Chapter 4.

9
WEATHER STATION SITING
Effects on Phenological Models

Richard L. Snyder1, Donatella Spano2, and Pierpaolo Duce3

1
Department of Land, Air, and Water Resources, University of California, Davis, CA, USA;
2
Department of Economics and Woody Plant Ecosystems, University of Sassari, Sassari,
Italy; 3Agroecosystem Monitoring Laboratory, Institute of Biometeorology, National
Research Council, Sassari, Italy

Key words: Temperature, Measurement, Standard surface, Heat units, Sensors

1. INTRODUCTION

The collection of accurate temperature data is critically important for

phenological model development as well as for predictions when a model is
being used. This is true everywhere, but especially in an arid climate where
advection and surface wetting (i.e., by rainfall or irrigation) can affect the
temperature measurements. In fact, the use of bad temperature data can lead
to errors as large as differences resulting from climate change. Therefore,
accurate temperature measurement is critical for both the application and
development of phenological models. Temperature readings are affected by
the local energy balance, so wetting frequency by irrigation or rainfall can
affect readings. Therefore standardization and management of the
underlying surface is critical to obtain useful models that can be universally
applied. In this paper, the history and proper measurement of temperature is
discussed including sensors, shielding, weather station siting, and surface
management.

Schwartz (ed.), PHENOLOGY: AN INTEGRATIVE ENVIRONMENTAL SCIENCE, 345-361

© 2003 Kluwer Academic Publishers. Printed in the Netherlands.
346 Phenology: An Integrative Environmental Science

2. BACKGROUND AND HISTORY OF

TEMPERATURE MEASUREMENT

Although everyone intuitively understands the meaning of temperature, it

is really not easily defined. One could describe it as a measure of the heat
content of the air. However, heat is an even more vague term. According
Horstmeyer (2001), who reported the following statistics, heat does not
really exist. What does exist are unimaginable numbers of air molecules
(about 2.69×1025 molecules/m3 at sea level) that are moving at incredible
speeds (increasing from about 1600 to 1700 km/h for dry air as the
temperature increases from 0°C to 40°C). The mass of individual air
molecules is small and the total volume of the molecules occupies less than
0.1% of the air volume, but there are about 5.95×1014 collisions/s between
molecules. When molecules collide, energy is not created or lost, but kinetic
energy is transferred between the molecules. In thermodynamics,
temperature is considered a measure of the average kinetic energy of the air
molecules expressed as: T M ν 2 R' , where ν 2 is the mean of the squared
molecular velocity, M is the molar mass, and R'' is the molar gas constant.
Therefore, when the speed of the molecules increases, the temperature
increases proportionally to the mean of the squared velocity.
As air molecules strike our skin, some kinetic energy is absorbed and
conducted from molecule to molecule into our bodies. We describe this
“sensation” as “heat” and the term “sensible heat” is used to indicate that it
is energy that we “sense”. If the temperature is higher, the molecules move
faster, more hit our skin, more kinetic energy is received, and the hotter we
feel. When a thermometer is placed in the air, its surface is also bombarded
with air molecules at near sonic speeds. These collisions transfer kinetic
energy to the thermometer and some is conducted inside. When the energy
reaches and heats the liquid in the thermometer, it expands and moves up the
capillary tube indicating a higher temperature.
Historically, temperature measurement was closely linked to the
development of thermometers. The earliest records pertaining to the concept
of temperature were based on the writings of Aristotle in about 300 A.D.
The first evidence of temperature measurement was in the late 1500's in
Europe, with temperature scales appearing in the early 1600’s. In the mid
1600’s, the first scales based on the freezing point of distilled water
appeared (Christopher Wren and Robert Hooke). By the 1700’s,
Fahrenheit’s scale, which defined 32 degrees as the ice point and 96 degrees
as the human body temperature was developed. Fahrenheit did not use the
boiling point in his definition, but the boiling point (212°F) was used to
manufacture thermometers. Celsius used the ice point and the boiling point
to define a temperature scale starting with 0 degrees at the boiling point and
Chapter 4.9: Weather Station Siting 347

