THEGOS

x1 ‘ .
,L (.3; O

This is to certify that the
thesis entitled

Simulation of Com ln-Field Drydown

presented by

Scott Daniel Piggott

has been accepted towards fulfillment
of the requirements for the

MS. degree in Biosystems Engineering

 

 

 

 

al/C/hvbz € 5, 3/7!

Date

MSU is an Afiinnative Action/Equal Opportunity Employer

 

LIBRARY
Michigan State
University

 

 

 

—. -A— -.-.-.-.—.-.—---.—.-

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KzlProj/Acc8-PreslclRC/DateDue.indd

Simulation of Com ln-Field Drydown

By

Scott Daniel Piggott

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Biosystems Engineering

2010

ABSTRACT

Simulation of Com ln-Field Drydown
By
Scott Daniel Piggott
Corn (Zea Mays L.) drying in the field is an important consideration for grain
quality and for economic assessment of corn production. With an accurate
model of in-field grain water loss, farmers can select a harvest date that
minimizes yield loss and mechanical drying and maximizes grain quality and
economic benefit. The purpose of this work was to develop a model of corn
drydown with readily available weather inputs. The model structure evolved from
reviewed literature and is of differential form. Model function is dependent on
CERES-Maize, a process oriented corn growth and development model, for
phenological timing. The model also incorporates simulation of grain-filling and
vapor pressure. The resultant model was tested and calibrated using data
collected from a 1999 experiment in Michigan as well data available from the
literature. The 1999 Michigan experiment also reviewed the influence of in-field
precipitation variability and plant population on corn drydown. The work
eventuated in a robust corn drydown model that requires only daily maximum

and minimum temperature and precipitation as weather inputs.

Copyright by
Scott Daniel Piggott
2010

For my mothers and fathers ( great, grand, God, Mom'& Dad),
Without whom this work would lack place.

For my wife, Donna,
Without whom this work would lack passion.

For my children,
Vl/Ithout whom this work would lack purpose.

ACKNOWLEDGMENTS

This work was made possible with financial support in the form of

graduate assistantships and equipment provided by Dr. Joe T. Ritchie, Dr. Ted

Loudon, and Dr. Roger Brook. In addition, I would like to thank the professors at

Michigan State University for their expertise through committee participation

and/or coursework and projects:

Committee Members

Dr. Joe T. Ritchie, Homer Nowlin Chair, Major Professor
Dr. Roger Brook
Dr. Jeff Andresen

Other Professors

Dr. Fred Bakker-Arkema Dr. Ken Poff

Dr. Ted Loudon Dr. Roy Black

Dr. Roger Wallace Dr. Gerald Scwab
Dr. Jiaguo Qi Dr. H. Belcher

Dr. Daniel Hayes Dr. Gary VanEe
Dr. Jim Flore Dr. Francis Pierce

I would also like to offer special recognition to the following people for

their academic and personal support during the best academic experience of my

life.

To the owners of Piggott Farms (Vernon, David, and Daniel) for their
support during this year’s research project.

To the Homer Nowlin Research Family (Joe, Anne, Bruno, Carlos, Brian,
Irene, Aris, Alagarswamy, Ayman, Samira, Daniel, Tom) for their
acceptance of my family into theirs.

To my family (Donna, Danielle, and Kaitlin) for their sacrifice and support.

TABLE OF CONTENTS

INTRODUCTION ..................................................................

CHAPTER 1
LITERATURE REVIEW ......................................................

CHAPTER 2
METHODS AND MODELS ......................................................
Models .......................................................................
General drydown equation .....................................
Use of CERES-Maize as a model platform ................
Modeling Grain filling and developmental water loss...
Estimation of dewpoint temperature .........................
Implementation of a re-wetting parameter for precipitation.
Experiment materials and methods ....................................
Experiment Objective .............................................
Experiment Design ................................................

CHAPTER 3
RESULTS AND DISCUSSION ...................................................
Analysis of results ..........................................................
Formulation of drydown equation ..............................

CHAPTER 4

CONCLUSIONS .......................................................................
BIBLIOGRAPHY .......................................................................

vi

18
18
18
18
19
24
28
31
31
31

39
4O
5O

52
56

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

LIST OF TABLES
Regression analysis for C4 drydown curves
plotted in Figure 15 .....................................

Regression analysis for C5 drydown curves
plotted in Figure 16. ....................................

Regression analysis for C4 drydown curves
plotted in Figure 17. ....................................

Regression analysis for C1 drydown curves
plotted in Figure 18. ....................................

Regression analysis for C3 drydown curves
plotted in Figure 19. ....................................

Regression analysis for C6 drydown curves
plotted in Figure 20. ....................................

vii

4O

41

42

43

45

Figure 1

Figure 2
Figure 3
Figure 4
Figure 5
Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13
Figure 14

Figure 15

Figure 16

Figure 17

LIST OF FIGURES

Schmidt and Hallauer equations and drydown boundary

values. .......................................... f .............
Lansing comparison of min. temp and dewpoint.
Phoenix comparison of min. temp and dewpoint.
Nesmith and Ritchie grain filling data ................
Brooking grain filling data ...............................
Kiniry et.al. grain filling data ............................

Tampa,FL comparison of estimated and measured
dewpoint. ....................................................

Des Moines, IA comparison of estimated and
measured dewpoint. .....................................

Lansing, MI comparison of estimated and
measured dewpoint. .....................................

Boise, ID comparison of estimated and
measured dewpoint. .....................................

Phoenix, AZ comparison of estimated and
measured dewpoint. .....................................

Waco, TX comparison of estimated and
measured dewpoint. .....................................

Farm overview schematic ..............................

Image if field C5 with selected measurement points.

Modeled and measured drydown curves for DK 535
Hybrid in field C4 (reference figure 13). ...............

Modeled and measured drydown curves for DK 535
Hybrid in field C5 (reference figure 13) .................

Modeled and measured drydown curves for DK 493
Hybrid in field C4 (reference figure 13) ................

viii

9
13
14
21
21
22

24

25

25

26

26

27
31

35

4O

41

42

Figure 18

Figure 19

Figure 20

Modeled and measured drydown curves for DK 477
Hybrid in field C1 (reference figure 13) ................ 43

Modeled and measured drydown curves for DK 477
Hybrid in field C3 (reference figure 13) ................ 44

Modeled and measured drydown curves for DK 471
Hybrid in field 06 (reference figure 13). ................ 45

ix

LIST OF ABBREVIATIONS

e°'|'¢jeW = Saturation vapor pressure at the dewpoint temperature.

eTambiem = Actual Vapor pressure of the air.

DBT = Dry Bulb Temperature

rh = relative humidity (%)

Twat: Wet bulb Temperature

WBD = Wet bulb depression

R1= silking phase

R2= blister phase

R3= milk phase

R4= dough phase

R5= dent phase

R6= . physiological maturity

DTT3= Daily thermal time with an 8 degree C base temperature.
TTx = Accumulated thermal time for x phase of development.
P1= The accumulated thermal time from seedling emergence to the end

of juvenile phase where x = 1.

P5: The accumulated thermal time from silking to physiological maturity
where x = 5.
WCWB = water content wet basis

= water content at time t (% dry basis)
Me = equilibrium moisture content (% dry basis)

k = proportionality constant

Mo = initial water content (% dry basis)

t = time (days)

Td = daily average dewpoint temperature (K)

Tmin = minimum daily temperature (K)

Tmax = maximum daily temperature (K)

EF = ZEp/dayl lpANN = daily potential evapotranspiration (m) I annual

precipitation (m)

(gap/day = [(Ep/pw) tday] =daily potential evapotranspiration (m)

tday= period when the sun is above the horizon (estimated from station
Iattitude)

pw= density of. water (kg/m3)

Ep = potential

k devebpmema. = proportionality constant for developmental water content loss

phase = the rate of DTT3 accumulations expressed as a
percentage of the P5 coefficient.

k post.matumy = proportionality constant for developmental water content loss

phase
P = precipitation (mm)
a) = wetting parameter (dimensionless)
t6 = time since the end of linear grain filling (days)
5 = drying constant (dimensionless)

xi

INTRODUCTION

Knowledge of the rate at which corn (Zea Maize L.) dries in the field is
significant to grain quality and farm profitability. An improved understanding of
this process would empower farmers to assess risk regarding weather factors
that may impede harvest timing and economic risk associated with an increased
need for mechanical drying that may degrade grain quality.

The objective of this work was to develop a process-oriented model of
pre-harvest corn kernel water content that is usable at the farm level utilizing a
limited number of environmental inputs. For the duration of this work, the pre-
harvest loss of water from corn kernels is referred to as drydown. In addition to.
drydown simulations, models of daily dewpoint temperature and grain-filling rate
were employed or developed in support of the stated objective. The drydown
model will use CERES-Maize (Jones and Kiniry, 1986), a process oriented corn
growth and development model, for daily weather inputs and for estimation of

plant phenology.

Chapter 1

LITERATURE REVIEW

The displacement of kernel water by dry matter in corn kernels is a
biological process that has received attention from agronomists, plant
physiologists, and engineers. Analysis of this process is dependent on research
scope and has involved complex examination of geometric relationships of
molecules, osmotic potentials, plant physiological development, and atmospheric
interaction that facilitates heat and mass transfer. The focus of this research is
functional plant-atmosphere relationships influencing changes in kernel water
content.

Formulation of a drydown model for corn is dependent on knowledge of
plant characteristics and weather variables that influence kernel-atmosphere
water exchange. Ritchie and Hanway (1997) defined the stages of plant growth
and development for corn. The stages are divided into vegetative and
reproductive periods. Of these phasic stages, the timing and duration of the
reproductive stages are important to the plant’s drydown pattern. Knowledge of
the plant’s progression through reproductive stages as influenced by the weather
is essential for the formulation of an accurate drydown model.

