Land Use/cover Classification and Rice Mapping Based on Envisat Asar Data

The main objective of this project is to develop and test methodology using ENVISAT ASAR data for agricultural applications, with an emphasis on land use, land cover classification and rice mapping. An optimal pre-processing chain for ASAR data is first constructed to provide input data to the classification steps. Experiments at the Zhangzhou test site, Fujian province, southern China, indicate that rice mapping based on Principal Components Analysis is effective at packing multi-temporal information on rice fields into a dominant component and gives results similar to a rule-based approach. A radial basis function neural network approach provides reasonable accuracies for broad land cover classification, with clear gains in statistical accuracy as more data are added, but without major impacts on the visual quality of the classified images Key word: ENVISAT ASAR data, Agriculture, Land use and land cover classification, Rice mapping, Zhangzhou region, Fujian Province