100 degrees as the ice point. However, shortly after Celsius proposed his
temperature scale, others began to use degrees centigrade for the scale
defined with 100 gradations between the ice point and boiling point (i.e., the
reverse of the original Celsius scale). Because there were several
temperature scales in usage during the late 1800’s, Callendar recommended
a single practical temperature using the ice and boiling points of water with
standard platinum resistance thermometers for interpolation.
During the 20th century, several conferences on the international
temperature scale were held in 1927, 1948, 1968, and 1990 (i.e., ITS-27,
ITS-48, ITS-68, and ITS-90). The ITS-27 adopted the concepts of Callender,
with the addition of thermocouples being the standard instrument. The ITS-
48 changed the name of the scale from centigrade to Celsius to credit Celsius
for developing the scale and because the word centigrade represents a unit of
angle measure in French. The ITS-48 also defined the absolute temperature
scale as having exactly 100 gradations between ice point and the boiling
point of water. The ITS-68 fixed the Kelvin temperature scale to absolute
zero by setting 1.0 K = 1/273.16 of the triple point (i.e., the equilibrium
between the phases of ice, liquid water, and water vapor). The triple point of
water was set at exactly 273.16 K and the ice point was given the value of
273.15 K = 0°C. The boiling point was still set at 373.15 K = 100°C. They
also eliminated the use of the term degrees for the absolute temperature scale
(i.e., the unit is Kelvin rather than degrees Kelvin). Later, the ITS-90
reported the boiling point was really 99.975°C, so the definition of the
boiling point was eliminated from the absolute temperature scale. However,
at standard temperature and pressure, the Celsius scale still is defined as
going from 0°C at the freezing point to 100°C at the boiling point, and the
Fahrenheit scale goes from 32°F at the freezing point to 212°F at the boiling
point. Today, absolute temperature (K) and temperature in Celsius, where
(0°C = 273.15 K), are commonly used in science.

3. MEASUREMENT THEORY

3.1 Sensors

Primary thermometers measure temperature in the sense that they

measure variables, which are directly dependent on temperature, with
coefficients that are virtually independent of temperature. For example, a
gas thermometer, which measures temperature by the thermodynamic
relationship between gaseous pressure, volume and temperature, is a primary
thermometer. Secondary thermometers (e.g., thermistors, diodes, transistors,
thermocouples, liquid-in-glass, liquid-in-metal, and metal deformation
348 Phenology: An Integrative Environmental Science

thermometers) measure variables that depend on temperature with

coefficients that may be highly dependent on temperature. Secondary
thermometers are the main types used for meteorological standards and
operational meteorological instruments. Because they are the most
commonly used in phenological studies, mercury-in-glass thermometers and
thermistors will be discussed in this chapter.
A mercury-in-glass thermometer is a secondary type thermometer
because the liquid expansion involves coefficients, which are temperature
dependent and not necessarily theoretically predictable. Most meteorological
thermometers are full immersion types, so the entire glass thermometer is
fully exposed to the air. By making the capillary (the bore in which the
liquid rises) small relative to the liquid bulb (reservoir), the sensitivity and
resolution of the thermometer increase. However, a limit is reached when
the capillary surface tension forces degrade performance.

Figure 4.9-1. Maximum and minimum liquid-in-glass thermometers.

Maximum and minimum thermometers are two main types of liquid-in-

glass thermometers that operate on different principles (Figure 1). The
maximum thermometer has a constriction in the capillary bore at a short
Chapter 4.9: Weather Station Siting 349

