Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/255710905

Benefits From Tying Satellite-Based Energy Balance To Reference


Evapotranspiration

Conference Paper · November 2005


DOI: 10.1063/1.2349336

CITATIONS READS
9 157

3 authors, including:

Richard G Allen Ricardo Trezza


University of Idaho University of Idaho
240 PUBLICATIONS   22,025 CITATIONS    69 PUBLICATIONS   3,411 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Advanced METRIC processing View project

Application of MODIS Land Products to Assessment of Land Degradation of Alpine Rangeland in Northern India with Limited Ground-Based Information View project

All content following this page was uploaded by Ricardo Trezza on 28 May 2014.

The user has requested enhancement of the downloaded file.


Benefits From Tying SatelliteBased Energy Balance To Reference
Evapotranspiration
R. G. Allen, M. Tasumi, and R. Trezza

Citation: AIP Conf. Proc. 852, 127 (2006); doi: 10.1063/1.2349336


View online: http://dx.doi.org/10.1063/1.2349336
View Table of Contents: http://proceedings.aip.org/dbt/dbt.jsp?KEY=APCPCS&Volume=852&Issue=1
Published by the AIP Publishing LLC.

Additional information on AIP Conf. Proc.


Journal Homepage: http://proceedings.aip.org/
Journal Information: http://proceedings.aip.org/about/about_the_proceedings
Top downloads: http://proceedings.aip.org/dbt/most_downloaded.jsp?KEY=APCPCS
Information for Authors: http://proceedings.aip.org/authors/information_for_authors

Downloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permissions
Benefits From Tying Satellite-Based Energy
Balance To Reference Evapotranspiration
R.G. Allen, M.Tasumi, R.Trezza

University of Idaho, 3793 N. 3600 E., Kimberly, Idaho USA 83341

Abstract. Satellite-based surface energy balance is very useful for determining actual
evapotranspiration (ET) over large areas. One uncertainty, however, is in regard to the amount
of error or bias introduced to the surface energy balance and ET by various calculation
components. These include biases in atmospheric correction, albedo calculation, net radiation
calculation, thermal band and surface temperature, air temperature gradient used in sensible heat
flux calculation, soil heat flux function, and extrapolation to 24-hour and longer periods. At the
University of Idaho, we use calculated hourly reference evapotranspiration (ETr) to calibrate the
surface energy balance during processing and to remove these biases. In addition to its use to
calibrate the energy balance, ETr is used to extrapolate ET images to 24-hour and longer
periods. The use of ETr provides equivalency and congruency with ET as estimated using the
traditional crop coefficient x ETr approach.

Keywords: evapotranspiration remote-sensing Landsat reference-evapotranspiration


PACS: 92.40.Je 45.20.dh 92.60.Vb 7.07.Df 44.25.+f, 44.27.+g

INTRODUCTION
The University of Idaho uses the METRICTM (Mapping Evapotranspiration at high
Resolution and with Internalized Calibration) image processing procedure to
determine ET over large land areas. METRIC calculates ET as a residual in the
surface energy balance (EB) and is a variant of SEBAL, an important EB process
model developed by Bastiaanssen [4, 5, 7]. METRIC was extended to provide tight
integration with ground-based reference evapotranspiration (ETr) [1, 13, 14].
METRIC is auto-calibrated for each image using ETr based on hourly weather data
where accuracy of the ETr estimate has been established by lysimetric studies in which
we have high confidence. The calibration corrects the EB for bias and error in:
• albedo calculation, including atmospheric correction
• net radiation calculation
• surface temperature
• air temperature gradientATTACHMENT
function used in sensible heat flux calculation
• soil heat flux function
CREDIT LINE (BELOW) TO BE INSERTED ON THE FIRST PAGE OF EACH PAPER
• aerodynamic resistance calculation including buoyancy impacts
• extrapolation of instantaneous ET to 24-hour periods
• to some degree the bias in wind speed field
CP852, Earth Observation for Vegetation Monitoring and Water Management,
edited by G. D’Urso, M. A. Osann Jochum, and J. Moreno
© 2006 American Institute of Physics 0-7354-0346-5/06/$23.00
127

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
These biases plague essentially all surface EB computations that utilize satellite
imagery as the primary spatial resource. The transport of vapor and sensible heat from
land surfaces is strongly impacted by aerodynamic processes including wind-induced
turbulence and buoyancy, which are essentially invisible to satellites. Therefore,
assumptions must be made during computation of spatial distribution of sensible heat
that introduce unknown biases manifested in the ET calculation.

