Search

An experiment was carried out over a flat agriculture area located at Gongzhuling, in the Jilin province of China. Four adjacent corn fields were selected as the test targets and the plant and soil parameters were collected over a growing season for the models inputs. Six multitemporal and multiangle ASAR AP images (C band, HH and HV) were acquired during the experiment. This paper presents a semi-experience model for retrieval of soil moisture from the corn field during the growing cycle using the measurement and radar data. Firstly, angular normalization of ASAR data using a coherent scattering model. Then, using the bare soil backscattering model AIEM and the ground measurements in one of the corn fields, the ratio of the modeled bare soil scattering contribution and the observed backscattering coefficient after angular normalization was expressed as the function of vegetation water content. Finally, the neural network approach was used to retrieve the soil moisture. The inversion results are validated by the in situ measurements of the other corn fields.