リモートセンシングとGISのジャーナル

概要

Application of Temporal Sentinel 1 SAR Data for Multiple Crop Type Classification in a Command Area

Sobhan Mishra, Annie Maria Issac, Syama S Rao, Ronald Singh, P V Raju, V V Rao

Assessment of cropped area during kharif season is a difficult task due to the presence of cloud, cloud shadows and haze. So microwave datasets provide a good alternative as they have cloud penetration capacity. But deriving crop related information from microwave datasets is difficult task as it is subjected to different factors like phonological stage of the crop during satellite image acquisition, presence of speckle, polarization and the classifier used. In this study suitable filter polarization and classifier is identified by systematic analysis of time series, back scatter values derived from Sentinel 1 Synthetic Aperture Radar (SAR) data. As per the study, for the selected study area and time period, Sentinel 1-SAR images subjected to speckle removal by Intensity Driven Adaptive Filter (IDAN) filter proved to perform well in classification as against other filters. The time series of speckle removed VH polarization images classified using Random Forest classifier gave an accuracy of 45 percent in classifying paddy, nonpaddy and fallow.