The reproductive stages of corn plant development as defined by Ritchie
and Hanway (1997) include silking (R1), blister (R2), milk (R3), dough (R4), dent
(R5), and physiological maturity (R6). Kernel water content assessment begins

with silking and continues beyond physiological maturity. The focus of the R1

period is kernel structure development. This stage is highly susceptible to
atmospheric stress, especially soil water deficit that desiccates the silks and
pollen grains (Ritchie and Hanway, 1997). The R2, R3, and R4 stages are
associated with grain filling, a time when the water content of the kernels is
displaced by plant assimilates. The influence of weather factors on grain filling
declines with kernel maturity. The R5 or dent stage signals a slowed movement
of assimilates into the kernels. The R6 stage is physiological maturity. This
phase marks the end of dry matter accumulation into the kernels, after which
change in water content is the primary influence of kernel Weight change.

The loss of kernel water initiates with kernel formation in the silking phase.
Silking phase duration is approximately 2-3 days, the time necessary for all silks
on a single ear to be exposed and pollinated (Ritchie and Hanway, 1997).
Knowledge of fertilization initiation is critical to drydown modeling as it marks the
beginning of a nearly constant decline in kernel water content.

The grain filling stages include blister (R2), milk (R3), dough (R4), and
dent (R5). The time between silking and blister is considered a lag period of
grain filling as kernel structure is the focus of kernel development and not dry
matter accumulation. For this reason, the lag period leading to blister (R2) is
segmented from the other grain filling stages with respect to calculation of dry
matter accumulation. The time difference between silking and beginning of dry

matter accumulation is approximately 24 - 28 days (Ritchie and Hanway, 1997).

This time differential can also be expressed in thermal time (TT3). W3 is the

summation of daily thermal time (DTT3) that is used to express the duration of

phenological periods and is expressed in °C/day (Ritchie and Hanks, 1991).
DTT3 refers to the daily thermal time associated with a base temperature of 8 °C.

This system of temperature/time expression is used in CERES-Maize to simulate
all processes except photoperiodic induction. There is little variation in thermal
time for the post-silking lag period of different corn genotypes (Cross, 1975).
This period is approximated at 170 °C/day in CERES-Maize.

The R2 - R4 stages represent the “effective” grain-filling period. At R2
initiation, the kernel water content is 85% water content wet basis (WCWB).
Water contents numerically expressed in this work will use a wet basis where
model variables may be expressed using dry basis. Differences in water content
calculation is due to integration of several research efforts, but all model outputs
and associated measured values for water content will be expressed in wet
basis. Water loss from the kernels during effective grain filling is termed as
“developmental” change in kernel water content as water is being displaced by
translocated assimilates (Brooking, 1985). The effective grain filling period has
been the subject of extensive research efforts. These efforts concentrate on
atmospheric influence over rate and duration of grain filling for various genotypes
of corn. The primary conclusions have three common findings:

1. The rate of dry matter accumulation during the effective grain fill period

is nearly linear.

2. The duration of the effective grain filling period is influenced by

temperature.

3. The rate of dry matter accumulation is affected by plant development

prior to the period.

Several studies have shown the linearity of dry matter accumulation with
respect to time during the effective grain fill period (Sprague, 1936; Kiesselbach,
1950; Johnson, 1972; Cross, 1975; Badu-Apraku, 1983; Brooking, 1985;
NeSmith, 1992; Overrnan, 1995;). This linear trend is a function of the plant’s
ability to translocate assimilates from the vegetative plant parts to the kernels.
Due to this constant movement of materials into the kernels, developmental
water loss from the kernels is dependent on the rate of dry matter accumulation.
Research has found that the rate of grain fill has a direct, positive effect on
drydown rate during the grain filling period (Kang et.a|., 1986). This relationship
is reinforced by the findings of a study that measured the influence of six (6)
weather variables on drydown (i.e. the potential evaporation, wind speed,
precipitation, solar radiation, relative humidity, thermal time). Of these factors,
only thermal time showed a consistent association with kernel water loss during
grain filling (Hallauer, 1960).

Several researchers have worked to associate grain fill rate and
accumulated daily thermal time. Johnson and Tanner (1972) calculated the
length of effective grain fill in terms of thermal time. CERES-Maize utilizes a
genotype specific thermal time from silking to maturity. This value, termed as the
P5 coefficient, includes the period of approximately 170 °C/day from silking to
initiation of the effective grain fill time. The P5 phenotype coefficient minus the

thermal time representing the lag phase is the thermal time required to achieve

physiological maturity from the beginning of the effective grain filling period. The
effective grain filling period ends when 95% of P5 has occurred (Kiniry, 1985).
Research has demonstrated that under high temperature conditions, assimilates
that remobilize from other plant parts account for a greater proportion of kernel
weight gain than assimilates produced by photosynthesis during grain filling
(Badu-Apraku, 1983). Extreme temperatures (high and low) and soil water
content deficits prior to grain-filling decrease yield by truncating the duration of
the effective grain-filling period (Badu-Apraku, 1983; NeSmith, 1992).

With continued focus on the effective grain filling period, researchers now
better understand developmental water loss from kernels as a function of grain
filling parameters and weather variables. Schmidt and Hallauer (1966) developed
a model that utilizes a series of linear equations to represent drydown for several
water content ranges. The equation used for the effective grain fill period is a
function of daily average temperature. The authors claimed a 95% confidence
interval for modeled values around measured infield drydown for “any field, any
year”. Other experiments utilized statistical regression to determine relationships
between drydown per unit thermal time and measured values (Kang, 1986;
Cavalieri, 1984; Dwyer et.a|., 1994). The proposed models are limited in
application as simulations are dependent on assumed relationships between
physiological stages and water content.

Near the end of the effective grain filling period, assimilate transfer into
the kernels slows. The stage is called dent, or the R5 phase. The duration of

the R5 phase can be truncated by adverse weather conditions such as an early

frost that can promote premature physiological maturity. This circumstance can
cause a decrease in yield due to premature stoppage of assimilate flow into the
kernels and can cause delays in harvest operations as frost-damaged corn is
slow to dry (Ritchie and Hanway, 1997). Johnson and Tanner (1972) paired this
phase of grain fill with physiological maturity and labeled it as the “leveling off of
dry matter accumulation”. In their research, this phase accounted for
approximately 10% of the total dry matter accumulation. 5% of P5 is used to
represent the duration of this phase in CERES-Maize (Kiniry, 1985).

The dent phase of corn kernel development terminates in physiological
maturity, or the R6 phase. Near the end of dent, a region of cells several layers
thick forms between the basal endosperm of the kernel and the vascular area of
the pedicel. As maximum dry matter accumulation is approached, these cells
compress into a dense layer that appears black to the naked eye (Daynard,
1969). The formation of this black, abscission layer represents the end of
assimilate transfer into the kernels and maturity. Several efforts have been
made to link black layer formation to a common kernel water content (Schmidt,
1966; Rench, 1971; Daynard, 1972; VanEe, 1979; Brooking, 1990; Ritchie and
Hanway, 1997). However, these values range from 25% - 43% for a variety of
genotypes and weather conditions. Due to this wide range of values, attempts to
model drydown processes using fixed water content values to represent black
layer formation can lead to inaccurate simulation. Schmidt and Hallauer’s work
made no mention of black layer, but found 30% to be a break-point in model

input requirements with temperature dominating the model above 30% and wet

bulb depression below. This approximation was later changed to 37% by Van
Ee (1979) upon further analysis of the data. Rench (1971) stated that use of
30% is only a rough estimate of physiological maturity. Daynard (1972)
calculated mean values Of 33.7% and 33.9% for a black layer water content in
1969 and 1970, respectively. Based on the reviewed literature, attempts to link
the formation of “black layer” with a fixed water content is an approximation at
best

The literature reviewed thus far concentrates on kernel development and
the loss of “developmental water”. A black abscission layer has stopped the flow
of assimilates into the kernel and maximum dry matter has been attained. The
kernel water content at this time is between 24 - 43%. The exchange of fluids is
no longer between the plant and the kernel but between the kernel and the
atmosphere. The following section will discuss research attempts to understand
post-maturity drydown in maize.

Most corn growth and development models, such as CORNF (Stapper
and Arkin, 1980) and CERES-Maize, end simulation of plant phenology with
physiological maturity. One of the first works to model post-maturity drydown
was the work performed by Schmidt and Hallauer in 1966. The equations and
the associated water content boundary values are presented in Figure 1. Notice
the only weather inputs used in the calculations are dry bulb temperature (DBT in

degrees F) and wet bulb depression (WBD in degrees F).

Figure 1: Schmidt and Hallauer equations and drydown boundary values.

 

 

 

 

 

 

 

 

 

 

 

75% (Milky roasted ear)

50% (Complete Dent) R = -2.00 + 0.047*DBT
30% (Average Maturity) R = -0.54 + 0.021*DBT
25% (Matured) R = -0.08 + 0.119*WBD
20% (Drying) R = -0.432 + 0.146*WBD

 

 

 

. DBT = Dry bulb temperature
. WBD = Wet bulb depression
a R = Kernel Water Content Reduction (%lday)

Van Ee (1979) changed the model to better statistically represent the
Schmidt and Hallauer data. The resultant model was implemented into a corn
production simulation called CORNSIM. Only the post-maturity drydown was
changed as the equation representing developmental water loss remained as
Schmidt & Hallauer hypothesized. The major changes to the model included
changing the water content that represents the end of developmental water loss
from 30 to 37% and the incorporation of equilibrium water content. While the
change of break-point water content enhanced model output correlation with the
data, the introduction of equilibrium water content is a major improvement of the
model’s ability to mimic the asymptotic behavior of corn infield drydown.
Equilibrium water content is defined as the tendency of a stored farm product
toward a water content value that is controlled by the ambient environment
(Henderson, 1952). Equilibrium water content establishes a lower limit to which
grain can dry (Thompson, 1969).

While the model fit the base data, the structure of the model limits robust
application in several ways. First is the dependence of the model on use of fixed

water content values. Secondly, the model does not take into account possible

re-wetting of the kernels by precipitation. Third, the model input of wet bulb
depression as a function of wet bulb temperature is not readily measured nor
available. Fourth, the model does not differentiate plant genotype
characteristics. Many research findings have emphasized that genotype specific
traits such as phenologic timing, ear size, husk number, angle of ear, and husk
and pericarp permeability influence the rate of drydown (Crane et.al, 1958; Purdy
and Crane, 1967; Troyer and Ambrose, 1971; Cavalieri and Smith, 1985; Baron
and Daynard, 1984);

A break from the use of discontinuous drydown models came when Bruns
(1975) developed a post-maturity drydown equation using the following

differential form.