BACKGROUND ON ENERGY BALANCE


Although satellites routinely measure surface reflectance and some measure surface
temperature, none measure near surface vapor content. Therefore, ET is frequently
determined from satellite imagery by applying an EB at the surface, where energy
consumed by the ET process is calculated as a residual of the surface EB equation:
LE = R n − G − H (1)
where LE is the latent energy consumed by ET, Rn is net radiation, G is sensible heat
flux conducted into the ground, and H is sensible heat flux to the air. A disadvantage
of Eq. (1) is that LE is only as accurate as the estimates for Rn, G, and H.

Net Radiation
Rn is generally computed by subtracting all outgoing radiant fluxes from all
incoming radiant fluxes and includes solar and thermal radiation:
Rn = RS↓ - α RS↓ + RL↓ - RL↑ - (1-εo)RL↓ (2)
where RS↓ is incoming shortwave radiation, α is surface albedo, RL↓ is incoming
longwave radiation, RL↑ is outgoing longwave radiation, and εo is broad-band surface
thermal emissivity. Surface albedo is calculated by integrating reflectivities from
representative satellite bands, for example, bands 1 – 5 and 7 of Landsat, bands 1 – 7
of MODIS and bands 1- 9 of ASTER. At-satellite reflectance is calculated from at-
satellite directional radiance and is corrected to at-surface reflectance (ρs,b) to account
for atmospheric attenuation. The broadband α is calculated by integrating band
reflectances within the short-wave spectrum using a weighting function:
n
α= ∑ [ρs, b ⋅ w b ] (3)
b =1

where wb is a weighting coefficient representing the fraction of at-surface solar


radiation occurring within the spectral range represented by a specific band and n is
the number of bands integrated. Tasumi et al., [14] derived a method for simple, yet
accurate atmospheric correction and for integrating α. Surface temperature (Ts) is
generally computed for Landsat images using a modified Plank equation after some
form of simple atmospheric and surface emissivity correction such as by [16]. Ts can
is generally estimated from MODIS and ASTER using split window techniques based
on availability of multiple thermal bands or by using MODIS science products.

128

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
Soil Heat Flux
Many satellite-based EB applications compute G from Rn. A method common to
SEBAL and METRIC is an empirical equation by [6] representing midday:
G
Rn
= (Ts − 273) (0.0038 + 0.0074α ) 1 − 0.98 NDVI 4 ( ) (4)

where Ts is in K and NDVI is a vegetation index.

Sensible Heat Flux


The classical equation for H, termed Ho, uses a temperature gradient established
between the surface and an air temperature (Ta) at some distance above the surface:
T − Ta
H o = ρ air C p o (5)
rah o

where ρair is air density, Cp is specific heat of air at constant pressure, To is


aerodynamic temperature of the heat source, and rah o is aerodynamic resistance
between the surface and some height z above the surface. Eq. (5) is hindered by
difficulty in estimating aerodynamic surface temperature (To) due to uncertainty in
atmospheric attenuation, contamination and radiometric calibration of the sensor and
because radiometric temperature Ts can deviate from To that drives the heat transfer
process by several degrees [9, 10]. In addition, spatial variation in Ta is unknown.
In SEBAL and as copied by METRIC, H is estimated between two heights, both
above the surface, to obviate problems associated with Ho estimation:
dT
H = ρ air C p (6)
rah 1− 2

where rah 1-2 is aerodynamic resistance between two near surface heights, z1 and z2
(commonly 0.1 and 2 m above zero plane displacement, d) and dT is air temperature
difference between the two heights. dT “floats” above the surface, beyond the height
for sensible heat roughness (zoh) and d, and therefore eliminates the need for zoh. dT
is approximated as a linear function of Ts as pioneered by Bastiaanssen [4, 5]:
dT = a + b Ts datum (7)
where a and b are empirically determined constants for a given satellite image and Ts
datum is surface temperature lapse adjusted to a common elevation datum. [4]
provided empirical evidence for using the linear relation between dT and Ts and [2]
provides some theoretical basis. METRIC differs from traditional applications of
SEBAL principally in how the H function is calibrated.