distance from the fluid reservoir. The liquid expands past the constriction,
and expands to a maximum temperature. On cooling, the liquid above the
constriction contracts leaving a vacuum barrier, while the liquid below the
constriction contracts into the reservoir. This leaves the upper end of the
liquid at the maximum temperature. The minimum thermometer has a small
rider, which is pushed by the surface forces of the meniscus of the liquid
within the capillary bore. The upper end of the rider is left at a minimum
temperature position when the liquid rises again.
Thermistors are made of sintered semiconductor materials (e.g.,
manganese, nickel, copper, iron, cobalt, and uranium oxides) that are pressed
into the thermistor form and aged to promote stability. The resistance to
electrical current decreases as the temperature rises and simple electronics
are used to monitor the resistance. The physics involved in resistance sensor
operation is described in Quinn (1983). With a thermistor, a curvilinear
relationship between resistance and temperature is needed to calibrate the
sensors. Older thermistors were somewhat unstable and, during their
lifetime, the resistance would drift requiring frequent calibration. However,
now thermistors are aged before calibration to minimize drift. Because the
bead thermistors are manufactured with their leads inside the powdered
material before they are pressed, whereas disk thermistors are manufactured
prior to the leads being sprayed or printed on the material, thermistors made
in the shape of beads are more stable than thermistors pressed into disk
shapes. A major advantage of thermistors is that they are small sensors.
The wire/metal resistors may be made very thin, but they still must have an
appreciable length (several millimeters at the minimum) to have a practical
level of resistance (tens of ohms) for temperature measurement. Thermistors
can be made as small as 0.03 mm.

3.2 Shielding

Although having accurate sensors solves part of the problem with

temperature measurement, proper shielding and ventilation are needed to
minimize radiation effects on the sensor energy budget. The three main
energy forms that affect energy balance on a thermometer are radiant,
convective/sensible heat transfer, and thermal conduction. To avoid energy
transfer from the mount to the sensor, the contact area should be small and
insulated with plastic, ceramic, or cardboard materials that reduce
conduction. For electronic sensors, the leads should be small, non-
conductive, and maintained in the same shelter environment as the sensor to
avoid direct exposure to radiation or other heat sources.
Temperature sensors are frequently shielded from short wave radiation
by putting them in a Stevenson Screen (Figure 2), which is made of wood
350 Phenology: An Integrative Environmental Science

and painted white to reflect away most solar radiation. For electronic
sensors, white plastic shields like the Gill shield (Gill 1983) are often used
(Figure 2). Long wave radiation is unavoidable, because it is emitted by all
surfaces at terrestrial temperatures.

Figure 4.9-2. A standard Stevenson Screen, image by Jack Kelly Clark, © Regents of the
University of California, used with permission (left), and a Gill radiation shield (right).

While the goal is to measure air temperature with a thermometer, the

actual temperature recorded is a weighted mean of the air temperature and
the housing or shield temperature around the sensor. Assuming that short
wave radiation reaching the thermometer is negligible, Monteith and
Unsworth (1990) expressed thermometer temperature (T Tt) in terms of the
shield temperature (T Ts), air temperature (T),
T and resistances to sensible (rH)
and radiation (rrR) heat transfer as:

rH Ts rRT (1)
Tt =
rR + rH

by assuming the net radiation and convective heat transfer are equal ((Rn = C)
C
as illustrated in Figure 3. When estimating air temperature, the goal is to
have Tt approach T by making rH much smaller than rR and/or by making Ts
very close to T T. To make rH much smaller than rR, one can maximize
convective heat transfer by making the sensor very small or by ventilating
the sensor. Generally, ventilation of a few meters per second is adequate.
When ventilating, the air should be pulled over the sensor to avoid heating
from the fans. In naturally ventilated Stevenson Screens or Gill shields,
Chapter 4.9: Weather Station Siting 351

errors can result when wind speed is low. Using double walls and roof and
painting the shelter white to reflect solar radiation makes Ts closer to T.
T

Figure 4.9-3. Drawing of the energy balance between a thermometer and its surrounding
radiation shield where Rn is the net radiation, C is convective heat transfer, σ is the Stefan-
Boltzmann constant, Ts, Tt, and T are the shield, thermometer, and air temperatures (K), ρ is
the air density, Cp is the specific heat of air at constant pressure, and rR and rH are resistances
to radiative and sensible heat transfer.