M1 = Mu -t) - (AMi/At) * t

- (AMi/At) is change in water content (%lday) = f (evaporation rate,

precipitation, occult precipitation, vapor resistance factor, wetting resistance
factor, and number of days since physiological maturity)
- M is water content (%)

- l is day number

. t is time interval in days

While this model’s input parameters are less readily measured on farms,
the differential structure of the model, lack of reliance on subsequent water
content values for change in parameters, and use of a wetting parameter warrant
further investigation. For the samples collected, Bruns observed a 90%

confidence interval for the daily mean observed water content, with

10

simulated/measured difference values ranging from 0.18% to 0.48% with an
average of 0.34% and a sample standard deviation of 1.0%.

The drydown models reviewed herein provide a collection of attributes
that contribute to a greater understanding of drydown processes. First, the
preceding work provides an understanding that certain weather factors influence
drydown. These variables include forms of temperature, vapor pressure, and
wetting. A second beneficial characteristic of the reviewed models is the use of
a differential equation structure as posed by Bruns. This model structure allows
for a continuous simulation of processes based on daily weather variables. Third
is the use of equilibrium water content in drydown analysis, as proposed by Van
Ee (1979). Implementation of these characteristics is essential to accurate
modeling of drydown. ‘

The relative accuracy of Bruns post-maturity drydown model acts as a
bridge to another segment of drydown research, that of mechanical grain drying
simulation. Henderson and Perry (1955) reported that the declining water
content of grain was inversely proportional to the water to be removed. This
statement takes the following equation form:

dM/dt = -k(M - Me)
- M = water content at time t (% dry basis)
- Me = equilibrium water content (% dry basis)
- k = proportionality constant

The solution to this equation is as follows:
k
MR = water content ratio = (M - Me)/(Mo - Me) = e' t

- Mo = initial water content (% dry basis)

11

This model utilizes equilibrium water content. Henderson (1952) developed the

following equation to calculate Me:
h = e(4: (t+460) MeAn)

1 - r
rh = relative humidity
c=1.1*10'5

- n = 1.90

Several models for equilibrium water content have been developed for corn
(Strohman, 1967; Thompson et.al, 1968; Bakker—Arkema et.a|.,1974). Two of
the most adopted models are those formulated by Thompson (1969) and
Bakker—Arkema (1974). Thompson’s model is a revision of Henderson’s (1952)

model with the following changes:

(-c (t+50) Me"n)

1 - rh = e
- rh = relative humidity
c=3sz*1o5
- n = 2.0

Bakker—Arkema’s model has been found to be more accurate than Thompson’s
model in controlled environments, but Thompson’s model may be accurate
enough to estimate equilibrium water content estimation for grain that has
constant atmospheric exposure.

A common requirement for equilibrium water content calculation is relative
humidity. Relative humidity is not an input variable for most com production
models, including CERES-Maize, and it is not often measured on farms.
Weather variables utilized by CERES-Maize include minimum daily temperature

(degrees C), maximum daily temperature (degrees C), daily precipitation (mm),

112

and daily solar radiation (MJ/m2). To keep this model applicable at the farm

level, an existing model should be utilized or a new model created to simulate
relative humidity as a function of commonly measured daily weather variables.

Several models use daily minimum temperature as a surrogate for
dewpoint temperature and corresponding vapor pressures in the absence of
recorded vapor pressure (Abawai, 1995; Koon, 1986). While this approximation
is relatively accurate for climates with relatively high annual rainfall averages, this
assumption is imprecise when considering more arid climates. Figures 1 and 2
chart the average daily dewpoint temperature versus the minimum daily
temperature for thirty year average data sets for Lansing, Michigan (relatively
high annual rainfall) and Phoenix, Arizona (Arid).

Figure 2: Lansing comparison of min. temp and dewpoint

 

Minimum Daily Temperature Vs. Average Daily Dewpoint for
Lansing, MI, USA

 

 

 

Tm'n

 

 

I Tdew

 

Degrees C

 

 

 

0 . t t r t 1. t t
0 50 100 150 200 250 300 350 400

Day of the year

 

 

 

13

Figure 3: Phoenix comparison of min. temp and dewpoint

 

Minimum Daily Temperature Vs. Average Daily Dewpoint for
Phoenix, AZ, USA

 

 

30 -—

25 ~~

20 +
3 15 * .Tmin
g: 10 .t I Tdew
o

 

 

 

 

       

 

L
Y

200 250 300

  

 

350

 

 

Days of the year

 

 

 

The graphs illustrate both the accuracy and inaccuracies of
implementing this assumption. Researchers have developed a few regression
based models that attempt to model the daily dewpoint temperatures using the
daily minimum temperature (Vasic, 1981; Koon, 1986; Abawai, 1994; Kimball
et.a|., 1997). The most comprehensive of these is the work performed by
Kimball, Runnings, and Nemani (1997). Their work developed an empirical
model to improve the accuracy of the minimum temperature estimates of
dewpoint temperature using daily air temperature, annual precipitation, and
estimated daily evapotranspiration. The model used daily meteorological
weather data for 52 weather stations in the continental United States and Alaska.

The model structure and associated sub-calculations are as follows:

14

Td = Tmin [ -o.127 +1.121(1.003 - 1.444 EF + 12.312 (EF)2 - 32.766 (EF)3

+0.0006(Tmax - Tmin)]
Td = daily average dewpoint temperature (K)
Tmin = minimum daily temperature (K)
Tmax = maximum daily temperature (K)

EF = (Ep/dayl (pANN = daily potential evapotranspiration (m) I annual precip. (m)
(Eta/day = [(Ep/pw) tday] =daily potential evapotranspiration(m)

tday= period when the sun is above the horizon (estimated from station Iattitude)
pw= density of water (kg/m3)

Ep = potential evapotranspiration (Priestly & Taylor method, 1972)

Accurate values of Me are dependent on accurate estimation of relative
humidity. The model of relative humidity model developed by Kimball et.al.
(1997) reduced humidity estimation errors by up to 80% for the 52 sites used in
the original analysis. A model that requires Me for operation should rely on such
a model if measured values for vapor pressure are not available.

Henderson and Perry’s (1955) grain drying model and the equilibrium
water content models were formulated for mechanical drying simulation and
discretion must be exhibited when attempting their use for infield drydown. As
mentioned, these equations represent controlled or closed systems. Application
of these models to in-field drydown requires an understanding of
plant/atmosphere relationships. The most glaring difference between the

controlled and uncontrolled system is re-wetting of the grain by precipitation.

15

The purpose of an equilibrium water content term in mechanical drying
models is to represent re-wetting of the grains. When the equilibrium water
content is greater than the existing water content of the grains, the grain water
content trends toward equilibrium with the atmosphere. The phenomenon where
the adsorption isotherm obtained by placing a dry biological material in
atmospheres of increasing relative vapor pressures of water lags behind the
desorption isotherm obtained by placing the wet product in atmospheres of
decreasing relative humidity is termed hysteresis (Ngoddy & Bakker-Arkema,
1975). Many changes to Henderson’s original equation (1955) have been
formulated to better model this hysteresis effect. The bulk of these modifications
are formulated for input into Thompson’s model for drying simulation (Thompson
et.a|., 1968), which is derived from Henderson’s findings. Many of these
modifications are attempts to increase the accuracy of the model for precisely
controlled systems where slight adjustments in a constant or use of a different
model for a system variable calculation are required for hysteresis manifestation
(Javas et.a|., 1991; Krueger and Bunn, 1985; 1988).

Bruns addressed the effects of re-wetting on infield drydown in his model,
reporting that the amount of wetting due to dew formation and higher equilibrium
water content seems to be small compared to the wetting due to precipitation,
but the duration of those conditions is important (Bruns et.a|., 1975). The use of
an equilibrium water content variable is important in drydown modeling as it sets
a lower limit of water content that the grain must approach and follow

asymptotically. Therefore, the incorporation of precipitation into a drydown model

16

must amplify the hysteresis revealed by differences between grain water content
and the equilibrium moisture content. Researchers have concluded that the
magnitude of water content change due to hysteresis, called the hysteresis loop,
is sometimes dependent on the speed and frequency with which the loop is
traversed, lending that the effect of precipitation on drydown may be time
dependent (Ngoddy and Bakker—Arkema, 1975).
In conclusion, the conception of a robust, easily applied model for infield
drydown of corn is conditional to the inclusion of the following characteristics:
1. The model must work relatively independent of assumed relationships
between water content values and plant phenological stages.
2. The model’s environmental inputs should be limited to relatively easily
measured parameters.
3. The model should incorporate equilibrium water content and precipitation for
the purpose of modeling grain hysteresis and for setting a lower boundary of
water content.

4. The model must simulate genotype specific plant phenology.

17

Chapter 2
METHODS AND MODELS

Models
General drydown equation

The objective of this work is to develop a process-oriented drydown model
that is usable at the farm level utilizing a limited number of weather inputs. The
basic structure of the model will be the first order differential equation proposed

by Henderson and Perry (1955).

MR = moisture ratio = (M - Me)/(Mo - Me) = e'kt

Use of CERES-Maize as a model platform

The functional corn growth and development model CERES-Maize will be
utilized as the platform of operation for the drydown model. The new model will
rely on CERES-Maize for calculation of plant phenology with respect to genotype
specific variables and for daily weather parameters. These measured

parameters include minimum and maximum daily temperature (degrees C), daily

accumulated precipitation (mm), and daily accumulated solar radiation (MJ/mz).

The accuracy of the drydown model is dependent on accurate calibration of
CERES-Maize with respect to site conditions (soils, weather) and experiment

design (plant population, fertility, genotype, management practices).

18

The proposed starting point of the model is the CERES-Maize
approximation for time of silking. The model will progress from silking to initiation

of the effective grain filling period by accumulating 170 °C/day.

Modeling Grain filling

Progression from silking to the beginning of the effective grain filling
period is an exclusive function of daily temperature in the proposed model. The
next stage of the model is simulation of the grain-filling period. The genotype
specific P5 coefficient of CERES-Maize is an estimation of the amount of daily
thermal time required for a corn plant to progress from silking to physiological

maturity. As outlined in the literature review, CERES-Maize uses the difference
between 95% of P5 and the TT3 required to connect the R1 (silking) and R2

(blister) phases of reproductive development to model linear grain filling duration.