129

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
REDUCING BIASES IN ENERGY BALANCE COMPONENTS
Table 1 lists major biases that plague estimation of satellite-based EB and estimated
minimum biases in ET caused by the component biases. Values are estimates and
subject to change. Some of the biases are overcome through the design of SEBAL and
use of the floating dT function. The following section provides a description of how
SEBAL and METRIC employ internal calibration within the surface energy balance
using anchor pixels and how METRIC uses ETr to eliminate or reduce biases. The
design of SEBAL reduces biases 1-6, with substantial reduction to bias 4. Remaining
biases are largely overcome in METRIC by use of ETr during the internal calibration.

Calibration of the sensible heat equation


The use of the floating dT function in SEBAL and its internal calibration are
effective in reducing biases in H associated with uncertainties in aerodynamic
characteristics including Ts. In traditional SEBAL applications [5, 6], parameters a
and b in Eq. (7) are determined by setting dTcoldSEBAL = 0 for an extreme “cold” pixel,
where Ts cold is set equal to the Ts of a local water body, or in its absence, a well
vegetated field. SEBAL anticipates H = 0 at the Ts cold. For a selected “hot” pixel in
the image, it is assumed dry enough for LE = 0 and H = Rn – G. Eq. (6) is rearranged:

(R n − G ) rah1−2 hot (8)


dThot SEBAL =
ρ air hot C p

where rah1-2 hot is rah computed for the roughness and stability conditions of the hot
pixel (between z1 and z2) and ρair hot is ρair at the hot pixel. Using dTcold SEBAL = 0
and dThot SEBAL from Eq. (8), values for a and b in Eq. (7) are determined. This
approach for dT in SEBAL reduces or cancels biases 1-6 of Table 1 by incorporating
these into the calibration of H and dT through the “anchoring” of H at the cold and hot
pixels. Bias no. 4 is especially reduced by the use of the floating dT, and impacts of
biases 1-3 and 6 are reduced by defining the value for H at the hot pixel as being equal
to Rn – G. Any biases in Rn or G at the hot pixel in the image are transferred into H.
However, this bias introduced into H is transferred back out of the energy balance
during the calculation of LE from Eq. (1), since the bias is present in both Rn – G and
H, and thus cancels. This is one of the valuable benefits of the inverted internal
calibration procedure developed by [4, 5] and employed in SEBAL and METRIC.
In METRIC, a similar approach and assumptions are made as in SEBAL for
determining dThot for the hot pixel and for eliminating biases, although a daily surface
soil water balance such as that of FAO [1] is run for a bare soil condition for each
image date to confirm that ET = 0 for the hot pixel or to supply a nonzero value if
there is evaporation from antecedent precipitation. Thus, in METRIC:
H hot METRIC = R n − G − LE hot (9)

as derived from inversion of Eq. (1), where LEhot is the latent residual evaporation at
the hot pixel determined from a daily water balance of the top 0.1 to 0.15 m soil layer.

130

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
At the “cold” pixel of METRIC, we define the sensible heat flux as:
H cold METRIC = R n − G − LE cold (10)

Experience shows that the top 2 to 3% of agricultural fields in an image that are at
full cover have ET rates that are about 5% above the alfalfa reference ETr [12, 13].
This phenomenon occurs because, for a large population of fields, some fields have
wet soil beneath a vegetation canopy that tends to increase the ET rate to 5% or more
above that of the ETr standard. The ETr definition represents the “mean” evaporative
condition for a 0.5 m tall alfalfa having LAI ~ 5 and surface resistance of 30 s m-1
during daytime. Hence, the value for ETcold is based on a representative “cold” pixel
selected from the satellite scene and is set at 1.05 × ETr. Exceptions to the assumption
ETcold = 1.05 ETr occur during nongrowing periods and early in growing seasons
where the amount of vegetation for all pixels is less than that of the 0.5 m reference.
In these cases, a ratio ETcold/ETr = f(NDVI) is developed for the particular image.
The near surface temperature gradients over the cold pixel, dTcoldMETRIC, and hot
pixel, dThotMETRIC are computed using Eq. (6) where rah is corrected for the roughness
and stability conditions at each pixel and ρair is calculated from Ts. In both SEBAL
and METRIC, the a and b coefficients of Eq. (7) are determined using the two pairs of
values for dT and Ts where:
dThot − dTcold
a= (11)
Ts hot − Ts cold
and

