Microwave radiometric measurements of soil moisture in Italy

Abstract

Within the framework of the MAP and RAPHAEL projects, airborne experimental campaigns were carried out by the IFAC group in 1999 and 2000, using a multifrequency microwave radiometer at L, C and X bands (1.4, 6.8 and 10 GHz). The aim of the experiments was to collect soil moisture and vegetation biomass information on agricultural areas to give reliable inputs to the hydrological models. It is well known that microwave emission from soil, mainly at L-band (1.4 GHz), is very well correlated to its moisture content. Two experimental areas in Italy were selected for this project: one was the Toce Valley, Domodossola, in 1999, and the other, the agricultural area of Cerbaia, close to Florence, where flights were performed in 2000. Measurements were carried out on bare soils, corn and wheat fields in different growth stages and on meadows. Ground data of soil moisture (SMC) were collected by other research teams involved in the experiments. From the analysis of the data sets, it has been confirmed that L-band is well related to the SMC of a rather deep soil layer, whereas C-band is sensitive to the surface SMC and is more affected by the presence of surface roughness and vegetation, especially at high incidence angles. An algorithm for the retrieval of soil moisture, based on the sensitivity to moisture of the brightness temperature at C-band, has been tested using the collected data set. The results of the algorithm, which is able to correct for the effect of vegetation by means of the polarisation index at X-band, have been compared with soil moisture data measured on the ground. Finally, the sensitivity of emission at different frequencies to the soil moisture profile was investigated. Experimental data sets were interpreted by using the Integral Equation Model (IEM) and the outputs of the model were used to train an artificial neural network to reproduce the soil moisture content at different depths. Keywords: microwave radiometry, soil moisture mapping, river basins, vegetative biomass, neural networks