3.3 Height

When using temperature data in phenological models the height of the

temperature measurement can have a big effect on the results. In the USA,
the National Weather Service typically reports 10.0 m temperatures, whereas
many climate and agricultural weather networks report temperatures for
heights varying between 1.5 and 3.0 m. To illustrate the problem, Figure 4
shows the vertical temperature profile changes during the day before and
during the night of a spring freeze event in a northern California mountain
valley. Clearly, big differences exist between data taken at 0.5 m and 10.0 m
heights. Therefore, when developing phenological models, temperature data
should be used from a station with a temperature measurement height similar
to the sensor heights of stations where it is likely to be used. Similarly,
models from the literature should report the temperature measurement height
and the model should only be used with data collected at the same height.

4. HEAT (THERMAL) UNITS

For many applications (e.g., pest management, crop modeling, and

irrigation scheduling) it is useful to predict when a crop or pest will develop
to a particular phenological stage. When temperatures are higher, organisms
352 Phenology: An Integrative Environmental Science

develop in fewer days because they are exposed to the greater heat
accumulation than organisms grown under cooler conditions. The
accumulation of heat is called “physiological time” and “heat units” are a
measure of physiological time, which typically are better than “calendar

Figure 4.9-4. Temperature profile data from a walnut orchard at 0.5, 1.0, 2.0, 6.0, and 10.0 m
height prior to and during a radiation freeze night in a northern California valley.

Figure 4.9-5. Days and cumulative degree-days to mature cotton bollworm from neonate
larvae to adult stages for different temperature ranges (after Wilson and Barnett 1983).
Chapter 4.9: Weather Station Siting 353

time” for predicting days between phenological stages. For example, the
cotton bollworm development from neonate larvae to adult (Figure 5) is
closely related to the number of accumulated heat units regardless of the
large differences in number of days between the stages due to temperature
differences during growth. The cumulative number of degree-days (oD) to
develop is about the same, whereas the number of days to develop is greatly
different for the different growth temperatures.
Most organisms show little growth or development below a lower
threshold temperature (T TL) and often there is little increase in development
rate above an upper threshold (T TU), but the number of hours between TL and
TU provide a measure of heat units for estimating the days between
phenological stages. Each hour of time that the organism is exposed to
temperature between the two thresholds is called a “degree-hour”. Dividing
the accumulated degree-hours by 24 converts the heat units to degree-days,
which are commonly used in phenological models.
There are numerous papers on methods to estimate degree-days from
daily maximum and minimum temperature data and on how to determine
threshold temperatures and the cumulative degree-day requirement for
various crops and pests (Baskerville and Emin 1969; Allen 1976; Johnson
and Fitzpatrick 1977; Parton and Logan 1981; Kline et al. 1982; Snyder
1985; Wann et al. 1985; Reicosky 1989; de Gaetano and Knapp 1993;
Kramer 1994; Yin et al. 1995; Pellizzaro et al. 1996; Roltsch et al. 1999;
Snyder et al. 1999, 2001; Cesaraccio et al. 2001). In general, researchers
have attempted to model diurnal temperature changes using mathematical
models with a range of complexity (Zalom et al. 1983; Snyder et al. 1999).
Calculating the number of degree-days between the two phenological
stages of interest using selected lower and upper threshold temperatures and
all appropriate data sets best identifies threshold temperatures. Then the
mean cumulative degree-days (oDm) over all of the data sets are computed.
The same thresholds are used to calculate degree-days for each data set until
the cumulative degree-days adds up to oDm. Then the mean absolute
difference or root mean square error between the predicted and observed
number of days over all of the data sets is calculated and recorded. This
process is repeated for a range of threshold temperatures until the smallest
mean absolute difference or root mean square error is identified. The
threshold temperatures with the smallest difference between predicted and
observed days between stages and the corresponding oDm provide the best
prediction model.
354 Phenology: An Integrative Environmental Science