b = dThot − a Ts hot (12)

where Ts hot and Ts cold are the surface temperatures associated with the hot and cold
pixels adjusted to a common elevation datum and customized lapse rate.
In METRIC, we use the ASCE standardized Penman-Monteith equation for the
alfalfa reference ETr [3], which is typically 20 to 30 percent greater than grass
reference ETo. METRIC can be applied without crop classification, so that the
specific crop type at any pixel is generally not known. Because the crop coefficient,
Kc, typically peaks at 1.0 for many crops when using the alfalfa reference basis [8,
15], the cold pixel can be nearly any agricultural crop having complete ground cover
and LAI greater than approximately 4. The freedom from requiring crop classification
constitutes a significant cost savings, as accurate crop classification is labor intensive.
Our confidence in accuracy and dependability of the standardized ASCE Penman-
Monteith equation for calibration of the EB is strong. The equation has exhibited
good sensitivity to hourly changes in weather, as demonstrated in Fig. 1, where
measured and computed ETr are compared for two days near Kimberly Idaho. The
equation provides a robust approximation of ET expected from extensive surfaces of
full vegetation cover. Each satellite image is auto-calibrated in METRIC using ETr
(based on hourly or shorter period weather data).
The calibration of the equations for H, and in essence the entire EB, to ETr corrects
the EB for lingering systematic computational biases 1-6 of Table 1. The use of ETr
to calibrate the EB accounts for influence of advection on ET and EB for vegetated

