Observed plant available soil moisture. Bondville, IL.
100

Figure 31b: Comparison of SWAT plant available soil moisture to Observed plant available soil moisture. Champaign, IL.
101

Figure 31c: Comparison of SWAT plant available soil moisture to Observed plant available soil moisture. Topeka, IL.
102

Figure 19: Rate of Change: Precipitation (mm/year) 1900-2008. Contours are at 0.5 mm/year intervals.

103

Figure 20: Rate of change of Evapotranspiration. Average of all soils, Corn land cover (mm/year), 1900-2008. Contours are 0.1
mm/year.

104

Figure 21: Rate of change of Evapotranspiration. Average of all soils, Forest land cover (mm/year). Contours are 0.1 mm/year.

105

Figure 22: Rate of change of Evapotranspiration. Average of all soils, Pasture land cover (mm/year). Contours are 0.1 mm/year.

106

Figure 23: Rate of change of Evapotranspiration. Average of all soils, Urban land cover (mm/year). Contours are 0.1 mm/year.

107

Figure 24: Rate of change of Evapotranspiration. Average of all soils, Wheat land cover (mm/year). Contours are 0.1 mm/year.

108

Figure 25: Rate of change of Percolation. Average of all soils, Corn land cover (mm/year). Contours are 0.25 mm/year.

109

Figure 26: Rate of change of Percolation. Average of all soils, Forest land cover (mm/year). Contours are 0.25 mm/year.

110

Figure 27: Rate of change of Percolation. Average of all soils, Pasture land cover (mm/year). Contours are 0.25 mm/year.

111

Figure 28: Rate of change of Percolation. Average of all soils, Urban land cover (mm/year). Contours are 0.25 mm/year.

112

Figure 29: Rate of change of Percolation. Average of all soils, Wheat land cover (mm/year). Contours are 0.25 mm/year.

113

Figure 30: Rate of change, return of soil moisture from shallow aquifer. Average of all soils, Corn land cover (mm/year). Contours are
0.01 mm/year.

114

Figure 31: Rate of change, return of soil moisture from shallow aquifer. Average of all soils, Forest land cover (mm/year). Contours
are 0.01 mm/year.

115

Figure 32: Rate of change, return of soil moisture from shallow aquifer. Average of all soils, Pasture land cover (mm/year). Contours
are 0.01 mm/year.

116

Figure 33: Rate of change, return of soil moisture from shallow aquifer. Average of all soils, Urban land cover (mm/year). Contours
are 0.01 mm/year.

117

Figure 34: Rate of change, return of soil moisture from shallow aquifer. Average of all soils, Wheat land cover (mm/year). Contours
are 0.01 mm/year.

118

Figure 35: Rate of change, Runoff. Average of all soils, Corn land cover (mm/year). Contours are 0.1 mm/year.

119

Figure 36: Rate of change, Runoff. Average of all soils, Pasture land cover (mm/year). Contours are 0.1 mm/year.

120

Figure 37: Rate of change, Runoff. Average of all soils, Urban land cover (mm/year). Contours are 0.1 mm/year.

121

Figure 38: Rate of change, Runoff. Average of all soils, Wheat land cover (mm/year). Contours are 0.1 mm/year.

122

Figure 39: Rate of change, Day of Year of the beginning of soil water discharge. Average of SL, L, and SiL. Corn land cover
(days/year). Contours are 0.1 days/year.

123

Figure 40: Rate of change, Day of Year of the beginning of soil water discharge. Average of SL, L, and SiL. Forest land cover
(days/year). Contours are 0.1 days/year.

124

Figure 41: Rate of change, Day of Year of the beginning of soil water discharge. Average of SL, L, and SiL. Pasture land cover
(days/year). Contours are 0.1 days/year.

125

Figure 42: Rate of change, Day of Year of the beginning of soil water discharge. Average of SL, L, and SiL. Urban land cover
(days/year). Contours are 0.1 days/year.

126

Figure 43: Rate of change, Day of Year of the beginning of soil water discharge. Average of SL, L, and SiL. Wheat land cover
(days/year). Contours are 0.1 days/year.