5. PLANT VERSUS AIR TEMPERATURE

Whereas phenological development is driven by plant temperature, most

phenological models use air temperature as the main input. Ideally, plant
rather than air temperature should be used in phenological models. Because
the plant temperature depends on the surface energy balance, there is no
simple method to estimate plant surface temperature from air temperature.
Based on the energy balance over a wet surface, the following equation
(after Monteith and Unsworth 1990) estimates the surface temperature (T
To):

es − e r γ *( n ) (2)
To = T + − a
Δ + γ * ρC p Δ + γ *

where T is the air temperature (°C), es is the saturation vapor pressure (kPa)
at T,
T e is the vapor pressure (kPa), ra is the aerodynamic resistance (s/m), Rn
is the net radiation (W/m2), G is the soil heat flux density (W/m2), ρ is the air
density (g/m3), Cp is the specific heat of air at constant pressure (J/g K), Δ is
the slope of the saturation vapor pressure at air temperature (kPa/°C), γ is the
psychrometric constant (kPa/°C), and γ* = γ (rs/ra), where rs is the surface
resistance (s/m). The adiabatic part (i.e., the middle term on the right) is
directly proportional to the vapor pressure deficit (es – e), which is
approximately an exponential function of the air temperature. The diabatic
term (i.e., the right-hand term) depends on the external energy supply
(mainly net radiation). Therefore, a wet surface temperature will be warmer
than air temperature when the vapor pressure deficit is high enough that the
adiabatic term is bigger than the diabatic term.

Clearly, Equation 2 is complex, involving parameters that are not readily

available for phenological models, and it only applies to relatively smooth,
wet surfaces (e.g., for unstressed plant canopies). However, considerable
past research with infrared thermometers has shown that the temperature of
unstressed canopies is predictable using the air temperature and vapor
pressure deficit (Idso et al. 1981; Jackson et al. 1981).

6. SITING EFFECTS

6.1 Underlying Surface

The main indirect source of energy for heating the air is the sum of direct
and diffuse short-wave (solar) radiation minus that which is reflected away
Chapter 4.9: Weather Station Siting 355

from the surface. Although smaller in magnitude, a more direct source is

long-wave radiation from the surface upwards and from the sky downwards.
By convention, downward radiation that adds energy to the surface is
positive, and radiation away from the surface is negative. During a clear,
mid-summer day at noon between about 30° and 50° latitude, the downward
short-wave radiation is on the order of 1000 W/m2 and, from a grass surface,
about 25% of the energy received is reflected away leaving a net short-wave
balance of about 750 W/m2. Under clear skies at the same latitude, the long-
wave radiation is about -450 W/m2 upwards and about 350 W/m2
downwards leaving a net long-wave radiation balance of about -100 W/m2.
Combining the short- and long-wave balances, the mid-latitude net radiation
near noon on a clear summer day is about Rn = 650 W/m2 over a grass
surface. Under clear skies, the net long-wave radiation changes little with
time, but, under overcast skies, because the clouds are warmer than clear
sky, it decreases to near -10 W/m2. Short-wave radiation changes
considerably over the day with the angle of the sun above the horizon and it
mostly decreases with cloud cover.
Energy from positive radiation does one of the following:
1. heats the soil surface and conducts downward into the soil,
2. vaporizes water that contributes to upward latent heat flux,
3. heats air near the surface, which then convects upwards as sensible heat
flux, and contributes to
4. miscellaneous consumption for heating the plants and photosynthesis.
However, energy flux also reverses direction and does one of the
following:
1. conducts heat upward from the soil to the surface,
2. condenses water vapor and convert the energy to sensible heat,
3. transfers sensible heat to a colder soil surface, and
4. miscellaneous release losses due to cooling plants and respiration.
Using the convention that positive Rn at the surface is partitioned into
energy that heats the soil (G), vaporizes water (λE), heats the air (H), or
contributes to miscellaneous energy consumption ((M M) used for heating the
plants and photosynthesis, the energy balance equation is written as:

Rn = G + λE + H + M (3)

Usually, M is relatively small and is ignored for energy flux calculations.