131

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
pixels and for biases in Rn and G. Some estimation error and bias is of course
introduced to the EB by the ETr itself. This error and bias is estimated to be on the
order of 5and
pixels to 10%.
for biases in R and G. Some estimation error and bias is of course
n
introduced to the EB by the ETr itself. This error and bias is estimated to be on the
Table 1.order
Estimated biases in energy balance components and their causes and impacts.
of 5 to 10%.
Est. Bias
Table 1. Estimated biases in energyinbalance
Est. Bias ET components and their causes and impacts.
Bias in Est. Bias
expressed Cause of Bias Impact of Bias
Est. Bias as
component in%ETof
Bias (%)in expressed
full ET Cause of Bias Impact of Bias
component as % of Error/bias in sensor, in est. Bias in net radiation
Albedo Other: 10-20 Other: 5-10
(%)
SEBAL: 15
full ET
SEBAL: 3 path reflectance, in
1 calculation
Albedo Other: 10-20 Other: 5-10 Error/bias in sensor,
transmissivity in est.
of short-wave Bias in net radiation
METRIC: 10 METRIC: 2 path
1 calculation SEBAL: 15 SEBAL: 3 and reflectance,
atmosphericincorrection,
METRIC: 10 transmissivity of short-wave
METRIC: 2 and in integration of bands
and atmospheric correction,
and in integration of bands
Net radiation Other: 3-7 Other: 3-7 Error/bias in incoming-long- Bias in est. ET from
Other: 6
SEBAL: 3-7 SEBAL: 3
Other: 3-7
wave andinoutgoing
Error/bias emissivityBiasenergy
incoming-long- bal.from
in est. ET
2 Net radiation
calculation and Tand
SEBAL:56
METRIC: SEBAL: 33
METRIC: wave s and in albedo
outgoing (LEbal.
emissivity energy = Rn – G – H)
2 calculation and Ts and in
Thermal band / 2 5
METRIC:
Other: METRIC:
Other: 5 3 Error/bias inalbedo
narrow band (LEBias
= Rnin
–G – H)
estimation of
Thermal band / 2
Other: 2
SEBAL: Other: 55
SEBAL: emissivity,
Error/bias in atmospheric
narrow band Biasthermal gradients
in estimation of and
3 surface SEBAL:22 SEBAL: 55 emissivity, atmospheric
correction, satellite sensor thermal gradients and
sensible heat flux for
3 surface METRIC: METRIC: correction, satellite sensor sensible
someheat flux for
temperature METRIC: 2 METRIC: 5 land uses
temperature some land uses
Air Air
temperature Other: 20-50 Other: 10-40 Other: bias in radiometric vs. Bias and error in
Other: 20-50
temperature SEBAL: Other: 10-40 Other: bias in radiometric vs. Bias and error in
10-20 SEBAL: 5 aerodynamictemperature
aerodynamic temperature if if calculated
calculated sensible
sensible
4 gradient
4 gradient SEBAL: 10-20 SEBAL: 5
METRIC:
METRIC:10-20
10-20 METRIC: 3
METRIC: 3
using direct gradient, and error
using direct gradient, and error heatheat flux flux
and and
function used in
function used in in zzoh. . SEBAL
in oh
SEBAL(lack
(lack
of of LE LE
= Rn=–RGn -–HG - H
sensible heat advection
advection consideration)
consideration) andand
sensible heat METRIC:potential
METRIC: potential calibration
calibration
fluxflux
calculation
calculation biasfrom
bias fromassumptions
assumptions at the
at the
cold and hot pixels
cold and hot pixels
5 Aerodynamic Other: 10-30 Other:
10-30 10-20
Other:10-20 Error/bias
Error/biasininroughness, near
roughness, near BiasBias
and and
errorerror
in in
5 Aerodynamic Other:
SEBAL: 10-20 SEBAL: 5 surface wind speed and calculated sensible
resistance SEBAL: 10-20 SEBAL: 5 surface wind speed and calculated sensible
resistance METRIC: 10-20 METRIC: 5
boundary layer stability
boundary layer stability
heat flux and
METRIC: 10-20 METRIC: 5 LE heat
= Rn flux
– G –andH
LE = Rn – G – H
Soil heat flux Other: 50-200 Other: 5-20 Error and potential bias in Bias and error in
Soil heat flux Other: 50-200
SEBAL: 20-100 Other:
SEBAL:5-20
3 Error and for
estimation potential
variousbias in calculated
landuse Bias and error in
6 function SEBAL: 20-100
METRIC: 20-200 SEBAL: 3 estimation
types for various
and vegetation andlanduse
soil LE calculated
= Rn – G - H
6 function METRIC: 2
METRIC: 20-200
METRIC: 2 types and vegetation and soil LE = Rn – G - H
conditions
conditions
Wind speed Other: 20-50 Other: 10 Variation in wind speed at Spatial bias in dT
40 blending
Variationheight and speed
in function, H,bias
and LE =
Wind speed
7 field Other: 20-50
SEBAL: SEBAL:
Other: 105 in wind
extrapolation to ground
at Spatial
Rn –function,
G – H as H,
in dT
a and LE =
40 40
METRIC:
SEBAL: METRIC:
SEBAL: 55 blending height and in
7 field extrapolation to ground
function of location
METRIC: 40 METRIC: 5 andR n – use
land G – H as a
function of location
8 Extrapolation SEBAL: 15 Impacts of advection and and land use
Generally
METRIC: 10 changes in wind speed and underestimation of
of SEBAL: 15 Impacts
vapor of advection
pressure and
deficit during Generally
24-hour ET under
8 Extrapolation
instantaneous METRIC: 10 changes
the in wind
day, bias speed and
in G estimate underestimation
advective conditions of
of ET to daily vapor pressure deficit during w/o24-hourETET
use of r under
instantaneous the day, bias in G estimate advective conditions
ET to daily w/o use of ETr