127

Figure 44: Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Corn land cover (µm/(days*year)). Contours are
µm /(days/year).

128

Figure 45: Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Forest land cover (µm/(days*year)). Contours are
µm /(days/year).

129

Figure 46: Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Pasture land cover (µm/(days*year)). Contours are
µm /(days/year).

130

Figure 47: Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Urban land cover (µm/(days*year)). Contours are
µm /(days/year).

131

Figure 48: Rate of change, Rate of soil water discharge. Average of SL, L, and SiL. Wheat land cover (µm/(days*year)). Contours are
µm /(days/year).

132

Figure 49: Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and SiL. Corn land cover, days/year. Contours
are 0.1 days/year.

133

Figure 50: Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and SiL. Forest land cover, days/year. Contours
are 0.1 days/year.

134

Figure 51: Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and SiL. Pasture land cover, days/year.
Contours are 0.1 days/year.

135

Figure 52: Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and SiL. Urban land cover, days/year. Contours
are 0.1 days/year.

136

Figure 53: Rate of change, Day of Year of Minimum Soil Moisture. Average of SL, L, and SiL. Wheat land cover, days/year.
Contours are 0.1 days/year.

137

Figure 54: Rate of change, Minimum Yearly Soil Moisture value. Average of SL, L, and SiL, Corn land cover, mm/year. Contours are
0.1 mm/year.

138

Figure 55: Rate of change, Minimum Yearly Soil Moisture value. Average of SL, L, and SiL, Forest land cover, mm/year. Contours
are 0.1 mm/year.

139

Figure 56: Rate of change, Minimum Yearly Soil Moisture value. Average of SL, L, and SiL, Wheat land cover, mm/year. Contours
are 0.1 mm/year.

140

Figure 57: Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Corn land cover (µm/(days*year)). Contours are
µm /(days/year).

141

Figure 58: Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Forest land cover (µm/(days*year)). Contours are
µm /(days/year).

142

Figure 59: Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Pasture land cover (µm/(days*year)). Contours are
µm /(days/year).

143

Figure 60: Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Urban land cover (µm/(days*year)). Contours are
µm /(days/year).

144

Figure 61: Rate of change, Rate of soil water recharge. Average of SL, L, and SiL. Wheat land cover (µm/(days*year)). Contours are
µm /(days/year).

145

Figure 62: Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and SiL. Corn land cover, days/year.
Contours are 0.1 days/year.

146

Figure 63: Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and SiL. Forest land cover, days/year.
Contours are 0.1 days/year.

147

Figure 64: Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and SiL. Pasture land cover, days/year.
Contours are 0.1 days/year.

148

Figure 65: Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and SiL. Urban land cover, days/year.
Contours are 0.1 days/year.

149

Figure 66: Rate of change, Day of Year of Recharge of soil moisture. Average of SL, L, and SiL. Wheat land cover, days/year.
Contours are 0.1 days/year.

150

Figure 67: Rate of change, Soil Moisture Days. Average of SL, L, and SiL. Urban land cover, (soil moisture days / days) / year.
Contours are 0.1 (soil moisture days / days) / year.

151

Figure 68: Rate of change, Soil Moisture Days. Average of SL, L, and SiL. Wheat land cover, (soil moisture days / days) / year.
Contours are 0.1 (soil moisture days / days) / year.

152

Figure 69: Rate of change: Winter Precipitation (mm/year) 1900-2008. Contours are at 0.25 mm/year intervals.

153

Figure 70: Rate of change: Spring Precipitation (mm/year) 1900-2008. Contours are at 0.25 mm/year intervals.

154

Figure 71: Rate of change: Summer Precipitation (mm/year) 1900-2008. Contours are at 0.25 mm/year intervals.

155

Figure 72: Rate of change: Fall Precipitation (mm/year) 1900-2008. Contours are at 0.25 mm/year intervals.

