Detection Paddy Field using dual Polarization SAR Sentinel-1 Data

Paddy field conversion monitoring is necessary conducted to ensure successful the harvest of rice. The monitoring can be done by using satellite data, both optical and radar data, which can cover a large area. In tropical area, cloudy day usually occurred, so that problems can’t be handle with optical data. Utilization of radar data that can penetrate the cloud condition can solve the problem, either as a complement of optical data or used alone to monitor the paddy field conversion. The research was conducted to investigate the capability of Sentinel 1 SAR multi temporal data to detect paddy field based on growth phenology of rice crop. This research explores SAR Sentinel-1A data (C-band, VV and VH polarization) for several growing seasons in 2017 (January-December) to detect paddy fields in Subang Regency, West Java. Stacking layer is carried out prior to classification, time series image and polarization composite image (VH/VV), stacking maximum, minimum, mean and range values, standard deviation and taking sample training and statistical analysis. Taking sample training takes into account the phenology of rice plants (phases of paddy crop) using references to appropriate Landsat imagery. Classification is done by the time series algorithm, while the accuracy is calculated with Kappa coefficients from the 1: 5000 paddy field map reference. The results found that SAR data in dual polarization (HV, VV), and Polarization Index (PI = 1-NDPI) can be used to detect paddy field. The best overall accuracy was obtained from the Min, Max, and Mean of PI 87%, Mean VH polarization 78%, Standard Deviation VH polarization 76%, and Range polarization VH 74%.