If the soil temperature decreases with depth below the surface, then G is
positive. Similarly, if the air temperature decreases with height above the
surface then H is positive. If the temperature profiles are reversed then G or
H are negative and heat is transferred to the surface. If more water vapor is
vaporized than condensed on the surface, then the water vapor flux is
356 Phenology: An Integrative Environmental Science

upwards and λE is positive. If more water condenses than vaporizes, then

λE is negative.
Temperature is a measure of the sensible heat content of air at the height
of the sensor. Because the temperature gradient with height is typically
much greater than with horizontal distance, vertical fluxes of sensible heat
are generally much more important than horizontal fluxes. Therefore,
energy balance at the surface, which largely determines the sensible heat
content at the sensor level, is extremely important. If the surface is warmer
than the air above, a positive H will likely increase the sensible heat content
and hence the temperature of air at the sensor level. If the surface is colder
than the air above, a negative H will tend to lower the sensible heat content
and temperature at the sensor height.
Assuming no horizontal advection and the same incoming long-wave and
solar irradiance, the outgoing long-wave and, hence, net radiation depend
mainly on the surface albedo and temperature, which is affected by G, H, H
and λE. Albedo is affected by surface properties and the angle of incidence
of the solar radiation. The relative partitioning of absorbed incoming
H, and λE determines the surface temperature. Soil heating is
radiation to G, H
affected by thermal conductivity and soil heat capacity. Latent heat flux is
mainly affected by the presence of water to vaporize and secondarily by the
water vapor content of the air and turbulence, which transfers sensible and
latent heat to and from the surface. Sensible heat flux is the residual energy
after the Rn contributions to heating the soil (G) and evaporating water (λE)
are removed. Sensible heat content of air near the surface determines the air
temperature. For a given incoming flux density of radiation, the presence or
absence of water is the main factor affecting changes in sensible heat content
of air near the surface and, hence, temperature. For example, recording
temperature above a transpiring grass surface will result in lower
temperature than measurements above a dry, bare ground surface. And the
difference between temperature recordings over the two surfaces is greater
as the climate becomes more arid.
In general, most climate and weather forecast stations in the USA are
located over plots of land with the natural vegetation of the region. Most
agricultural weather stations are located over grass or irrigated grass surfaces
to standardize the site for comparisons and to allow for model development
that can be transferred to other locations. In arid climates, variations in the
energy balance due to intermittent rainfall, can greatly affect temperature
readings. For example, Snyder et al. (2001) compared annual degree-day
calculations, with a 10°C lower threshold, from stations located over
irrigated grass and over bare soil in four regions of California and found that
the cumulative degree-days were between 3.2 and 10.7% higher for the un-
irrigated, bare soil. The difference is likely due to differences in energy
Chapter 4.9: Weather Station Siting 357

balance over bare soil that is similar to the grass during the rainy season but
can be quite different when the soil dries and the grass continues to transpire.
Also, it is likely that differences in vegetation around the weather stations
contributed to differences in the degree-day accumulation.
The type of underlying surface around a weather station should be
selected depending on the purpose of the data. If the purpose is to
characterize phenological development of natural vegetation, the stations
should be sited in a natural setting that represents the conditions of the
vegetation being studied. However, if the data are used to model phenology
of irrigated crops, it is best to site the stations over an irrigated grass surface
that removes variability due to intermittent rainfall.

6.2 Fetch

Fetch is the upwind distance of the same vegetation underlying a weather

station. Inadequate fetch is a problem for some micrometeorological
measurements and it can lead to errors in temperature measurements if the
purpose is to collect data over a standard surface (e.g., irrigated grass).
Under extremely arid conditions, the authors have observed systematic over-
estimation of temperature by about 4% from a station downwind from a
desert when compared to a station over irrigated grass 177 m from the edge
of the desert (Figure 6). While this amounts to only about 1.6°C higher
temperature at 40°C, the readings were systematic and they led to
considerable differences in degree-day accumulations (Snyder et al. 2001).