This is better outlined in the equation below.
Effective grain filling duration (TTa) = [.95 * P5(°C/day)] - 170(°C/day)

The proposed drydown model will follow this methodology but will change the .95
multiplier to .90 to more closely align with other work (Johnson and Tanner,
1972). Using data from performed research, CERES-Maize was calibrated for
each site and year of data collection (Kiniry and Ritchie, 1984; NeSmith, 1992;
Brooking, 1990). Genotype specific information, especially the P5 coefficient if
not known, was estimated to best fit the experimental data using CERES-Maize.
For each experiment, data sets were obtained that shared these common
elements:

1. Date of silking initiation.

19

2. Value of the P5 coefficient.

3. Daily accumulated values of DTT3 from date of silking to physiological

maturity.
4. Measured values of individual kernel weight with respect to time.

The duration of the effective grain filling period was calculated for each
data set as a function of respective genotype-specific P5 values. The resultant
thermal time was then expressed as 100% of the dry matter accumulation for the
effective grain filling period. The rate of effective grain filling as a function of
daily thermal time accumulation was then expressed as a percentage of the of
the total dry matter accumulation for the effective grain filling period. These
values are compared with measured grain filling for each data set (also
expressed as a percentage of maximum dry matter with respect to time for the

effective grain fill period) in Figures 4-6.

20

% accum of D'fT8 for effective

% accum of DTT for effective

grain filling

grain filling

and
% accum. of total dry matter

and

% accum. of total dry matter
0
01

Figure 4: Nesmith and Ritchie grain filling data

NeSmith, 1992

 

   

 

+ D‘lTB
+ Measured

    
 

 

 

 

 

 

FCOFCOV-(OFCOF
FFNNC‘OC‘OV’V‘ID

(O
In

Days after silking (days)

Figure 5: Brooking grain filling data

Brooking, 1990

 

 

0.6 1»

0.5 4

0.4 —»

0-3 ‘” + DTT8 l

0.2 4

0.1 4
0 _

FCDv-(DFCOFCOPCD
FFNNCOC‘OV'V

 

 

+ Measured i

 

 

FCC
Into

Days after silking (days)

21

 

Figure 6: Kiniry et.al. grain filling data

Ritchie 8: Kiniry, 1984

 

 

 

 

 

2 ~ -

.3 g 0.9 —r f

. 3: +

5 Z‘ ’ «I. ‘

8 E 3 06 ..

00 g '0 5 '

I: c g 3 0'5 4* K

e E '15 0.4 4» I

2 m g 0.3 «r +0113
3 3 0'2 T t" +Nbasured
g r: 0.1 4 q /

.\° °\ 0 -

 

 

FOFCDT—(O‘.
NNCOC‘OVV'LO

Days after silking (days)

1
6
11
16

(D
In

As shown graphically, the grain-filling rate and thermal time accumulation
closely correlate for the analyzed data sets (all r2 >0.98). A model of the grain-

filling rate during effective grain-filling period will be used in the drydown model to

represent the analogous displacement of developmental water from the kernels.
This function uses a smaller percentage of TTg to determine a linear function for

grain-filling as use of a high percentage of thermal time may conflict with the dent
stage when assimilate movement into the kernels slows and grain-filling is no
longer linear. The model does not attempt to simulate the period between the
end of linear grain-filling and physiological maturity.

As the rate of dry matter accumulation is not a dependent function of
temperature, simulating kernel water loss as a dependent function of
temperature can only be viewed as a model and not a dependent relationship.

Research has found no direct, positive correlation between temperature and

22

kernel water loss prior to maturity nor any other weather variable (Hallauer &
Russell, 1960). Research has found that the rate of grain fill has a direct, positive
effect on grain water content reduction per unit thermal time during the grain
filling period (Kang et.a|., 1986).

To understand the association of grain filling and kernel water loss, a
common time frame of reference is required. Research has found that the
approximate water content of the total kernel weight at the beginning of the
effective grain filling period is approximately 85%, leaving 15% of the total weight
to structural dry matter (Ritchie and Hanway, 1997). Using this fact as an origin,
the following conditions were set for analysis of the drydown model:

1. Two proportionality constants are needed to describe developmental and

post-maturity water content.
Ldeve|ogmenta| = proportionality constant for developmental water loss phase

= thermal time expressed as a fraction of the effective grain filling period

duration.
Lpost-maturim = proportionality constant for post-maturity water content loss

phase expressed as a constant.

2. The initial water content M0 is assumed to be 85% at the beginning of
effective grain filling.

As described above, the proportionality constant used to calculate the period of
developmental water loss has been initially set to the rate of thermal time
accumulation for the effective grain filling expressed as some fraction of P5.

This is only an initial setting.

23

Estimation of dewpoint temperature

The remaining input variable of the existing model that can be directly
calculated using a measured parameter is Me. Relative humidity is required to
calculate Me but is not a data requirement for CERES-Maize. To retain farm
level applicability, relative humidity is calculated using an empirical model to

estimate dewpoint temperature (Kimball, Runnings, and Nemani,1997).

Td = Tmin {-0.127 +1.121(1.003 - 1.444 EF + 12.312 (EF)2 - 32.766 (EF)3

+0.0006(Tmax - Tmin)]
The only additional model inputs required for site characterization are estimated

annual rainfall (m) and site latitude. Of these, only estimated annual rainfall is
not provided by standard CERES-Maize weather files.

The model was tested against 30-year weather data sets for sites in
Florida, Iowa, Michigan, Idaho, Arizona, and Texas. The associated graphs of

the modeled and measured daily dewpoint data are represented in Figures 7-12.

Figure 7: Tampa,FL comparison of estimated and measured dewpoint.

Estimated vs. Measured Daily Dewpoint for Tampa, FL, USA

 

 

 

o Tdew
. Tdew Estimated

 

 

 

Degreesc

 

 

 

 

0 50 100 150 200 250 300 350 400
Day of the year

24

Degrees C

 

Figure 8: Dew Moines, IA comparison of estimated and measured dewpoint.

Estimated vs. Measured Daily Dewpoint for Des Moines, IA, USA

 

 

 

e Tdew

 

 

D Tdew Estimated

 

Degrees C

 

 

 

 

 

Day of the year

Figure 9: Lansing, MI comparison of estimated and measured dewpoint.

Estimated vs. Measured Dally Dewpoint for Lansing, MI, USA

 

 

I

e Tdew
I Tdew Estimatedj

 

 

 

 

 

 

 

Day of the year

25

Degrees C

Figure 10: Boise, ID comparison of estimated and measured dewpoint.

12

         

Estimated vs. Measured Daily Dewpoint for Boise, ID, USA

 

 

    

 

_ '1
'Ill'u.
Elf-.- 5-

 

 

 

Day of the year

 

e Tdew
u Tdew Estimated

 

 

 

Figure 11: Phoenix, AZ comparison of estimated and measured dewpoint.

Degrees C

18

16 <-
14 .
12 4
10 <-

Estimated vs. Measured Daily Dewpoint for Phoenix, AZ, USA

 

 

ONAODCD

 

. - I“- A

I. ,A I; A ‘I ’
. , I

‘A'I’-"‘._Av"” _‘I'- s

‘ - a' T ‘
' A - A
. II A
J,‘ A 3

-
..... A .-.-_

......

 

 

Day of the year

26

 

 

A Tdew
II Tdew Estimated

 

 

Figure 12: Waco, TX comparison of estimated and measured dewpoint.

Estimated vs. Measured Daily Dewpoint for Waco, TX, USA

 

25 we

 

o Tdew
I Tdew Estirmted

 

 

 

 

 

 

 

 

 

[by of the year

For this analysis and the original error analysis performed by Kimball et.al.
(1997), the model relatively reduced humidity estimation errors by up to 80% in
semi-arid and arid sites (like Phoenix, Idaho, and Texas) and had minimal effects
when the minimum temperature based humidity estimates were relatively
accurate (Michigan, Iowa, and Florida). Based on these findings, this model was
integrated into the proposed drydown model calculations where the estimated

daily dewpoint temperature is used to solve for relative humidity.

Implementation of a re-wetting parameter for precipitation

The model proposed for simulation of drydown was originally
intended to represent post-harvest mechanical drying. This paper has described
the benefits of its application to pre-harvest drying. Adaptation of the model

structure to simulate drydown is reliant on the model’s ability to expand and

27

simulate plant/atmosphere interaction. The weather parameters that most
influence drydown are temperature, relative humidity, and precipitation. Of
these, only precipitation is not expressed in the proposed model. Temperature
and relative humidity are represented in the model via equilibrium water content
and the proportionality constants if they are defined as functions of thermal time.
The difference between the initial water content (Mo) and Me represents
both desorption and adsorption of water tolfrom the kernels depending on their
respective magnitudes. The difference between Mo and Me is initially very large
and reduces exponentially with time. The initial water content is 85% while Me is
a function of atmospheric conditions that are generally much lower in water
content. As the model increments daily, Mo takes on the value of the previous
day’s kernel water content minus the water lost to drydown during that day.
When Me > Mo, the grain is being wetted and the grain water content moves
toward the water content of the atmosphere. For grain drying in an enclosed
system, the model can simulate re-wetting of the grains with the difference of
kernel and atmospheric water content alone or with simple modifications to the
proportionality constant. Krueger & Bunn (1988) demonstrated hysteresis by
performing independent simulations for the same data sets with different models
for equilibrium water content. The difference in simulation results correlated with
measured values of hysteresis. These methods will not suffice for simulation of a
wetting with precipitation. To simulate wetting, a term must be added to the
original equation structure to increase the magnitude of Me when a precipitation

event occurs. Simulation of the effect of precipitation on drydown would only be

28

required for post-maturity simulation as precipitation has not been shown to
effect developmental moisture content changes. The following model structure

amendments are proposed to simulate the effects of precipitation.