132

132

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
EXTRAPOLATION TO 24-HOUR PERIODS
The reference ET fraction (ETrF) is calculated as the ratio of the instantaneous ET
(ETinst) from each pixel to the reference ET (ETr) computed from weather data:
ETinst
ETr F = (13)
ETr
where ETinst is from Eq. (1) (mm hr-1) and ETr is for the standardized 0.5 m tall alfalfa
reference at the time of the image (mm hr-1). Generally only one or two weather
stations are required to estimate ETr for a Landsat image that measures 150 km x 150
km. ETrF is similar to the well-known crop coefficient, Kc, and is used to extrapolate
ET from image time to 24-hour or longer periods. ETrF values range from 0 to about
1.0 [8, 15] so that ETr nearly represents an upper bound on ET for large expanses of
well-watered vegetation. At a completely dry pixel, ET = 0 and therefore ETrF = 0.
Kimberly Lysimeters - September 4,1990 Kimberly Lysimeters -September 7, 1990
Data from Dr. J.L Wright
1.10 1.10

0.90 0.90

0.70 0.70
ET, mm/hour
ET, mm/hour

0.50 0.50

0.30 0.30

0.10 0.10

-0.10 -0.10
010003000500070009001100130015001700190021002300
010003000500070009001100130015001700190021002300
Time of Day
Time of Day
Etr Lys. 2 alfalfa
Etr Lys. 2 alfalfa

FIGURE 1. Hourly calculated (ETr) and measured (Lys. 2) alfalfa reference ET for a clear day –
left and cloudy afternoon – right near Kimberly, Idaho (data from Dr. J.L. Wright, USDA-ARS)

24-Hour Evapotranspiration
In METRIC, ET24 is estimated assuming that instantaneous ETrF at image time
equals the average ETrF over the 24-hour period. Figure 2 shows graphs of ET and
ETrF vs. time, based on precision lysimeter measurements from a crop of peas near
Kimberly, Idaho, that demonstrate ETrF to be relatively constant throughout daytime,
but evaporative fraction (EF) to generally increase during the day under the advective
conditions of southern Idaho where large irrigated projects are surrounded by
expansive desert. EF, defined as ET / (Rn-G), is used in many traditional EB
approaches to extrapolate from image time to 24-hour periods and is often assumed
constant over the day. Figure 2 shows that for the most part, the 24-hour value for
ETrF agrees closely with the ETrF derived at image time (about 10:30 – 11:00 am for
Landsat), whereas EF for the 24-hour periods was often significantly higher than that
at image time. Thus, use of ETr to extrapolate to daily periods tends to remove the last
potential bias of Table 1, which is the extrapolation to 24-hour and longer periods.
Fig. 3 (left) shows measured EF for 24-hour periods vs. EF at image time for the
same crop of irrigated peas as in Fig. 2, as measured by weighing lysimeter and net

133

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
radiometer. Fig. 3 (right) shows similar for ETrF. Advection during afternoon and
early evening caused values of EF for the 24-hour period to exceed those at image
time. Agreement was relatively good between 24-hour ETrF values and ETrF at
image time, with some scatter. ETr tends to encapsulate regional advective effects on
ET. Fig. 4 shows measurements of ETrF at image time vs. ETrF for the 24-hour
period for sugar beets, as determined by weighing lysimeter at Kimberly. The
assumption of constant ETrF during a day appears to be generally valid for crops that
have been developed to maximize photosynthesis and thus stomatal conductance.
Similar trends were found for crops of corn, potatoes and beans at Kimberly [11].
ETrF may decrease during afternoon for some native vegetation under water short
conditions where plants endeavor to conserve soil water. Under these conditions, the
24-hour ETrF must be modeled as some fraction of instantaneous ETrF.
Finally, the ET24 (mm/day) is computed for each image pixel as:
ET24 = C rad (ETr F) ETr _ 24 ( (14) )
where ETrF is from satellite overpass time, ETr-24 is the cumulative 24-hour ETr for
the day of the image and Crad is a correction term used in sloping terrain to correct for
variation in ratios of 24-hr vs. instantaneous energy availability [2].

Seasonal Evapotranspiration
Monthly and seasonal evapotranspiration “maps” are often desired for quantifying
total water consumption from agriculture. Here again, ETr is valuable in that ET maps
are derived from a series of ETrF images by interpolating ETrF between processed
images and multiplying, on a daily basis, by ETr for each day. Interpolation of ETrF
between image dates is similar to the construction of a seasonal Kc curve, where
interpolation is between known values for Kc. Interpolation is made using linear or
curvilinear interpolation [15]. We assume that ET for the area of interest changes in
proportion to ETr at the weather station [5]. This is a generally valid assumption and is
similar to the assumptions in the conventional application of Kc x ETr. This approach
applies to clear and cloudy days between satellite image dates [13]. The assumption is
further supported by the use of ETr from the same weather location as is used to derive
ETrF at image time, so that biases caused by differing weather conditions in some part
of an image will be predominately cancelled by using the same ETrF for both
instantaneous and 24-hour period as well as for days between image dates.