156

BIBLIOGRAPHY

157

BIBLIOGRAPHY

Alexandersson H, Moberg A. (1997) Homogenization of Swedish temperature data. Part I:
homogeneity test for linear trends. International Journal of Climatology, 17, 25-34.
Andreadis K, Clark E, Wood A, Hamlet A, Lettenmaier D. (2005) Twentieth-Century Drought in
the Conterminous United States. Journal of Hydrometeorology, 6, 985–1001.
Andreadis K, Lettenmair D. (2006) Trends in 20th Century drought over the Continental United
States. Geophysical Research Letters, 33, L10403.
Andresen J (2007) Historical Climate Trends in Michigan and the Great Lakes Region, 2012.
Climate Change in the Great Lakes Region: Navigating an Uncertain Future. Michigan State
University Press, East Lansing MI, pp 17-34.
Andresen J, Winkler J. (2009). Michigan Geography and Geology. US: Pearson Custom
Publishing.
Arnold J, Srinivasan R, Muttiah R, Williams J. (1998) Large Area Hydrologic Modeling and
Assessment. Part I: Model Development. Journal of the American Water Resources Association,
34(1), 73-89.
Arnold R, Jones C. (1987) Soil and Climate Effects upon Crop Production Activity and Nutrient
Use. Soil Scientists Society of America Publication, 19, 9-17.
Baumgart P. (2005) Lower Green Bay and Lower Fox Tributary Modeling Report. Source
allocation of suspended sediment and phosphorus loads tp Green Bay from the lower Fox River
Subbasin using the Soil and Water Assessment Tool (SWAT). Prepared for the Onieda Tribe of
Indians of Wisconsin and the Science and Technical Advisory Committee of the Green Bay
Remedial Action Plan.
Beven K, Kirby M. (1979) A physically based, variable contributing area model of basin
hydrology. Hydrological Science Bulletin, 24, 43-69.
Cambardella C, Moorman T, Novak J, Parkin T, Karlen D, Turco R, Konopka A. (1994) Fieldscale variability of soil properties in central Iowa soils. Soil Science Society of America Journal,
58(5), 1501-1511.
Canadell J, Jackson R, Ehleringer J, Mooney H, Sala O, Schulze E (1996). Maximum rooting
depth of vegetation types at the global scale. Oecologia, 108, 583-595.

158

Cayan D, Georgakakos K. (1995) Hydroclimatolgy of continental watersheds 2. Spatial analyses.
Water Resources Research, 31(3), 677-697.
Changnon S, Jones D. (1972) Review of the influences of the Great Lakes on weather. Water
Resources Research, 8, 223-230.
Cosby B, Jornberger R, Clapp R, Ginn T. (1984) A statistical exploration of the relationships of
soil moisture characteristics to the physical properties of the soil. Water Resources Research, 20,
682-690.
Dai A, Trenberth KE, Qian T. (2004) A Global Dataset of Palmer Drought Severity Index for
1870-2002: Relationship with soil moisture and effects of surface warming. Journal of
Hydrometeorology, 5, 117-1130.
DeLiberty T, Legates D. (2003) Interannual and Seasonal Variability of Modeled Soil Moisture
in Oklahoma. International Journal of Climatology, 23, 1057-1086.
Delworth T, Manabe S. (1988) The Influence of Potential Evaporation on the Variabilities of
Simulated Soil Wetness and Climate. Journal of Climate, 1, 523-574.
Eichenlaub V, Harman J, Nurnberger F, Stolle H. (1990) The Climatic Atlas of Michigan.
University of Notre Dame Press, South Bend, IN.
Etin J, Robock A, Vinnikov K, Hollinger S, Liu S, Namkhai A. (2000) Temporal and Spatial
scales of observed soil moisture variations in the extratropics. Journal of Geophysical Research,
105, 11865-11877.
Findell K, Eltahir E. (1997) An analysis of the soil moisture-rainfall feedback, based on direct
observations from Illinois. Water Resources Research, 33(4), 725-735.
Foth H. (1990) Fundimentals of Soil Science 8th Ed. John Wiley and Sons, New York, NY. (360
pages)
Gaffen D, Wang J. (2001) Late-Twentieth-Century Climatology and Trends of Surface Humidity
and Temperature in China. Journal of Climate, 14, 2833-2845.
Gaffen D, Ross R. (1999) Climatology and Trends of U.S. Surface Humidity and Temperature.
Journal of Climate, 12, 811-828.
Gahrs G, Lanzante J. (2000) The “Clear-Sky Bias” of TOVS Upper_Tropospheric Humidity.
Journal of Climate, 13, 4034-4041.
Gassman P, Reyes M, Green C, Arnold J. (2007) The soil and water assessment tool: Historical
development, applications, and future research directions, Transactions of the ASABE, 50(4),
1211-1250.
159