IF (TTB< P5). MDEVELOPMENTAL= (Me) +(Mo - Me)e_kdevelopmental 1”

IF (TT8> P5):

MPOST-MATURITY = (Me + Pe-w(tg)) +(Mo - (Me + Pe-m(tg)))e_kpostmaturity it

P = precipitation (mm)
0) = wetting parameter
t9 = time since the end of linear grain filling (days)

The form and placement of the precipitation term into the drydown
equation was selected for experimental testing based on the following
assumptions. First, the exponential form was selected to better represent
reduction of the hysteresis loop as its size is sometimes dependent on the speed
and frequency with which the loop is traversed, sunnising that the effect of
precipitation on drydown may be time dependent. A hypothesized declining
effect of precipitation on drydown with respect to time gains feasibility as other
factors such as loss of kernel membrane permeability, increased water shedding
by the ear due to development of an obtuse ear angle with the top of the stalk,
and increased air movement between stalks due to leaf dropping contribute to
this reasoning. Secondly, making the term additive to the equilibrium water
content forces the term to zero when there is no precipitation thus retaining the

original model structure. Third, initiation of the precipitation term’s influence is at

29

the end of linear grain filling. Using this initial time removes extreme values from
the decline in the precipitation curves. The precipitation term is not used until
100% of the P5 thermal time has accumulated. This allows for an acclimation
period for the precipitation term that represents about 5% of the P5 duration.
Creation of this acclimation period reduces the dramatic effect of a precipitation

event that may occur near the end of grain filling.

Experiment Methods and Materials

Experiment Objective

The objective of this work is to test the hypothesized drydown model using
data sets collected spatially and temporally. The following questions are the
focus of experimental design and hypothesis testing:

1. Can the model work relatively independent of assumed relationships between
kernel water content values and plant phenological stages?

2. What is the influence of spatial variability of precipitation on drydown? Are
there other required model inputs for accurate implementation at the
producer level?

3. Does the proposed precipitation term in the model accurately describe the
hysteresis effect? Are there other requirements for explaining this effect?

4. How sensitive is the model to changes in the CERES-Maize P5 coefficient?

Experimental Design

The experimental phase of this project was performed near Fowler,

Michigan, US, during the summer growing season of 1999. The farm utilizes

30

approximately 1200 acres of land to raise corn, soybeans, wheat, and alfalfa.
The corn, soybean, and wheat enterprises shift spatially on a three- year crop
rotation. Soybeans and wheat utilize conservation tillage practices with respect
to planting where as the corn enterprise relies on prepared soils and
conventional planting. The farm soils range from sandy loam to heavy clay
loams. Most of the farm is moderately well drained.

The 1999 corn enterprise included 270 acres in six different field locations

within the 3-mile radius of the farm as shown below in Figure 13.

Figure 13: Farm overview schematic

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Approx.1
mile
AI H —"—'
A 0T6
B F
ccsE L
A D 'H
B
04 1: G
C E
0
Cl
C 8 North Side of Farm
—— —--A‘— ——-__._I_
2 South Side of Farm
C2
8 A
c C1 E * iWeather Stations

31

Planting dates ranged from May 4th through May 14‘", 1999. The hybrids used
for the experiment were Dekalb brands 535, 493, 477, and 471 that range in
relative maturity’s from 103 - 97 days, respectively. The farmer selected the
hybrids based on previous success. Soil characteristics, management practices
regarding fertilization and planting, and genotype specific estimates of plant
phenology were quantified for CERES-Maize input for field and genotype specific
simulation.

With the base information for CERES-Maize set, the remaining inputs to
the model were measured weather variables. Two (2) Ll-COR weather stations
were installed on the farm, one representing the “north” ‘/2 of the corn enterprise
and the 2"d representing the “south” one-half. The weather stations were
calibrated with respect to one another twice during the season by placing the
stations side-by—side for a period of two days and changing the calibration
coefficients of a single unit. The weather stations began logging data prior to
planting and continued through harvest.

In addition to weather station measured variables of temperature and
solar radiation, a network of 32 wedge style Tru-Chek® rain gauges was
installed throughout the corn fields on the farm. Rain gauges were placed to
best characterize the total area of the field as one rain gauge represented
approximately 8 - 9 acres. The reasoning for intense sampling of precipitation
was two-fold. First, the sample points for drydown were undetermined at the

beginning of the season and as precise precipitation quantities were imperative

32

to model accuracy, a network approach to rain measurement reduced the need
for interpolation. Second, spatial variability of precipitation does exist at the field
level (Basso, 1998) and needs to be quantified for its effects on plant growth and
development as well as its unknown effects on in-field drying.

Precipitation was initially measured after every independent rainfall event
for all gauges. This practice eventually became extensive as the crop
developed, requiring a global positioning system to locate the gauges. The
measurement of precipitation was eventually broadened to larger time intervals
for the vegetative and reproductive phases of development with more intense,
event specific measurement during the post-maturity period. A differential global
positioning system was utilized for the duration of the experiment for data
collection of drydown and precipitation measurements. To eliminate the pending
canopy interference with precipitation measurement, the rain gauges were
placed atop 4 foot sections of PVC pipe that were secured to the top of a steel
fence post that was driven into the ground past the base spade. Vertical
leveling of the gauges was checked intermittently throughout the season.

A third source of weather data in the form of radar estimated weather
parameters was provided by WSI, Inc. Two measurement “sites” were provided
to represent the northern and southern halves of the farm as shown in Figure 13.
Each data set was based on 2 km resolution radar data. These quantities were
used for comparison with measured values and not for model inputs.

Limited inspections of crop growth and development were performed

throughout the season to check the CERES-Maize modeled phenology. Airplane

33

derived remote sensing images were obtained on July 6‘", 1999 for each of the
corn fields. The images represented a 1 m resolution aerial view of in-field crop
growth and developmental differences for the approximate V15 stage (about 10 -
12 days prior to tassel. The infrared band of the images was singled out for
further analysis as it visibly showed the greatest amount of spatial variability
when compared to the other bands. The correlation of a single band of an
interlaced remotely sensed image and yield components of corn has been
reported in the literature (Senay et.a|., 1998). The remote sensing images were
used to select sites for experimental measurement of in-field drydown that
represented the greatest spatial variability with respect to hybrid and field level
development. A sample of one of these images with the selected measurement

points is shown in Figure 14.

34

Figure 14: Image if field C5 with selected measurement points.

 

 

 

 

3.? .. Index

§ f; Blue 0-62
3'» I Black 63—111
- DarkPink 112—145
.1: LightPink 146—300
I

r-'

'..._ Reddots=RainGauges

e .- . .

,2i.‘ ' Circles = 30 m sanplmg
=53: radius

{5 '_

ra-

31-.

1? -{‘.

.M’Vr
0's.

- .0 :.I.. h ..:.J ?.‘:*? .:.:- V O .I. . e . .02: . ...'..: '0'. .

    
 

  
  

.1
1 (It-

I , I: - -
EDT-a" ' A

" 1... ‘ s s al'
-I urethral”. l. ..l-‘-'.‘.ll.|

     

    

 

Tassel initiation and silking were observed for each selected field location
and were compared with modeled dates in CERES-Maize. The model’s P1
coefficient, the thermal time from seedling emergence to the end of the juvenile
period, was calibrated for each genotype to conform the modeled phenology for

monitored plant development.

35

Kernel sampling for drydown began on September 1”t and continued
through harvest on approximately 3-day intervals. Intervals of measurement
were decreased or increased depending on timing and duration of precipitation
events. Drydown was monitored by removing several kernels (20 - 30) from ears
of two neighboring plants per dasiteta point and collection event. Husks were
gently pulled back from each ear and the kernels were removed from the center
section of the ear using a knife if needed. The husk was then replaced against
the ear and secured with a rubber band. This method has been used in past
research and has been found to have a negligible effect on grain-filling and water
loss (Tollenaar and Daynard, 1978; NeSmith, 1992). When more than one-half
of the kernels from the central portion of an ear were removed, a second kernel
sample was taken from an adjacent plant in an attempt to suppress plant to plant
variability in drydown patterns. This process was repeated at the stated intervals
through harvest with no more than three plants per site required for data
collection. Samples were frozen immediately after collection for collective drying
at a later time. For each sample, kernels were inspected upon collection for
black layer formation. Black layer began in the first planted field on September
17th as a light brown layer and concluded as complete black layer in the last
planted field on September 213‘, 1999. The P5 coefficients were genotype-
calibrated for CERES-Maize and drydown simulations using black layer
formation timing.

Each sampling area was hand harvested on a per row basis.

Measurement cell size ranged from 900 (30 x 30 ft) - 100 (10 x 10 ft.) square feet

36

with size of cell changing due to time constraints. In addition to yield component
information, plant population data including plant number and spacing was also
recorded per cell. Hand-harvested yield samples were shelled and analyzed for
their yield components including water content, grain weight, and sample
temperature. The crop was mechanically harvested after hand harvest of the
cells using a yield monitor with integral water content sensor. Hand measured
information will be used for model formulation as the spatial information provided
by the harvester water content sensor and yield monitor were less accurate.

The collected kernel samples were weighed separately and collectively
dried at 80 C for 5 days at the season’s end (NeSmith, 1992). The kernels were
then re-weighed and water contents for each sample were calculated. Site-
specific samples were arranged chronologically to represent site-specific
drydown curves.

With data collection complete, site-specific CERES-Maize simulations
were performed using soil texture, measured plant population, and precipitation
data from the field. Soil texture differences were simple field approximations
based on observations and existing county-level soil maps. Precipitation data
sets were either stand-alone precipitation measurements from a single gauge or
a linear interpolation of the two closest gauges depending on distance from the
sampling point to the gauges.

With site-specifically calibrated CERES-Maize simulations, the proposed
drydown model could be tested using the required phenological timing

information from CERES-Maize and the measured, site—specific drydown curves.

'37

Initial formulation and testing of the model utilized spreadsheet analysis. Non-

developmental postmaturity by

linear optimization was used to determine k and k

minimizing the sum of squared differences between measured and modeled
values of moisture content for each site-specific drydown curve. The drydown
curves were collectively analyzed such that single values of constants could be
used to represent the complete data set as well as in- reld drydown processes for

other places and times.