SUMMARY AND CONCLUSIONS


Many of the biases that plague energy balance components are accounted for by
SEBAL and METRIC’s internal calibration of sensible heat flux. Additional biases
are accounted for in METRIC by the use of alfalfa reference ET. Reduction in total
bias in ET, as compared to other models that are not routinely internally calibrated,
may be as much as 30%. A disadvantage of METRIC is that it requires high-quality
hourly weather data, which can be difficult to obtain. In contrast, SEBAL can be
applied with a minimum of ground data, which is an important attribute.

134

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
700 3.2 3.6
Peas, Kimberly, ID, 1977, DoY 164 3.0 Peas, Kimberly, ID, 1977, DoY 168 3.4
600 2.8 700 3.2
2.6 3.0
Clear, warm, E. 2.8
500 Clouds 2.4
ET and Rn-G, W/m2

wind. Strong 2.6

ET and Rn-G, W/m2


during 2.2 gusty winds late
500 2.4
400 2.0 afternoon from 2.2
afternoon

Fraction
1.8

Fraction
SW with heavy 2.0
300 and late 1.6 clouds and 1.8
1.4 300 advection and 1.6
afternoon
significant 1.4
200 advection 1.2
nighttime ET 1.2
1.0
100 0.8 1.0
100 0.8
0.6 0.6
0 0.4 0.4
0.2 0.2
-100 0.0 -100 0.0

0 4 8 12 16 20 24 0 4 8 12 16 20 24
Hour of the Day Hour of the Day
Lysimeter ET Rn-G EF ETrF Lysimeter ET Rn-G EF ETrF

EF_24 ETrF_24 ETr EF_24 ETrF_24 ETr

700 3.2 700 3.2


Peas, Kimberly, ID, 1977, DoY 169 3.0 Peas, Kimberly, ID, 1977, DoY 171 3.0
600 2.8 600 2.8
2.6 2.6
500 Clear, warm 2.4 500 2.4

ET and Rn-G, W/m2


ET and Rn-G, W/m2

day 2.2 2.2


400 2.0 400 2.0

Fraction
1.8
Fraction
1.8
300 1.6 300 1.6
1.4 1.4
200 1.2 200 1.2
1.0 1.0
100 0.8 100 0.8
0.6 0.6
0 0.4 0 0.4
0.2 0.2
-100 0.0 -100 0.0

0 4 8 12 16 20 24 0 4 8 12 16 20 24
Hour of the Day Hour of the Day
Lysimeter ET Rn-G EF ETrF
Lysimeter ET Rn-G EF ETrF

EF_24 ETrF_24 ETr


EF_24 ETrF_24 ETr

.
FIGURE 2. Hourly ET and ETrF for an irrigated pea crop vs. time, based on lysimeter observations by
Dr. J.L. Wright, USDA-ARS, Kimberly, Idaho, for four days in June (EF and ETrF for the 24-hour
period are the larger triangle and circle plotted at 11:00, which is general satellite over pass time).

1.6 1.6

Evaporative Fraction Reference ET Fraction


1.4 1.4
Peas, Kimberly, Idaho 1977 Peas, Kimberly, Idaho 1977
1.2 1.2
ET rF for 24-hours
EF for 24-hours

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

EF at 1100 ET rF at 1100

FIGURE 3. Left--EF for the 24-hour period vs. EF at Landsat time and right--ETrF for the 24-hour
period vs. ETrF at Landsat time for an irrigated crop in 1977 (data from Dr. J.L. Wright, USDA-ARS)

135

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
ET r F for Sugar Beets
May - September 1989
1.2
FIGURE 4. ETrF for the 24-hour
period vs. ETrF at satellite time for
1.0
sugar beets near Kimberly, Idaho in
1989 (lysimeter data from Dr.
0.8 J.L.Wright, USDA-ARS)
24 hour ET r F

1:1 line
0.6

0.4

0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Instantaneous ET rF at 11:00 am

ACKNOWLEDGEMENTS
The testing of the various EB and ET models, including ETr, would not be possible
without the long-term, high quality lysimetric data collection by Dr. James L. Wright
at the USDA-ARS research center near Kimberly, Idaho. Dr. Wright established and
managed two precision weighing lysimeters from 1967 through 1992.

