392
S. Paloscia
as well, looking for a sensitivity to the soil moisture content. Some interesting results have
been achieved in this sense by Dobson et al. (1992) using a first-order vector radiative transfer
model (MIMICS) (Ulaby et al., 1990) with which the aO values over forest and grasslands has
been predicted. In order to estimate the usefulness of ERS-1 SAR for estimating near-surface
soil moisture, MIMICS simulations of aO have been compared to ERS-1 data on grasslands.
The results showed that model outputs and ERS-l measurements agreed to within 0.4 dB,
whereas on forests the ERS-1 backscattering is dominated by properties of the crown layer and
consequently none relationship with soil moisture can be obtained. The conclusion can be that
the retrieval of soil moisture from ERS-1 data is feasible for surfaces covers characterized by
low biomass « 1 kg/m 2 ) such as short grasses.
JERS-1 is a satellite really conceived for soil and vegetation studies. Interesting results have
been obtained in estimating biomass both of herbaceous and arboreous crops. On soil moisture
some works have been carried out mainly by Japanese researchers and presented on the final
report of JERS-I/ERS-1 system verification program (1995). Musiake et al. (1995) presented
a comparison between ERS-1 and JERS-1 data in detecting soil moisture of agricultural fields.
In general a fairly good relationship has been obtained between JERS-1 aO and the saturation
degree S (which is the soil moisture content normalized to the saturation of soil), obtaining a
regression equation of this type: aO = 0.04585 - 14.219, with a correlation coefficient of 0.93.
This means that when saturated water content is 50% the sensitivity to soil water content is 0.1
dB/%. As far as ERS-1 data are concerned, a correlation with soil moisture variations is only
obtained using multi temporal data sets, when the roughness and vegetation characteristics of
the observed fields do not change in time.
MAC Europe campaign
A milestone in the improvement of SAR research has been represented by the NASAl JPL Aircraft Campaign named MAC-91 which has been conducted within the framework of SIR-C/XSAR Project. It is well-known that, in general, multifrequency Imultipolarization observations
make it possible to obtain more information than single channel measurements do. For example: a combination of airborne active microwave (C band) and passive thermal infrared remote
sensing (Soares et al., 1987), has shown that the spatial behaviour of surface temperature is
correlated to the fields' canopy whilst the backscatter depends on soil parameters and mainly
on SMC. Indeed, this experiment has been a systematic investigation, which aimed toward a
better understanding of the information obtainable in agro-hydrology from the multi-frequency
polarimetric SAR, and allowed the gathering of a large and useful data set which includes
multi-temporal, multi-frequency and multi-incidence angle, polarimetric SAR measurements
and ground truth.
In summer 1991, a DC-8, carrying a multifrequency, multipolarization SAR at P, Land C
bands on board (AIRSAR), was flown over several European and American test-sites which
were imaged on several dates and with different incidence angles. The SAR resolution in the
multilook data was 6.66 m in slant range and 12 m in azimuth. In the framework of this
experiment also the test site of Montespertoli (Italy), which was selected as a SIR-C I X-SAR
'supersite' for Hydrology, was imaged by the AIRSAR (Canuti et al., 1992). The landuse is
represented by vineyards, oliveyards and some woods in the hills and agricultural crops (wheat,
corn, sorghum, sunflower, alfalfa) in the flat areas along the rivers.
An example of recorded data is represented in Fig. 14.22, which shows the slant range images
of the agricultural area at P-, L- and C-band in VH polarization. A simple photointerpretation
of this composite image shows that the Pes a river which crosses the scene can easily be identified, as well as the village of Cerbaia on the left side which shows a very strong and diffused
S. Paloscia
as well, looking for a sensitivity to the soil moisture content. Some interesting results have
been achieved in this sense by Dobson et al. (1992) using a first-order vector radiative transfer
model (MIMICS) (Ulaby et al., 1990) with which the aO values over forest and grasslands has
been predicted. In order to estimate the usefulness of ERS-1 SAR for estimating near-surface
soil moisture, MIMICS simulations of aO have been compared to ERS-1 data on grasslands.
The results showed that model outputs and ERS-l measurements agreed to within 0.4 dB,
whereas on forests the ERS-1 backscattering is dominated by properties of the crown layer and
consequently none relationship with soil moisture can be obtained. The conclusion can be that
the retrieval of soil moisture from ERS-1 data is feasible for surfaces covers characterized by
low biomass « 1 kg/m 2 ) such as short grasses.
JERS-1 is a satellite really conceived for soil and vegetation studies. Interesting results have
been obtained in estimating biomass both of herbaceous and arboreous crops. On soil moisture
some works have been carried out mainly by Japanese researchers and presented on the final
report of JERS-I/ERS-1 system verification program (1995). Musiake et al. (1995) presented
a comparison between ERS-1 and JERS-1 data in detecting soil moisture of agricultural fields.
In general a fairly good relationship has been obtained between JERS-1 aO and the saturation
degree S (which is the soil moisture content normalized to the saturation of soil), obtaining a
regression equation of this type: aO = 0.04585 - 14.219, with a correlation coefficient of 0.93.
This means that when saturated water content is 50% the sensitivity to soil water content is 0.1
dB/%. As far as ERS-1 data are concerned, a correlation with soil moisture variations is only
obtained using multi temporal data sets, when the roughness and vegetation characteristics of
the observed fields do not change in time.
MAC Europe campaign
A milestone in the improvement of SAR research has been represented by the NASAl JPL Aircraft Campaign named MAC-91 which has been conducted within the framework of SIR-C/XSAR Project. It is well-known that, in general, multifrequency Imultipolarization observations
make it possible to obtain more information than single channel measurements do. For example: a combination of airborne active microwave (C band) and passive thermal infrared remote
sensing (Soares et al., 1987), has shown that the spatial behaviour of surface temperature is
correlated to the fields' canopy whilst the backscatter depends on soil parameters and mainly
on SMC. Indeed, this experiment has been a systematic investigation, which aimed toward a
better understanding of the information obtainable in agro-hydrology from the multi-frequency
polarimetric SAR, and allowed the gathering of a large and useful data set which includes
multi-temporal, multi-frequency and multi-incidence angle, polarimetric SAR measurements
and ground truth.
In summer 1991, a DC-8, carrying a multifrequency, multipolarization SAR at P, Land C
bands on board (AIRSAR), was flown over several European and American test-sites which
were imaged on several dates and with different incidence angles. The SAR resolution in the
multilook data was 6.66 m in slant range and 12 m in azimuth. In the framework of this
experiment also the test site of Montespertoli (Italy), which was selected as a SIR-C I X-SAR
'supersite' for Hydrology, was imaged by the AIRSAR (Canuti et al., 1992). The landuse is
represented by vineyards, oliveyards and some woods in the hills and agricultural crops (wheat,
corn, sorghum, sunflower, alfalfa) in the flat areas along the rivers.
An example of recorded data is represented in Fig. 14.22, which shows the slant range images
of the agricultural area at P-, L- and C-band in VH polarization. A simple photointerpretation
of this composite image shows that the Pes a river which crosses the scene can easily be identified, as well as the village of Cerbaia on the left side which shows a very strong and diffused