17:15 〜 18:30
[ACG36-P18] Soil Moisture and Plant Water Content estimate in Pilot areas of the Mediterranean basin through multitemporal analysis of AMSR-E/2 images
キーワード：Microwave radiometry, Soil Moisture, Vegetation Biomass
This study aims contributing to the analysis of climatic change evaluation on regional scale in sensible areas like the Mediterranean basin. Two target areas have been identified in North-Africa in the framework of a European project devoted to the analysis of drought conditions and climate change impacts (ERANET-MED OptiMed-Water). The selected areas are the Nile Delta region in Egypt and two areas in Eastern Tunisia, whose main land use is represented by agriculture both rainfed and irrigated.
The goal of this research was to carry out a multi-temporal analysis on soil moisture content (SMC, in cm3/cm3) and vegetation biomass, expressed as Plant Water Content (PWC, in Kg/m2), through large spatial resolution microwave images. The scarce ground resolution (about 10Km at all frequencies) of microwave radiometers from satellite platforms (AMSR-E and AMSR2) can be considered a limit for in-depth analysis related to drought/water waste phenomena at local scale. However, the availability of long-term series of data starting from 2002 makes the investigation of climate change cycles possible through the observation of the seasonal variations of brightness temperatures (Tb) at C, X and Ku bands and the corresponding polarization differences, the latter expressed as polarization indices (PI). Both Tb and PI are related to SMC and PWC. In addition to the frequent revisit time, the possibility of two daily passages (night/day) may also provide useful information on the evapotranspiration processes.
As a preliminary investigation, the temporal trends of Tb and PI at C, X and Ku bands collected on these areas have been analyzed from 2002 to today, pointing out the occurred seasonal variations. Successively, these microwave parameters have been processed through an inversion algorithm (HydroAlgo) implemented at IFAC-CNR, which is able to estimate both SMC and PWC. HydroAlgo is based on neural network approach and is trained using both experimental and model data derived from the tau-omega model. The model, which was validated in the entire Italian territory by using the SMC product derived from a hydrological model as a comparison, provided an overall Pearson’s correlation coefficient between the estimated and the target SMC higher than 0.8 and the corresponding root mean square error less than 0.055 m3/m3. Slightly different results were obtained using ascending or descending overpasses.
Comparisons with MODIS products have been carried out to identify spatial correlations during the analyzed period and validate the PWC products.
The goal of this research was to carry out a multi-temporal analysis on soil moisture content (SMC, in cm3/cm3) and vegetation biomass, expressed as Plant Water Content (PWC, in Kg/m2), through large spatial resolution microwave images. The scarce ground resolution (about 10Km at all frequencies) of microwave radiometers from satellite platforms (AMSR-E and AMSR2) can be considered a limit for in-depth analysis related to drought/water waste phenomena at local scale. However, the availability of long-term series of data starting from 2002 makes the investigation of climate change cycles possible through the observation of the seasonal variations of brightness temperatures (Tb) at C, X and Ku bands and the corresponding polarization differences, the latter expressed as polarization indices (PI). Both Tb and PI are related to SMC and PWC. In addition to the frequent revisit time, the possibility of two daily passages (night/day) may also provide useful information on the evapotranspiration processes.
As a preliminary investigation, the temporal trends of Tb and PI at C, X and Ku bands collected on these areas have been analyzed from 2002 to today, pointing out the occurred seasonal variations. Successively, these microwave parameters have been processed through an inversion algorithm (HydroAlgo) implemented at IFAC-CNR, which is able to estimate both SMC and PWC. HydroAlgo is based on neural network approach and is trained using both experimental and model data derived from the tau-omega model. The model, which was validated in the entire Italian territory by using the SMC product derived from a hydrological model as a comparison, provided an overall Pearson’s correlation coefficient between the estimated and the target SMC higher than 0.8 and the corresponding root mean square error less than 0.055 m3/m3. Slightly different results were obtained using ascending or descending overpasses.
Comparisons with MODIS products have been carried out to identify spatial correlations during the analyzed period and validate the PWC products.