6.3 Surrounding Environment

Even when temperatures are recorded over the same standard surface,
sometimes big differences are recorded due to the surrounding environment.
For example, Figure 7 shows the corresponding temperatures recorded at
Torrey Pines and Miramar Naval Air Station near San Diego, California.
Torrey Pines is on the coast and Miramar is located about 30 km inland.
Clearly, although these sites are not far apart, there were big differences in
the temperature data with the coastal site having warmer temperatures when
values are low and the inland site having warmer temperatures when values
are high. If phenological models were to be applied in this region, many
stations would be needed to account the temperature effects of the ocean and
how they change with distance from the coast.
358 Phenology: An Integrative Environmental Science

Figure 4.9-6. Corresponding hourly temperature data (May-June 2000) measured on the west
edge of a grass field and 177 m east of the west edge of a large grass field surrounded by
desert near Indio, California. The prevailing wind was from the west.

Figure 4.9-7. Corresponding hourly temperature data (June-November 2000) from Torrey
Pines on the coast and Miramar Naval Air Station, which is 30 km inland.

In areas that are far from the ocean, there are often big differences in
temperature over similar underlying surfaces. For example, Figure 8 shows
the corresponding temperatures measured at the Indio CIMIS station and the
Vintage Country Club, which are only about 30 km apart in the below sea
Chapter 4.9: Weather Station Siting 359

level desert in southern California. In this example, the regression statistics

indicate that the temperature at the Vintage Country Club is about 3% less
than at the Indio CIMIS station, but that is because the regression was forced
through the origin and the regression line is strongly influenced by the high
temperatures during midday, which are similar. However, at lower
temperatures, during the morning and afternoon, the temperatures at the
Vintage Country Club were considerably more variable and lower. The
difference is mainly due to considerably less wind and mountains that shade
the station in the afternoon at the Vintage Country Club. Because the
diurnal temperature curves are different, the same temperature curve model
cannot be used to estimate degree-days at both sites. Therefore, the use of
temperature driven phenological models is problematic in an area with
multiple microclimate zones.

Figure 4.9-8. Corresponding hourly temperature data (June - November 2000) measured at
the Vintage Country Club near Indio, California and the Indio CIMIS station. The wind speed
at the Vintage Country Club averaged about 8% of the Indio CIMIS site.

7. CONCLUSIONS

Temperature is the driving factor in most phenological models, and

proper measurement is critical for both development and use of the models.
In addition to selecting accurate sensors, they should be mounted at an
appropriate height and properly shielded from short-wave radiation (double
shielding is best). Choosing small sensors that respond rapidly, protecting
electronic leads, and ventilation (in areas with little wind) can improve
360 Phenology: An Integrative Environmental Science

accuracy of the temperature measurements. Data should be collected at a

height that is typical of other weather stations in the area where the model
will be used. Generally, agricultural weather stations collect temperature
data at 1.5 to 2.0 m height and weather services tend to measure at 10.0 m
height. For phenological models of natural vegetation, it is best to site the
weather station in a similar environment without irrigation. However, when
the models are used for irrigated crops, the stations should be sited over an
irrigated grass surface to avoid temperature fluctuations due to intermittent
rainfall at the measurement site. Strong temperature gradients can occur
near large water bodies (e.g., the ocean or large lakes) and in hilly or
mountainous regions where sunlight is blocked during part of the day. In
such regions, more weather stations are needed to better characterize
microclimate differences. However, even when the temperature data are
accurately determined, inaccuracies in model predictions can occur because
it is plant temperature rather than air temperature that truly drives the
phenological development. Although little or no literature on the topic
exists, perhaps using vapor pressure deficits to estimate plant from air
temperature could improve models and make them more universally
applicable.