38

Chapter 3
RESULTS AND DISCUSSION

Analysis of results

The systematic analysis of this experiment follows an outline of questions
that initially focuses on the collected data and culminates in reviewing the
proposed model’s accuracy and applicability. Below is a detail of these
examinations:

1. Based on a statistical comparison of measured drydown values, which data
sets would be used as reference curves for model parameter development?

A. Were site-specific data sets affected by any unaccounted influence?

B. What are the possible sources of error for the collected data?

2. Did the proposed model meet the experiment/thesis objective?

A. Did the integral “parent models” (CERES-Maize, grain filling, dewpoint

as a function of minimum daily temperature) perform accurately?

B. Is the model accurate for other data sets?

Initial analysis concentrates on the collected data. Regression analysis of
the measured and modeled drydown curves was performed per genotype and
field. Examples of the complete data set are shown in the following tables and

figures.

39

Figure 15: Modeled and measured drydown curves for DK 535 Hybrid in field C4

(reference Figure 13).

 

 

 

 

 

 

 

Modeled vs. Measured drydown for C4 (Dekalb 535)
90% 0048 Measured
Moisture
80%
IMoisture (%)
70% New Model
2:; 60% C4D Measured
c .
g 50% Morsture
0 o ' C4D Moisture
S 401’ (%) New Model
a 30%
2 *C4F Measured
20% Moisture
10% OC4F Moisture
% New Model
0% ( )
7/19/1999 8/8/1999 8/28/1999 9/17/1999 10/7/1999 10/27/1999
Time (days)

 

 

 

Table 1: Regression analysis for C4 drydown curves plotted in Figure 15.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Plant Population: 6.53 7.18
Regression CoeflF): 0.99 0.98 0.95
C48 C4D C4F

Measured Measured Measured Standard
Date Model Moisture Moisture Moisture MEAN Deviation
7/27/1999 85.0% 85.0% 85.0% 85.0% 85.0% 0.00%
9/1/1999 43.2% 42.3% 39.5% 40.9% 40.9% 1.44%
9/6/1999 37.6% 38.5% 36.2% 35.2% 36.6% 1.71%
9/10/1999 31.7% 33.9% 26.0% 29.9% 5.55%
9/15/1999 26.6% 29.8% 28.1% 33.6% 30.5% 2.79%
9/18/1999 23.1% 29.0% 25.4% 32.9% 29.1% 3.72%
9/22/1999 19.8% 24.6% 21.6% 25.0% 23.7% 1.87%
9/24/1999 18.7% 25.7% 18.4% 27.7% 23.9% 4.94%
10/6/1999 17.6% 17.6% 16.4% 25.1% 19.7% 4.74%
10/15/1999 16.1% 12.3% 15.6% 13.9% 2.30%
10/19/1999 17.1% 19.0% 18.6% 18.8% 0.34%

 

 

40

 

Figure 16: Modeled and measured drydown curves for DK 535 Hybrid in field C5

(reference Figure 13).

 

Modeled vs. Measured Drydown For C5 (DeKalb 535)

 

0 C5F Measured
Moisture

I 05F Moisture (%) New
Model

CSG Measured
Moisture

. CSG Moisture (%) New
Model

1" 05H Measured
Moisture

O 05H Moisture (%) New
Model

+ C5l Measured Moisture

Moisture Content (%)

- C5| Moisture (%) New
Model

CSJ Measured
Moisture
C5J Moisture (%) New
Model

 

 

 

 

10/27/1999

10/7/1999

8/28/1999 9/17/1999

Time (days)

7/19/1999 8/8J1999

 

 

 

Table 2: Regression analysis for C5 drydown curves plotted in Figure 16.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Population (plants’m 2): 5. 77 6.41 6.33 6.46 I
Regression Coefi.(r2): 0.987 0.965 0.991 0.991 0.987
C5F CSG C5H C5! C5J

Measured Measured Measured Measured Measured Standard
Date Model M olsture Moisture Moisture Moisture Moisture MEAN Deviation
7/27/1999 85.0% 85.0% 85.0% 85.0% 85.0% 85.0% 85.0% 0.00%
9/1/1999 43.2% 42.9% 39.7% 41.5% 41.4% 43.5% 41.8% 1.48%
9/6/1999 37.6% 41.1% 31.3% 36.8% 37.4% 36.6% 4.06%
9/10/1999 31.7% 37.7% 32.1% 28.3% 32.1% 32.6% 3.89%
9/15/1999 26.6% 33.2% 29.4% 31.3% 2. 73%
9118/1999 23.1% 27.9% 27.6% 28.8% 24.0% 31.0% 27.9% 2.55%
9/22/1999 19.8% 22.3% 21.4% 19.4% 21.9% 21.3% 1.31%
9/24/1999 18.7% 21.1% 18.7% 20.4% 19.4% 24.7% 20.9% 2.30%
10/6/1999 17.6% 15.8% 16.8% 18.0% 19.4% 17.5% 1.52%
10/15/1999 16.1% 18.8% 18.6% 18.7% 0.14%
10/19/1999 17.1% 18.9% 20.6% 19.8% 1.22%

 

 

41

Figure 17: Modeled and measured drydown curves for DK 493 Hybrid in field C4

(reference Figure 13).

 

Modeled vs. Measured Drydown for C4 (Dekalb 493)

 

0 04A Measured
Moisture

IC4A Model

C40 Measured
Moisture

C4C Model

Moisture Content (%)

x 04E Measured
Moisture

 

OC4E Model

 

 

 

 

8/8/1999 10/7/1999 10/27/1999

8/28/1999 9/17/1999
Time (days)

7/19/1999

 

 

 

Table 3: Regression analysis for C4 drydown curves plotted in Figure 17.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Population (plants’mz): 8.29] 6.21 6.07
Regression Coefl'.(r2): 0.980I 0.987 0.991
C4A C4C 04E
Measured Measured Measured Standard
Date Model Moisture Moisture Moisture MEAN Deviation
7/27/1999 85.0% 85.0% 85.0% 85.0% 85.0% 0.00%
9/1/1999 42.7% 43.6% 45.6% 40.5% 43.2% 2.57%
9/6/1999 36.0% 32.3% 37.8% 32.8% 34.3% 3.07%
9/10/1999 30.3% 31.3% 29.1% 30.2% 1.58%
9/15/1999 25.1% 23.1% 25.8% 24.5% 24.5% 1.37%
9/18/1999 21.8% 21.6% 17.4% 25.6% 21.5% 4.11%
9/24/1999 17.9% 24. 5% 24. 5%
9/25/1999 17.3% 21.4% 20.6% 16.7% 19.6% 2.54%
10/6/1999 17.5% 17.2% 14.8% 14.6% 15.5% 1.46%
10/9/1999 17.1% 19.9% 16.8% 16.8% 17.8% 1.80%

 

42

 

Figure 18: Modeled and measured drydown curves for DK 477 Hybrid in field C1

(reference Figure 13).

 

Modeled vs. Measured Drydown forC1 (Dekalb 477)

 

0 C1A Measured
Moisture

IC1A Model

C1 B Measured
Moisture

' C1B Model

X C1C Measured
Moisture

OC1C Model

Moisture Content (%)

+ C1D Measured
Moisture

-C1D Model

 

 

 

 

7/19/1999 8/8/1999

8/28/1999 9/17/1999
Tlme (days)

10l7/1999 10/27/1999

 

 

 

Table 4: Regression analysis for C1 drydown curves plotted in Figure 18.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Population (plants/m 2 ): 5. 92' I
Regression Coefi.(r2): 0.991 0.994_ 0.988 0.991
C1A 015:: me 015
Measured Measured Measured Measured Standard
Date Model Moisture Moisture Moisture Moisture MEAN Deviation
8/2/1999 85.0% 85.0% 85.0% 85.0% 85.0% 85.0% 0.00%
9/1/1999 45.7% 48.3% 47.4% 51.4% 40.1% 46.8% 4.75%
9/6/1999 41.5% 39. 9% 43. 0% 32.6% 38. 5% 5. 32%
9/10/1999 35.1% 33. 3% 30. 7% 35.8% 26. 5% 31.6% 3.99%
9/15/1999 31.5% 28.6% 28.6%
9/20/1999 25.0% 23.2% 24.8% 20.3% 22.8% 2.27%
9/22/1999 22.9% 22.2% 23.0% 15.7% 20.3% 3.98%
9/25/1999 20.4% 23.4% 17.4% 22.1% 15. 1% 19.5% 3.93%
10/8/1999 17.8% 15.9% 15.1% 14.0% 12.8% 14.5% 1.35%
10/20/1999 17.5% 17.9% 14.7% 17.9% 12.6% 15.8% 2.57%

 

 

 

 

 

 

 

 

43

 

Figure 19: Modeled and measured drydown curves for DK 477 Hybrid in field C3

(reference figure 13).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Modeled vs. Measured Drydown for C3 (Dekalb 477)
9 C3A
0'9 - - - " Measured
L" ." -- " . : : . ; ' ‘ f .7 5 "3 ' 3 '--’ “7 T'2"“".‘:';': :‘13‘ 1:21; ' : ; :?:§:3:?; :5: 5:353 3293.73.1- MOIStUre
0-3 ., I C3A Model
A 0.7 33555
1?: 0 6 C38
5 ' Measured
”E 0.5 Moisture
o
0 5.3:; " C38 Model
93 0.4 ji55§35
3 11:2
.9 3:15-51
g 0.3 $31; X C3C
0 2 ' Measured
' =13?- Moisture
0.1 - 0C3C Model
0 _
7/19/1999 8/8/1999 8/28/1999 9/17/1999 10/7/1999 10/27/1999
Time (days)

 

 

 

Table 5: Regression analysis for C3 drydown curves plotted in Figure 19.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Papulation (plants/m2): 7.26 57' 66'
Regression Coefi.(r2): 0.984 0988' 0.990I
C3A 035- cac
Measured Measured Measured Standard
Date Model Moisture Moisture Moisture MEAN Deviation
8/1/1999 85.0% 85.0% 85.0% 85.0% 85.0% 0.00%
9/1/1999 44. 9% 49.6% 47. 7% 47.4% 48.2% 1.18%
9/6/1999 39.9% 42.2% 40.3% 39.1% 40.6% 1.56%
9/10/1999 33.7% 35.0% 29.0% 32.0% 4.21%
9/15/1999 29.1% 27.4% 28.1% 27. 8% 0.46%
9/16/1999 27.9% 29.6% 29.6%

9/18/1999 25.2% 21 .6% 24.4% 28. 8% 24.9% 3.62%
9/22/1999 21.3% 25.2% 17.0% 19.9% 20.7% 4.18%
9/24/1999 20.0% 19.4% 17.2% 18.3% 1.55%
10/6/1999 17.8% 16.1% 14.9% 14.0% 15.0% 1.06%
10/20/1999 17.4% 17.9% 16.8% 15.0% 16.5% 1.47%

 

 

 

Figure 20: Modeled and measured drydown curves for DK 471 Hybrid in field C6

(reference figure 13).