REFERENCES
1. Allen, R.G., L.S. Pereira, D. Raes, and M. Smith. Crop Evapotranspiration: Guidelines for Computing Crop
Water Requirements. UN FAO, Irrig. Drain. Pap. 56. Rome, Italy. 300p. (1998).
2. Allen, R.G., M. Tasumi and R. Trezza. Satellite-based energy balance for mapping ET with internalized
calibration (METRIC). J. Irrig. Drain. Engrg., ASCE (submtd). (2005).
3. ASCE – EWRI. The ASCE Standardized reference evapotranspiration equation. ASCE-EWRI Task Comm.
Report, available at http://www.kimberly.uidaho.edu/water/asceewri/. 180 p. (2005).
4. Bastiaanssen W.G.M. Regionalization of surface flux densities and moisture indicators in composite terrain:
A remote sensing approach under clear skies in Mediterranean climates. Ph.D. Diss., CIP Data Koninklijke
Bibliotheek, Den Haag, the Netherlands. 273 p. (1995).
5. Bastiaanssen, W.G.M., M. Menenti, R.A. Feddes, and A.A.M. Holtslag. A remote sensing surface energy
balance algorithm for land (SEBAL): 1. Formulation. J. Hydrol., 212-213:198-212. (1998).
6. Bastiaanssen, W.G.M. SEBAL-based sensible and latent heat fluxes in the irrigated Gediz Basin, Turkey. J.
Hydrol., 229:87-100. (2000).
7. Bastiaanssen W.G.M. et al. SEBAL for spatially distributed ET under actual management and growing
conditions, ASCE J. Irrig. Drain. Engrg 131(1):85-93. (2005).
8. Jensen, M.E., R.D. Burman and R.G. Allen. (eds.) Evapotranspiration and Irrigation Water Requirements,
ASCE Man. on Engineering Practice No. 70. ISBN 0-87262-763-2, 332 p. (1990).
9. Kustas, W. P., et al.,. Surface energy balance estimates at local and regional scales using optical remote
sensing over semiarid rangelands. Water Res. Research, 30(5), 1241-1259. (1994).
10. Norman, J.M., W.P. Kustas, and K.S. Humes. Source approach for est. soil and vegetation energy fluxes in
obs. of direct. radiomet. surface temp. Ag. and For. Meteorol. 77:263-293. (1995).
11. Romero M.G. Daily evapotranspiration estimation by means of evaporative fraction and reference ET
fraction. Ph.D. Diss., Utah State Univ., Logan, Utah. (2004).

136

wnloaded 25 Jul 2013 to 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss
12. Tasumi, M. Progress in operational estimation of regional evapotranspiration using satellite imagery. Ph.D.
Dissertation, University of Idaho, Moscow, ID, 357p. (2003).
13. Tasumi M., Allen R.G., Trezza R. and Wright J.L. Satellite-based energy balance to assess within-population
variance of crop coefficient curves. ASCE J. Irrig. Drain. Engrg.131(1):94-109(2005a).
14. Tasumi M., Allen R.G. and Trezza R. Estimation of at-surface reflectance and albedo from satellite for
routine, operational calculation of land surface energy balance, Agric. Wat. Man. (subm).(2005b).
15. Wright, J.L. New Evapotranspiration Crop Coefficients. J. of Irrig. Drain. Div. (ASCE), 108:57 74. (1982).
16. Wukelic, G.E., D.E. Gibbons, et al., Radiometric calibration of Landsat thematic mapper thermal band.
Remote Sensing of Environment 28:339-347. (1989).

137

Viewto
publication stats
wnloaded 25 Jul 2013 129.101.83.201. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://proceedings.aip.org/about/rights_permiss

You might also like