Georgakakos K, Bae D, Cayan D. (1995) Hydroclimatology of continental watersheds, 1,
Temporal analyses. Water Resources Research, 31(3), 655-675.
Giorgi F, Mearns O, Shields C, Mayar L. (1996) A regional model study of the importance of
local versus remote controls of the 1988 drought and the 1993 flood over the central United
States. Journal of Climate, 9, 1150-1162.
Giorgi F. (2002) Variability and trends of sub-continental scale surface climate in the twentieth
century. Part 1: Observations. Climate Dynamics, 18, 675-691.
Gordan H, Hunt B. (1987) Interannual variability of the simulated hydrology in a climatic model
– implications for drought. Climate Dynamics, 1, 113-180.
Grant R, Hesketh J (1992). Canopy structure of Maize (Zea Mays L.) at different populations:
simulation and experimental verification. Biotronics, 21, 11-24.
Groisman P, Knight R, Karl T, Easterling D, Sun B, Lawrimore J. (2004) Contemporary
Changes of the Hydrological cycle over the Contiguous United States” Trends derived from In
Situ Obervations. Journal of Hydrometeorology, 5, 64-85.
Guttman N. (1999) Accepting the Standardized Precipitation Index: A calculation algorithm.
Journal of the American Water Resources Association.
Hartkamp, AD, De Beurs K, Stein A, and White JW. (1999) Interpolation Techniques for
Climate Variables. NRG-GIS Series 99-01. Mexico, D.F.: CIMMYT.
Henderson-Sellers A. (1989) North American Total Cloud Cover Amount Variations this
Century. Palaeogeography, Palaeoclimatology, Palaeoecology (Global and Planetary Change
section), 75, 175-194.
Heim R. (2002) A review of 20th Century Drought Indices used in the United States. Bulletin of
the American Meteorological Society August 2002, 1149-1165.
Hoaglin D, Mosteller F, Tukey J. (1983) Understanding Robust and Exploratory Data Analysis.
John Wiley and Sons, New York, 447 pages.
Hollinger S, Isard S. (1994) A Soil Moisture Climatology of Illinois. Journal of Climate, 7, 822833.
Hollinger S, Reinke B, Peppler R. (1994) Illinois Climate Network: Site Descriptions,
Instrumentation, and Data Management. Illinois State Water Survey, Chapaign, IL, 62 pages.

