9 Soil Moisture
213
search need is to have the hydrology community try using the available data with
hydrologic models. This will undoubtedly lead to an improved understanding of
the value of these data band and the modification or development of new models
to effectively use these data.
There is a need to develop and improve algorithms to extract volumetric soil
moisture directly from the microwave measurement (backscatter coefficient or
brightness temperature). To do this, the other target characteristics of vegetation
and surface roughness will have to be parameterized. As discussed previously,
there is great progress being made in these areas but much more needs to be done.
Connected directly to this need is a need to better understand the effects of surface
roughness on the measured microwave response with respect to incidence angle,
azimuth angle, wavelength, and polarization. New methods to measure surface
roughness need to be explored. The dual frequency or Dk radar technique has been
quite effective to measure sea-surface roughness (Schuler et aI., 1991). Implementation of this technique to soil surfaces can provide accurate measure of
roughness over a very large scale. Also, there is a need to understand the effect of
the vegetation canopy on the microwave response. Progress has been made in this
area with both the modeling of radar (Ulaby et al. 1990) and a dual frequency, dual
polarization approach for passive microwave (Njoku, 1999). Vegetation variables
include the geometry for the individual plant as well as the canopy as a whole, the
water content (and perhaps the biochemical makeup) of the plant, and its stage of
growth. Microwave variables would include the incidence angle, the azimuth angle, wavelength, and polarization. Some difficult problems such as soil moisture
estimation from rocky soil, effects of discontinuous canopy or vegetation clumps
on soil moisture estimation, etc., also need to be addressed.
There is a need to investigate the use of change detection algorithms or statistical techniques like the principal components analysis (Verhoest et al. 1998) for
determining the relative soil moisture of an area and whether or not this information can be useful for hydrologists. Change detection should minimize the influence of target variables such as roughness and vegetation, at least over short time
intervals. With change detection it is assumed that the only target change occurring
is the soil moisture. Thus, any measured changes in brightness temperature or
backscatter can be related directly to changes in soil moisture. Fortunately, both
the brightness temperature and backscatter relationships with soil moisture are
approximately linear. There is also a reasonable basis for expecting change detection methods to provide adequate data for agricultural and hydrologic applications
if the data are collected from a long term orbiting platform. Long term (multi season or year) data will establish the upper (wet) and lower (dry) limits for the
change algorithm.
There is a need to develop software procedures for correcting the effects of terrain on the microwave response. Active microwave (SAR) is especially sensitive to
this. This includes foreshortening, layover, and local incidence angle effects. Also,
a potential issue is the relative accuracy of the DEM data with respect to the spatial resolution of the microwave data and the potential effect of subpixel variability
on the measured signal.
There is also a need to investigate the potential for polarimetric SAR and its potential for abstracting target information such as the surface roughness and vegeta-
213
search need is to have the hydrology community try using the available data with
hydrologic models. This will undoubtedly lead to an improved understanding of
the value of these data band and the modification or development of new models
to effectively use these data.
There is a need to develop and improve algorithms to extract volumetric soil
moisture directly from the microwave measurement (backscatter coefficient or
brightness temperature). To do this, the other target characteristics of vegetation
and surface roughness will have to be parameterized. As discussed previously,
there is great progress being made in these areas but much more needs to be done.
Connected directly to this need is a need to better understand the effects of surface
roughness on the measured microwave response with respect to incidence angle,
azimuth angle, wavelength, and polarization. New methods to measure surface
roughness need to be explored. The dual frequency or Dk radar technique has been
quite effective to measure sea-surface roughness (Schuler et aI., 1991). Implementation of this technique to soil surfaces can provide accurate measure of
roughness over a very large scale. Also, there is a need to understand the effect of
the vegetation canopy on the microwave response. Progress has been made in this
area with both the modeling of radar (Ulaby et al. 1990) and a dual frequency, dual
polarization approach for passive microwave (Njoku, 1999). Vegetation variables
include the geometry for the individual plant as well as the canopy as a whole, the
water content (and perhaps the biochemical makeup) of the plant, and its stage of
growth. Microwave variables would include the incidence angle, the azimuth angle, wavelength, and polarization. Some difficult problems such as soil moisture
estimation from rocky soil, effects of discontinuous canopy or vegetation clumps
on soil moisture estimation, etc., also need to be addressed.
There is a need to investigate the use of change detection algorithms or statistical techniques like the principal components analysis (Verhoest et al. 1998) for
determining the relative soil moisture of an area and whether or not this information can be useful for hydrologists. Change detection should minimize the influence of target variables such as roughness and vegetation, at least over short time
intervals. With change detection it is assumed that the only target change occurring
is the soil moisture. Thus, any measured changes in brightness temperature or
backscatter can be related directly to changes in soil moisture. Fortunately, both
the brightness temperature and backscatter relationships with soil moisture are
approximately linear. There is also a reasonable basis for expecting change detection methods to provide adequate data for agricultural and hydrologic applications
if the data are collected from a long term orbiting platform. Long term (multi season or year) data will establish the upper (wet) and lower (dry) limits for the
change algorithm.
There is a need to develop software procedures for correcting the effects of terrain on the microwave response. Active microwave (SAR) is especially sensitive to
this. This includes foreshortening, layover, and local incidence angle effects. Also,
a potential issue is the relative accuracy of the DEM data with respect to the spatial resolution of the microwave data and the potential effect of subpixel variability
on the measured signal.
There is also a need to investigate the potential for polarimetric SAR and its potential for abstracting target information such as the surface roughness and vegeta-