not a strong correlation with water budget factors in relation to the maximum T s
deviation composite method.
Three representative studies using MODIS data illustrate the potential of using
these data for ET estimation. Modeling of actual daily ET in combination with
MODIS data by Sánchez et al. (2007) allowed for the determination of surface fluxes
over boreal forests on a daily basis from instantaneous information registered from a
conventional meteorological tower as well as the canopy temperatures (T c ) retrieved
from satellite. The comparison between T c ground measured with a thermal infrared
radiometer at the meteorological sites and T c retrieved from MODIS showed an
estimation error of ±1.4°C. Their modeling method was validated over the study site
using 21 MODIS images from 2002 and 2003. The results were compared with eddycorrelation ground measurements; with an accuracy of ±1.0 mm/day and an overestimation of 0.3 mm/day shown in daily ET retrieval. Mallick et al. (2007) used
MODIS optical and thermal band data and ground observations to estimate evaporative fraction and daily actual ET (AET) over agricultural areas in India. Five study
regions, each covering a 10 km ´ 10 km area falling over agricultural land uses, were
selected for ground observations at a time closest to MODIS overpasses. Eight
MODIS scenes collected between August 2003 and January 2004 were resampled to
1 km and used to generate surface albedo, land surface temperature, and emissivity.
Evaporative fraction and daily AET were generated using a fusion of MODIS-derived
land surface variables coincident with ground observations. Land cover classes were
assigned using a hierarchical decision rule applied to multidate NDVI and applied via
a triangle method to estimate the relationship between NDVI and surface temperature.
Energy balance daily AET from the fused MODIS data was found to deviate from
water balance AET by between 4.3 and 24.5% across five study sites with a mean
deviation of 11.6%. The RMSE from the energy balance AET was found to be 8% of
the mean water balance AET. Thus, the satellite-based energy balance approach can
be used to generate spatial AET, but as noted by the investigators, further refinement
of this technique should produce more robust results.
Remote sensing with multispectral infrared can improve regional estimates of ET
by providing new constraints on land surface energy balance. Current models use
visible and near-infrared bands to obtain vegetated cover and in some cases utilize
TIR data; these data together yield good ET estimates. However, it may be possible to
enhance theseET models by using emissivity estimates derived from TIR emissivity,
which is a property related to fractional vegetation cover but independent of plant
greenness (French and Inamdar, 2010). This is demonstrated in a study using MODIS
observations obtained over Oklahoma and Kansas which were compared with
changes in NDVI for winter wheat and grazing land. It was found that emissivity
changes were independent of NDVI and sensitive to standing canopies, regardless of
growth stage or senescence. Therefore, emissivities were seasonally dynamic, able to
detect wheat harvest timing, and helpful for modeling ET.
Data combinations from different satellite sensors potentially provide even more
useful information on soil wetness than by using data from one satellite platform
alone. Surface soil wetness determines moisture availability that controls the response
and feedback mechanisms between land surface and atmospheric process. Mallick
EVAPORATION/EVAPOTRANSPIRATION/SOIL MOISTURE
43
deviation composite method.
Three representative studies using MODIS data illustrate the potential of using
these data for ET estimation. Modeling of actual daily ET in combination with
MODIS data by Sánchez et al. (2007) allowed for the determination of surface fluxes
over boreal forests on a daily basis from instantaneous information registered from a
conventional meteorological tower as well as the canopy temperatures (T c ) retrieved
from satellite. The comparison between T c ground measured with a thermal infrared
radiometer at the meteorological sites and T c retrieved from MODIS showed an
estimation error of ±1.4°C. Their modeling method was validated over the study site
using 21 MODIS images from 2002 and 2003. The results were compared with eddycorrelation ground measurements; with an accuracy of ±1.0 mm/day and an overestimation of 0.3 mm/day shown in daily ET retrieval. Mallick et al. (2007) used
MODIS optical and thermal band data and ground observations to estimate evaporative fraction and daily actual ET (AET) over agricultural areas in India. Five study
regions, each covering a 10 km ´ 10 km area falling over agricultural land uses, were
selected for ground observations at a time closest to MODIS overpasses. Eight
MODIS scenes collected between August 2003 and January 2004 were resampled to
1 km and used to generate surface albedo, land surface temperature, and emissivity.
Evaporative fraction and daily AET were generated using a fusion of MODIS-derived
land surface variables coincident with ground observations. Land cover classes were
assigned using a hierarchical decision rule applied to multidate NDVI and applied via
a triangle method to estimate the relationship between NDVI and surface temperature.
Energy balance daily AET from the fused MODIS data was found to deviate from
water balance AET by between 4.3 and 24.5% across five study sites with a mean
deviation of 11.6%. The RMSE from the energy balance AET was found to be 8% of
the mean water balance AET. Thus, the satellite-based energy balance approach can
be used to generate spatial AET, but as noted by the investigators, further refinement
of this technique should produce more robust results.
Remote sensing with multispectral infrared can improve regional estimates of ET
by providing new constraints on land surface energy balance. Current models use
visible and near-infrared bands to obtain vegetated cover and in some cases utilize
TIR data; these data together yield good ET estimates. However, it may be possible to
enhance theseET models by using emissivity estimates derived from TIR emissivity,
which is a property related to fractional vegetation cover but independent of plant
greenness (French and Inamdar, 2010). This is demonstrated in a study using MODIS
observations obtained over Oklahoma and Kansas which were compared with
changes in NDVI for winter wheat and grazing land. It was found that emissivity
changes were independent of NDVI and sensitive to standing canopies, regardless of
growth stage or senescence. Therefore, emissivities were seasonally dynamic, able to
detect wheat harvest timing, and helpful for modeling ET.
Data combinations from different satellite sensors potentially provide even more
useful information on soil wetness than by using data from one satellite platform
alone. Surface soil wetness determines moisture availability that controls the response
and feedback mechanisms between land surface and atmospheric process. Mallick
EVAPORATION/EVAPOTRANSPIRATION/SOIL MOISTURE
43