REFERENCES CITED
Allen, J. C., A modified sine wave method for calculating degree days, Environ. Ent., 5, 388-
396, 1976
Baskerville, G. L., and P. Emin, Rapid estimation of heat accumulation from maximum and
minimum temperatures, Ecology, 50, 514-517, 1969.
Cesaraccio, C., D. Spano, P. Duce, and R. L. Snyder, An improved model for degree-days
from temperature data, Int. J. Biometeorol., 45, 161-169, 2001.
de Gaetano, A., and W. W. Knapp, Standardization of weekly growing degree day
accumulations based on differences in temperature observation and method, Agric. For.
Meteorol., 66, 1-19, 1993.
Gill, G. C., Comparison testing of selected naturally ventilated solar radiation shields, Final
Report Contract # NA-82-OA-A-266, NOAA, St. Louis, 15 pp., 1983.
Horstmeyer, S., Building blocks – What goes on in a cubic meter of air?, Weatherwise, 54(5),
20-27, 2001.
Idso, S. B., R. D. Jackson, P. J. Pinter, R. J. Reginato, and J. L. Hatfield, Normalizing the
stress-degree-day parameter for environmental variability, Agric. Meteorol., 24, 45-55,
1981.
Jackson, R. D., S. B. Idso, R. J. Reginato, and P. J. Pinter, Canopy temperature as a crop
water stress indicator, Water Resour. Res., 13, 651-656, 1981.
Johnson, M. E., and E. A. Fitzpatrick, A comparison of methods of estimating a mean diurnal
temperature curve during the daylight hours, Arch. Meteorol. Geophys. Bioklimatol.,
Series B, 25, 251-263, 1977.
Chapter 4.9: Weather Station Siting 361

Kline, D. E., J. F. Reid, and F. E. Woeste, Computer simulation of hourly dry-bulb

temperatures, Virginia Agricultural Experiment Station, No. 82-5, Virginia Politechnical
Institute and State University, Blacksburg, 18 pp., 1982
Kramer, K., Selecting a model to predict the onset of growth of Fagus sylvatica, J. Appl.
Ecol., 31, 172-181, 1994.
Monteith, J. L., and M. H. Unsworth, Principles of Environmental Physics, 2nd ed., Edward
Arnold, London, 291 pp., 1990.
Parton, W. J., and J. A. Logan, A model for diurnal variation in soil and air temperature,
Agric. For. Meteorol., 23, 205-216, 1981.
Pellizzaro, G., D. Spano, A. Canu, and C. Cesaraccio, Calcolo dei gradi–giorno per la
previsione delle fasi fenologiche nell’actinidia, Italus Hortus, 5, 24-30, 1996.
Reicosky, L. J., J. M. Winkelman, J. M. Baker, J. M., and D. G. Baker, Accuracy of hourly air
temperatures calculated from daily minima and maxima, Agric. For. Meteorol., 46, 193-
209, 1989.
Roltsch, J.W., F.G. Zalom, A.J. Strawn, J.F. Strand, and M.J. Pitcairn, Evaluation of several
degree day estimation methods in California climates, Int. J. Biometeorol., 42, 169-176,
1999.
Quinn, T. J., Temperature, Academic Press, New York, 495 pp., 1983.
Snyder, R. L., Hand calculating degree-days, Agric. For. Meteorol., 35, 353-358, 1985.
Snyder, R. L., D. Spano, C. Cesaraccio, and P. Duce, Determining degree-day threshold from
field observations, Int. J. Biometeorol., 42, 177-182, 1999.
Snyder, R. L., D. Spano, P. Duce, and C. Cesaraccio, Temperature data for phenological
models, Int. J. Biometeorol., 45, 178-183, 2001.
Wann, M., D. Yen, and H. J. Gold, Evaluation and calibration of three models for daily cycle
of air temperature, Agric. For. Meteorol., 34, 121-128, 1985.
Wilson, L. T., and W. W. Barnett, Degree-days: an aid in crop and pest management, Calif.
Agric., 37, 4-7, 1983.
Yin, X., M. J. Kropff, G. McLaren, and R. M. Visperas, A nonlinear model for crop
development as a function of temperature, Agric. For. Meteorol., 77, 1-16, 1995.
Zalom, F. G., P. B. Goodell, W. W. Wilson, and W. J. Bentley, Degree-Days: The calculation
and the use of heat units in pest management, Leaflet 21373, Division of Agriculture and
Natural Resources, University of California, Davis, 10 pp., 1983.