Modeled vs. Measure Drydown for 06 (Dekab 471)

 

 

 

 

 

 

 

 

0.9
0.8 ..
0'7 " e C6A Measured Moisture
0.6 «- I C6A Model
C6B Measured Moisture

°-5 “ C68 Model
0.4 ,_ o )1: CBC Measured Moisture

I C60 Model
0'3 " + -+ C6D Measured Moisture
02 ,_ x + -ceo Model

XX . v
0.1 «~
7/29/99 8/18/99 9l1l99 10/27l99

Table 6: Regression analysis for C6 drydown curves plotted in Figure 20.

 

 

 

 

 

 

 

 

 

 

 

 

Population (plenum 2): I
Regression Coeff.(r2): 0.992 0.987 0.962 0.963
CGA C68 CGC COD
Measured Measured Measured Measured Standard
Date M eds! Moisture Moisture Moisture Moisture MEAN Deviation

8/3/1999 85. 0% 85. 0% 85.0% 85.0% 85.0% 85.0% 0.00%

9/1/1999 46.6% 41.2% 41.2%

9/6/1999 42. 3% 39.3% 46.3% 37.2% 36.5% 39.8% 4.46%
9/10/1999 36. 7% 35.4% 39.4% 32.9% 29.8% 34.4% 4.08%
9/15/1999 35.4% 30. 5% 34.1% 24.4% 30.1% 29.8% 4.00%
9/20/1999 28.2% 23.6% 29.9% 16.8% 25.4% 23.9% 5.42%
9/22/1999 25. 7% 25.8% 17. 1% 24. 0% 22.3% 4. 58%
9/25/1999 22.8% 20. 3% 26.4% 22.7% 24. 5% 23.5% 2.61%
10/6/1999 18.7% 13.6% 19.2% 17. 1% 24.4% 18.6% 4. 54%

10/20/1999 17.5% 12.8% 14.1% 15.2% 16.0% 14.5% 1.36%

 

 

 

 

 

 

 

 

 

 

45.

 

The results provided two important findings. The first is spatial variability of
drydown rate with respect to position within a field of common genotype exists.
This effect is demonstrated in measurement locations nearest to the outside
edges of the field. These sites produced the greatest difference between
measured and modeled values and were primary sources of error in the
evaluation of model constants. This circumstance was most prevalent at
sampling point C1 D (Figure 18). Cell C1 D was located in the second row (east
to west orientation) from the edge of field that is adjacent to a heavily used road.
A possible reason is that increased air movement and increased availability of
solar radiation due to a lower “effective plant population” caused faster drydown.
The term “effective plant population” for this work refers to the fact that the
outside rows of a corn field are not sheltered by a local array of plants like those
in the center of the field. While intra-row plant spacing may be common, the
severing of a local plant array by removal of rows creates an effective plant
population that is less than the intended population. Evidence of this relationship
between plant population and drydown rate was observed at an interior site
(C5E). The measured plant population for this point was the lowest in the study
while it’s drydown rate was the fastest. Because the objective of this work is to
test a robust model that is applicable at the farm level, the effects of differing
plant population on drydown rates are not incorporated into the model.

A second finding from this result is the exponential decline in water
content from physiological maturity except for precipitation events. This is most

obvious for precipitation events that occurred early in the post-maturity phase of

46

drydown. The best example of this phenomenon is the difference between the
measured water contents on September 22"d and 24'", 1999. On September
23'“, a rainfall event of approximately 10.2 mm lasting approximately 2.5 hours
was recorded. On the 24‘", samples were collected in the late afternoon after
plant surface water had evaporated. Special care was taken not to manually wet
the kernels during the sampling process. As shown in the tables, a hysteresis
effect was not expressed in all data sets for these dates due to site-specific
development differences related to genotype or plantingdate. As mentioned in
the model & methods section, the hysteresis effect has been found to dissipate
due to changes in the number of times the hysteresis “loop” has been activated.
Precipitation influenced the water content of later maturing hybrids more so than
hybrids that matured earlier. This could be attributed to a lack of rainfall in the
initial weeks after the earlier hybrids reached physiological maturity when
changing water content would have been most susceptible to hysteretic effects
due to increased kernel membrane permeability. Precipitation had a great
influence for the later hybrids, particularly DK 471 & 477, as 2 sites did not show
increased water contents near precipitation events. Based on this analysis, the
proposed form of the drydown model’s precipitation term will be retained.

While the model is relatively sensitive to precipitation amount during early
post-maturity simulation, in-field spatial variability of rain events was not
significant relative to measured changes in water content. Further research
regarding the proposed wetting factor via manual re-wetting of corn is required

for a more comprehensive understanding of its value.

47

Using this information, cells that exhibited a water content reduction rate
due to decreased plant population would not be included in model formulation.

These cells include C1 D & C5E (for aforementioned reasons) and CGC (also due
to population of less than 5 plants/m2). Two other cell drydown curves, 05G &

C4F, have also been removed from model formulation due to possible erroneous
data points. Each of these measurement locations were relatively protected
from possible “effective plant population” effects and each had comparable plant
populations with neighboring cells. In each case, a single measurement point
was out of agreement with others near it and therefore that cell was removed
from model consideration. Possible causes for these errors include samples that
were water contaminated during kernel removal, kernel destruction during
removal that may have caused wetting of the sample, and weighing errors made

during the mechanical drying process.

Formulation of the drydown equation

With the remaining data sets, the proposed model structure was tested by
solving for undetermined constants and analyzing differences between model

output and measured data sets. Below is the proposed model.

IF (TTB< P5). MDEVELOPMENTAL= (Me) +(Mo - Me)e-kdeveIOp mental *t

IF (TT3> P5):

MPOST-MATURITY = (Me + Pe-m(tg)) +(Mo - (Me + Pe-m(tg)))e_kpostmaturity *t

48

The modeled grain filling rate as a function of fraction accumulation of the
P5 coefficient in thermal time was a good estimate of the k developmental

proportionality constant. This can be seen in Figures 15 - 20 where the modeled
water content fit well with measured values prior to the onset of physiological
maturity. This value ranged from .025 to .027 for early and late maturing hybrids,
respectively. As the P5 coefficient was re-calibrated for this analysis to reflect
measured dates of black layer development, tests could not be made to
determine if the P5 coefficient is a good predictor of physiological maturity.
However, the observed value of P5 was within one day of daily thermal time of
the manufacturers assessed time from mid-pollination to physiological maturity.
This relationship gives merit to the use of the P5 coefficient as a basis for

developmental timing in the model.

The k post-maturity constant was more difficult to calibrate. As mentioned

earlier, the hysteresis effect decreases with time after maturity and the number
of wetting events for biological materials. For this work, an exponential model of
the precipitation term was used successfully for modeling hysteresis effects.
Analogous to hysteresis is an increased ability of the kernel to dissipate water
when there is significant difference between grain water content and the

equilibrium water content. To simulate this effect, it was necessary to change

the k post-maturity term to reflect the kernel’s increased ability to dry when the

(Mo - Me) was relatively large. k post-maturity was chosen to represent this

drying function as it was not associated with a measurable event such as

49

precipitation. Changing drying model constants (k) to represent time dependent
processes is common throughout the literature (Jayas et.a|., 1991). The
following relationship is used in this work to model the increased rate with which

kernels change in water content after hysteresis:

k ”Mammy = 6/Mo

5 = drying constant
This change in model structure worked well for modeling the drying process for
post-precipitation drydown rate and required not equation. Use of M0 in the
formulation of the post-maturity proportionality constant was also found
successful for a like temperature regime in other research (Misra & Brooker,
1980).

The remaining unsolved constant is the precipitation constant 0). With
accurate phenological timing provided by CERES-Maize and observations, (0 and
6 were solved for using non-linear optimization to minimize the sum of squared
differences between measured and modeled drydown values. The results of this
analysis concluded that the following values, when used in the model, best

simulated corn drydown.

 

 

k 9&4“,th = 5/Mo = (drying constant = .027)/Mo

00 = wetting parameter = .49

 

The model simulated drydown accurately for the measured data sets

(Figures 15 — 20 and Tables 1 - 6). The mean difference between the measured

50

 

and modeled values decreased with respect to time as mean errors of less than
1.5% were common near harvest.

One of the major objectives of this work was accurate application of a
drydown model for different environments. To’test the models applicability to
other data sets, weather and moisture content data collected in Iowa in 1964
were used to test the model. Model accuracy mirrored those found during the

1999 Michigan experiment.

51

Chapter 4

CONCLUSIONS

The objective of this work is to forge a robust, farm-level applicable model
for infield drydown of corn from reviewed literature and experimental results.
Meeting this objective required adhering to the following conditions. First, the
model needs to work relatively independent of assumed water content values
and plant phenology. This constraint was observed with one exception, as a
I value of 85% water content is assumed at the beginning of linear grain filling.

A second constraint called for the model’s required environmental parameters to
be easily measured by farmers. This condition was meet and exceeded as only
daily maximum and minimum temperatures and precipitation are required model
inputs. These variables are common to the CERES-Maize weather file format
and will allow for the drydown model to be easily integrated into CERES-Maize.
A third constraint included incorporation of equilibrium moisture content and
precipitation into the model. This condition is of extreme importance with respect
to understanding the environmental processes that control changes in grain
water content with respect to time. The precipitation term showed relevance to
simulation of measured changes in water content in the performed experiment,
but it’s exact effect at all points in the post-maturity drydown curve requires
further research. The inclusion of an equilibrium moisture content term
increased the model’s ability to simulate the effects of the environment on kernel

water content as it approached atmospheric water content.