160

Huang J, Van Den Dool H, Georgakakos K. (1996) Analysis of Model-Calculated Soil Moisture
over the United States (1931-1993) and Applications to Long-Range Temperature Forecasts.
Journal of Climate, 9, 1350-1362.
Huff F, Angel J. (1992) Rainfall Frequency Atlas of the Midwest. Bulletin 71, MCC Research
Report 92-03. 141pp.
Jayawickreme D, Van Dam R, Hyndman D. (2008) Subsurface imaging of vegetation, climate,
and root-zone moisture interactions. Geophysical Research Letters, 35, L18404, 5 pages.
Jones C, Dyke P, Williams J, Kiniry J, Benson V, Griggs R. (1991) EPIC: An Operational Model
for Evaluation of Agricultural Sustainability. Agricultural Systems, 37, 341-350.
Jones P, New M, Parker D, Martin S, Rigor I. (1999) Surface Air Temperature and its Changes
over the past 150 years. Reviews of Geophysics, 37 (2), 173-199.
Karl T. (1986) The Relationship of Soil Moisture Parameterizations to Subsequent Seasonal and
Monthly Mean Temperature in the United States. Monthly Weather Review, 114, 675-686.
Karl T. (1988) Multi-year Fluctuations of Temperature and Precipitation: The Gray Area of
Climate Change. Climatic Change, 12, 179-197.
Karl T, Knight R. (1996) Indices of Climate Change for the United States. Bulletin of the
American Meteorological Society, 77 (2), 279-292.
Karl T, Knight R. (1998) Secular Trends of Precipitation Amount, Frequency, and intensity in
the United States. Bulletin of the American Meteorological Society, 79 (2), 231-241.
Karl T, Williams C. Quinlan F, Boden T. (1990) United States Historical Climatology Network
(HCN) serial temperature and precipitation data. Publication No. 3404, 83 pages. Available from
Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, TN
37831-6285.
Keetch J, Byram G. (1968) A drought index for forest fire control. USDA Forest Service
Research Paper SE-38, Southeastern Forest Experiment Station, Ashville, NC, 33 pages.
Khaliq M, Ouarda T. (2007) On the Critical values of the standard normal homogeneity test
(SNHT). International Journal of Climatology, 27, 681-687.
Knisel W. (1980) CREAMS: A field scale model for Chemicals, Runoff, and Erosion from
Agricultural Management Systems. U.S. Department of Agriculture. Conservation Research
Report 26. 643 pp.

161

Kunkel K, Andsager K. (1998) Long-Term Trends in Extreme Precipitation Events over the
Conterminois United States and Canada. Journal of Climate, 12, 2515-2527.
Kunkel K, Easterling D, Hubbard K, Redmond K, Andsager K, Kruk M, Spinar M. (2005)
Quality control of pre-1948 cooperative observer network data. Journal of Atmospheric and
Oceanic Technology, 22, 1691-1705.
Landsberg H. (1982) Climatic Aspects of Droughts. Bulletin of the American Meteorological
Society, 63 (6), 593-596.
Lanzante J. (1996) Resistant, robust and non-parametric techniques for the analysis of climate
data: Theory and examples, including applications to historical radiosonde station data.
International Journal of Climatology, 16, 1197-1226.
Letternmaier D, Wood E, Wallis J. (1994) Hydro-Climatological Trends in the Continental
United States, 1948-88. Journal of Climate, 7, 586-607.
Loomis R, Amthor J. (1999) Yield Potential, Plant Assimilatory Capacity, and Metabolic
Effciencies. Crop Science, 39, 1584-1596.
McGuire J, Palmer W. (1957) The 1957 drought in the eastern United States. Monthly Weather
Review, 85, 305-314.
McKee T, Doesken N, Kleist J. (1993) The Relationship of Drought Frequency and Duration to
Time Scales. 8th Conference on Applied Climatology, 6 pages.
Maurer E, Wood A, Adam J, Lettenmaier D. (2002) A Long-Term Hydrologically Based Dataset
of Land Surface Fluxes and States for the Conterminous United States. Journal of Climate, 15,
3237-3251.
Menne M, Durre I, Vose R, Gleason B, Houston T. (2011) An Overview of the Global Historical
Climatology Network Daily Database. Journal of Atmospheric and Oceanic Technology,
(unpublished)
Mishra V, Cherkauer K. (2010) Retrospective droughts in the crop growing season: Implications
to corn and soybean yield in the Midwestern United States. Agricultural and Forest Meteorology,
150, 1030-1045.
Mo K, Chelliah M, Carrera M, Higgins R, Ebisuzaki W. (2005) Atmospheric Moisture Transport
over the United States and Mexico as Evaluated in the NCEP Regional Reanalysis. Journal of
Hydrometeorology, 6(5), 710-728.
Mo K. (2008) Model-Based Drought Indices over the United States. Journal of
Hydrometeorology, 9, 1212-1230.
162

