et al. (2009) performed a study to estimate volumetric surface soil moisture content in
cropped areas in India at field (<10
2 -m) to landscape (£10
3 -m) scales. In situ data
collected at the field scale were used to obtain a soil wetness index (SWI) from which
soil moisture content (q V ) was derived using ASTER data for the field scale and
MODIS at the landscape scale.
Integration of satellite data with spatial data on vegetation and terrain features via
GIS methods have also been used to map ET. Accurate estimation of ET is difficult to
obtain over heterogeneous landscapes with diverse land covers and topographic
terrains (Liu et al., 2007). Mariotto et al. (2011) performed a study to build advanced
remote sensing and land surface energy balance algorithms to map ET in a heterogeneous semiarid area over the U.S. Department of Agriculture, Agricultural Research
Service, Jornada Experimental Range that encompasses parts of southern Arizona,
New Mexico, and Texas. ET of 12 different land covers was computed by applying
the SEBAL model. A GIS raster/vector system was used to integrate multispectral
TIR and reflectance imagery from ASTER with meteorological, terrain, and land
cover data. The study showed that SEBAL run with all these input data provided the
best agreement with ground measurements, in comparison with SEBAL run without
any modification for terrain features and associated data, and it could significantly
discriminate ET among 75.8% of vegetation types. SEBAL without ASTER integrated data set could not discriminate any vegetation types.
The influence of spatial scale on ET estimation using multiple satellite sensors
collected over heterogeneous land surfaces is a critical research need. McCabe and
Wood (2006) used Landsat ETM+ (60-m), ASTER (90-m), and MODIS (1020-m) data
to independently estimate ET. The range of satellite sensor resolutions allows for
analyses that span spatial scales from in situ measurements (i.e., pointscale) to the
MODIS kilometer scale. ET estimates derived at these multiple resolutions were
assessed against eddy covariance flux measurements during the SMACEX campaign
over the Walnut Creek watershed in Iowa. Together, these data allow a comprehensive
scale intercomparison of remotely sensed predictions that included intercomparison of
the ET products from the various sensors as well as a statistical analysis for the retrievals
at the watershed scale. A high degree of consistency was observed between the higher
spatial resolution sensors (ETM+ and ASTER). The MODIS-based estimates were
unable to discriminate the influence of land surface heterogeneity at the field scale but
did effectively reproduce the average ET of the watershed response, which illustrated
the utility of this sensor for regional scale ET estimates.
Further information on the assessment of ET and soil moisture content across
different scales of observation that has implications in deriving ET from satellitebased data is given by Verstraeten et al. (2008). They provide a summary of the
generally accepted theory of ET, a summary of ET assessment at different scales of
observation, a summary of data assimilation schemes for estimating ET using
reflectance and TIR remote sensing data, and a summary of soil moisture retrieval
techniques at different spatial and temporal scales. Another useful reference on
scaling of TIR data for evaporation estimation is given by Li et al. (2009). They
provide an overview of the commonly applied ET models using remote sensing data at
regional scales. They discuss the main inputs, assumptions, theories, advantages, and
44
THERMAL INFRARED REMOTE SENSING FOR ANALYSIS OF LANDSCAPE
cropped areas in India at field (<10
2 -m) to landscape (£10
3 -m) scales. In situ data
collected at the field scale were used to obtain a soil wetness index (SWI) from which
soil moisture content (q V ) was derived using ASTER data for the field scale and
MODIS at the landscape scale.
Integration of satellite data with spatial data on vegetation and terrain features via
GIS methods have also been used to map ET. Accurate estimation of ET is difficult to
obtain over heterogeneous landscapes with diverse land covers and topographic
terrains (Liu et al., 2007). Mariotto et al. (2011) performed a study to build advanced
remote sensing and land surface energy balance algorithms to map ET in a heterogeneous semiarid area over the U.S. Department of Agriculture, Agricultural Research
Service, Jornada Experimental Range that encompasses parts of southern Arizona,
New Mexico, and Texas. ET of 12 different land covers was computed by applying
the SEBAL model. A GIS raster/vector system was used to integrate multispectral
TIR and reflectance imagery from ASTER with meteorological, terrain, and land
cover data. The study showed that SEBAL run with all these input data provided the
best agreement with ground measurements, in comparison with SEBAL run without
any modification for terrain features and associated data, and it could significantly
discriminate ET among 75.8% of vegetation types. SEBAL without ASTER integrated data set could not discriminate any vegetation types.
The influence of spatial scale on ET estimation using multiple satellite sensors
collected over heterogeneous land surfaces is a critical research need. McCabe and
Wood (2006) used Landsat ETM+ (60-m), ASTER (90-m), and MODIS (1020-m) data
to independently estimate ET. The range of satellite sensor resolutions allows for
analyses that span spatial scales from in situ measurements (i.e., pointscale) to the
MODIS kilometer scale. ET estimates derived at these multiple resolutions were
assessed against eddy covariance flux measurements during the SMACEX campaign
over the Walnut Creek watershed in Iowa. Together, these data allow a comprehensive
scale intercomparison of remotely sensed predictions that included intercomparison of
the ET products from the various sensors as well as a statistical analysis for the retrievals
at the watershed scale. A high degree of consistency was observed between the higher
spatial resolution sensors (ETM+ and ASTER). The MODIS-based estimates were
unable to discriminate the influence of land surface heterogeneity at the field scale but
did effectively reproduce the average ET of the watershed response, which illustrated
the utility of this sensor for regional scale ET estimates.
Further information on the assessment of ET and soil moisture content across
different scales of observation that has implications in deriving ET from satellitebased data is given by Verstraeten et al. (2008). They provide a summary of the
generally accepted theory of ET, a summary of ET assessment at different scales of
observation, a summary of data assimilation schemes for estimating ET using
reflectance and TIR remote sensing data, and a summary of soil moisture retrieval
techniques at different spatial and temporal scales. Another useful reference on
scaling of TIR data for evaporation estimation is given by Li et al. (2009). They
provide an overview of the commonly applied ET models using remote sensing data at
regional scales. They discuss the main inputs, assumptions, theories, advantages, and
44
THERMAL INFRARED REMOTE SENSING FOR ANALYSIS OF LANDSCAPE