52

The result of this work culminates in a robust, highly applicable model of
in-field drydown of corn. Accurate use of the model is dependent on a calibrated
CERES-Maize platform with respect to site and environmental conditions. Both
CERES-Maize and the drydown model are highly reliant on accurate assessment
of genotype characteristics related to plant phenology, in particular the time from
silking to physiological maturity.

Needs for future research regarding in-field drydown should be
concentrated in the following areas:

1. The effect of precipitation with respect to time in the post-maturity
drydown phase. Mechanical wetting could help to better define the
influence of precipitation on a physiologically changing kernel.

2. The effect of plant population corn drydown. “Effective plant
population” for plants near open regions and for areas of the field that
have different plant populations due to problems with planting and
emergence needs to be further addressed.

3. The effect of an early killing frost is abscent from the model. This
season’s first killing frost came on September 22, 1999 and did not
cause pre-mature formation of black layer.

4. Applications in different places and seasons. This research is limited
to only a few data sets. Future data sets are needed to strengthen the
model’s spatial and temporal application vigor.

Continued work regarding in-field drydown may provide answers regarding

the aspects of plant growth that most determine spatial variability in com

53

production. If drydown is quantified as a function of hybrid selection and the
environment, producers and researchers can shift their focus to improving
management practices based on reflected changes in the vegetative period of
development.

Regardless of future research, this work will help farmer’s to maximize
grain quality and yield while minimizing mechanical drying and the risk of yield

loss due to harvesting in less than ideal weather conditions.

54

BIBLIOGRAPHY

55

Bibliography

Abawai, Y. 1996. Minimum daily temperature as a predictor of dewpoint
temperature. ACIAR proceedings 71:240-244.

Badu-Apraku, B., Hunter, R.B., Tollenaar, M. 1983. Effect of temperature during
grain filling on whole plant and grain yield in maize (Z. Mays L.). Can. J. Plant
Sci. 63: 357-363.

Bakker—Arkema, F.W., L.E. Lerew, S.F. DeBoer, and MG. Roth, 1974. “Grain
Dryer Simulation”, Research report 224, Agr. Exp. Sta. Mich. State Univ., East
Lansing, MI.

Baron, V.W., T.B. Daynard. 1984. Factors affecting grain dry-down in early-
maturing European and canadian corn hybrids. Can. J. Plant Sci. 64: 465-474.

Basso, B., Ritchie, J.T., Pierce, F.J., Braga, R.P., Jones, J.W. Integrating crop
models, GIS and remote sensing for yield map interpretation. Am. Soc. Of
Agron. (In Press).

Brooking, LR. 1990. Maize ear moisture during grain-filling, and its relation to
physiological maturity and grain-drying. Field Crops Res., 23:55-68.

Bruns, W.A., R.M. Peart, J.E. Newman. 1975. Simulation of the field dry-down
rate of corn. ASAE Paper No. 754050, American Society of Agricultural
Engineers, 2950 Niles Rd., St. Joseph, MI 49085.

Cavalieri, A.J., Smith, 08. 1985. Grain filling and field drying of a set of maize
hybrids released from 1930 to 1982. Crop Sci. 25:856-860.

Crane, P.L., S.R. Miles, J.E. Newman. 1959. Factors associated with varietal
differences in rate of infield drying in corn. Agron. J. 51:318-320.

Cross, H.Z. 1975. Diallel analysis of duration and rate of grain filling of seven
inbred lines of corn. Crop Sci. 15:532-535.

Daynard, T. B. , and W. G. Duncan, 1969. The black layer and grain maturity in
corn. Crop Sci. 92473-476. ‘

Daynard, T. B. 1972. Relationships among black layer formation and grain
moisture percentage, and heat unit accumulation in corn. Agron. J. 64:716-719.

Daynard, T.B. 1972. Relationships among black layer formation, grain moisture
percentage, and heat unit accumulation in Corn. Agron. J. 64:716-719.

56

Daynard, T.B., Duncan, W.G. 1972. The black layer and grain maturity in corn.
Crop Science 9: 473-476.

Dwyer, L.M., B.L. Ma, L. Evenson, R.l. Hamilton. 1994. Maize physiological traits
related to grain yield and harvest moisture in mid-to short-season environments.
Crop Sci. 34:985-992.

Elmore, R.W., Roeth, F .W. 1999. Corn kernel weight and grain yield stability
during post-maturity drydown. J. Prod. Agric. 12:300-305.

Hallauer, AR, and WA Russell. 1961. Effects of selected weather factors on
grain moisture reduction from silking to physiologic maturity in corn. Agron. J.
53:225-229.

Henderson, SM. 1952. A basic concept of equilibrium moisture content.
Agricultural Engineering 33(1): 29-32.

Henderson, S.M., R.L. Perry. 1955. Agricultural process engineering. John Wiley
& Sons, Inc. New York.

Jayas, D.S., S.Cenkowski, S Pabis, W.E. Muir. 1991 . Review of thin-layer drying
and wetting equations. Drying Technology. 9(3):551-588.

Johnson, D.R., Tanner, J.W. 1972. Calculation of the rate and duration of grain
filling in corn. Crop Sci. 12: 485-486.

Jones, CA, and JR. Kiniry. 1986. CERES-Maize: A simulation model of maize
growth and development. Texas A&M Univ. Press, College Station, TX.

Kang, M.S., Zuber, M.S., Colbert, T.R., Horrocks, RD. 1986. Effects of certain
agronomic traits on and relationship between rates of grain-moisture reduction
and grain fill during the filling period in Maize. Field Crops Research, 14: 339-
347.

Kiesselbach, TA. 1950. Progressive development and seasonal variations of
the corn crop. Nebr. Agric. Exp. Sta. Res. Bul. 166.

Kimball, J.S., S.W. Running, R. Nemani. 1997. An improved model for estimating
surface humidity from daily minimum temperature. Agricultural and Forest
Meteorology. 85(1-2):87-98.

Kiniry, JR. 1985. Responses of maize kernel number to shading stress: timing of
sensitivity in the reproductive stage and characteristics of genotypes differing in
this sensitivity. Ph.D. diss. Texas A&M Univ., College Station (Diss. Abstr. 85-
14258)

57

Kiniry, J.R., J.T. Ritchie. 1984. Correspondence between kernel moisture and
kernel dry matter of Maize. Unpublished.

Kiniry, JR. 1991. Maize Phasic Development. J. Hanks and J.T. Ritchie. 1991.
Modeling plant and soil systems. Agronomy Monograph, no. 31: 55-70.

Koon, J.L., C.A. Flood, R.D. Trumbull, R.N. Brewer. 1986. Predicting average
daily dewpoint temperatures. ASAE Paper No. 86-4020, American Society of
Agricultural Engineers, 2950 Niles Rd., St. Joseph, MI 49085.

Krueger,C.A. and J.M. Bunn, 1985. “Selection of a rewetting model for shelled
corn.” ASAE Paper No. 85-3511, American Society of Agricultural Engineers,
2950 Niles Rd., St. Joseph, MI 49085.

Misra, M.K., D.B. Brooker. 1980. Thin-layer drying and rewetting equations for
shelled yellow corn. Trans. ASAE 23:1254-1260.

NeSmith, D.S., Ritchie, J.T. 1992. Maize (Zea mays L.) response to a severe soil
water-deficit during grain filling. Field Crops Research 29: 23-35.

Ngoddy, P.O., F.W. Bakker—Arkema. 1975. A theory of sorption hysteresis in
biological materials. J. Agric. Engng. Res. 20:109-121.

Overman, A.R., Wilson, D.M., Vidak,W., Allhands,M.N., Perry, MN. 1995 Model
for partitioning of dry matter and nutrients in corn. Journal of Plant Nutrition,
18(5): 959-968.

Priestley, C.H.B., R.J. Taylor. 1972. On the assessment of surface heat flux and
evaporation using large scale parameters. Mon. Weather Rev. 100281-92.

Purdy, J. L. and PL. Crane. 1967. Inheritance of drying rate in “mature” corn
(Zea Maize L.). Crop Sci. 72294-297.

Rench, E.W., R.H. Shaw. 1971. Black layer development in corn. Agron. J.
63:303-305.

Ritchie,S.W., Hanway, J.J., Benson, GO. 1997. How a Corn Plant Develops.
Special Report No. 48: Iowa State University of Science and Technology
Cooperative Extension Service, Ames, Iowa.

Schmidt, J.L., and AR. Hallauer. 1966. Estimating harvest date of corn in the
field. Crop Sci. 6:227-231.

Senay,G.B., Ward, A.D., Lyon, J.G., Fausey,N.R., Nokes,S.E. 1998.

Manipulation of high resolution aircraft remote sensing data for use in site-
specific farming. Transactions of the ASAE Vol.41(2): 489-495.

58

Sprague, 1936. The relation of moisture content and time of harvest to
germination of immature corn. J. Am. Soc. Agron. 28:472-478.

Stapper, M., G.F. Arkin. 1980. CORNF: A dynamic grwoth and development
model for Maize (Zea Maize, L.). Texas Agric. Exp. Stn. Res. Ctr. Program and
Model Documentation. 80 - 2.

Strohman, R.D., R.R. Yoerger. 1967. A new equilibrium moisture content
equation. Transactions of the ASAE 10(5):675-677.

Thompson, T. L., Peart, R. M., Foster, G. H. 1968. Mathematical simulation of corn
drying- A new model. Trans. Of the ASAE 11: 582- 586.

Tollenaar, M., T.B. Daynard. 1978b. Dry weight, soluble sugar content, and
starch content of maize kernels during the early postsilking period. Can. J. Plant.
Sci. 58:199-206.

Troyer, AF, and W. B. Ambrose. 1971. Plant characteristics affecting file drying
rate of ear corn. Crop Sci. 11:529-531.

VanEe, G.R., Kline, G.L. 1979. CORNSIM - A com production model for central
Iowa. ASAE Paper 79-4518.

Vasic, M. 1981. Computation of water surface evaporation for meteorological

stations in Serbia (Yugoslavia) using air temperature data only. Agrohemija 0(3-
4): 155-162.

59

   

”'illlllllijllijlllljllilllllljlll'ES