Munger T. (1916) Graphic Method of Representing and Comparing Drought Intensities 1.
Monthly Weather Review, 44, 642-643.
Narasimhan B, Srinivasan R. (2005) Development and Evaluation of Soil Moisture Deficit Index
(SMDI) and Evapotranspiration Deficit Index (ETDI) for Agricultural Drought Monitoring.
Agricultural and Forest Meteorology, 113, 69-88.
Narasimhan B, Srinivasan R, Arnold J, Di Luzio M. (2005) Estimation of Long-Term Soil
Moisture using a Distributed Parameter Hydrologic Model and Verification using Remotely
Sensed Data. Transactions of the ASAE, 48(3), 1101-1113.
Palmer W. (1965) Meteorlogical Drought. Research Paper No. 45. Washington DC: Us
Department of Commerce, Weather Bureau.
Palmer W. (1968) Keeping track of crop moisture conditions, nationwide: The new crop
moisture index. Weatherwise, 21, 156-161.
Pan Z, Segal M, Turner R, Takle E. (1995) Model simulation of impacts of transient surface
wetness on summer rainfall in the United States Midwest during drought and flood years.
Monthly Weather Review, 123, 1875-1581.
Pielke R. (2001) Influence of the Spatial Distribution of Vegetation and Soils on the Prediction
of Cumulus Convective Rainfall. Reviews of Geophysics, 39(2), 151-177.
Price D, McKenney D, Nalder I, Hutchinson M, Kesteven J. (2000) A comparison of two
statistical methods for spatial interpolation of Canadian monthly mean climate data. Agricultural
and Forest Meteorology, 101, 81-94.
Richardson C, Wright D. (1984) WGEN: A model for generating daily weather variables.
USDA-ARS. ARS-8. Washington D.C.
Rodriguez-Iturbe I, Porporato A. (2004) Ecohydrology of Water-Controlled Ecosystems. Soil
Moisture and Plant Dynamics. Cambridge University Press.
Rossby, C. (1939) Relations between variations in the intensity of the zonal circulation and
displacement of the semi-permanent centers of action. Journal of Marine Research, 2, 38-54.
Rowntree P, Bolton J. (1983) Simulations of the Atmospheric Response to Soil Moisture
Anomalies over Europe. Quarterly Journal of the Royal Meteorological Society, 109, 501-526.
Schubert S, Suarez M, Pegion P, Koster R, Backmeister J. (2004) Causes of Long-Term Drought
in the US Great Plains. Journal of Climate, 17, 485-503.

163

Sen P. (1968) Estimates of the regression coefficient based on Kendall’s tau. Journal of the
American Statistical Association, 63, 1379-1389.
Sharply, A and Wiliams J. (1990) EPIC-Erosion Productivity Impact Calculator, 1. Model
documentation. US Department of Agriculture, Agricultural Research Service, Tech. Bull. 1768.
Sawyer J. (1964) Notes on possible physical causes of long-term weather anomalies. WMOIUGG Symposium. Research Development Aspects of Long-Range Forecasting. WMO
Technical Note 162. TP. 79, Geneva, 227-248.
Sheffield J, Goteti G, Wen F, Wood E. (2004) A simulated soil moisture based drought analysis
for the United States. Journal of Geophysical Research, 109D, D24108. 19 pages
Sheffield J, Wood E. (2008) Global Trends and Variability in Soil Moisture and Drought
Characteristics, 1950-2000, from Observations-Driven Simulations of the Terrestrial Hydrologic
Cycle. Journal of Climate, 21, 432-458.
Shepard D. (1968) A two-dimensional interpolation function of irregularly-spaced data.
Proceedings 23rd ACM National Conference, 517-524.
Sun B, Groisman P, Bradley R, Keimig F. (2000) Temporal Changes in the Observed
Relationship between Cloud Cover and Surface Air Temperature. Journal of Climate, 13, 43414357.
Svoboda M. (2000) An introduction to the Drought Monitor. Drought Network News, 12, 15-20.
United States Department of Agriculture (2009) 2007 Census of Agriculture. United States
Summary and State Data. 1, 51, 739 pp.
Wilhite D, Glantz M. (1985) Understanding the drought phenomenon: The role of definitions.
Water International, 10, 111-120

164