MAPPING RICE AREAS IN MINDANAO USING THE FIRST IMAGES FROM SENTINEL-1A: THE PRISM PROJECT EXPERIENCE

摘要

The Philippine Rice Information System (PRISM) project aims to develop a monitoring and information system for rice production in the Philippines using data from remote sensing, crop models and in-field crop surveys. PRISM predominantly uses multi-temporal Synthetic Aperture Radar (SAR) imagery to monitor rice growing regions in the Philippines. This study is a first attempt to use the first images of Sentinel-1A, a radar imaging satellite, for mapping rice areas. Sentinel-1 A Interferometric Wide Swath mode (IWS) 20 m resolution images acquired from October 2014 to May 2015 which coincides with the dry season rice cropping in the island of Mindanao in southern Philippines were used for mapping. This C-band time series SAR data were processed using a dedicated semi-automated processing chain including a knowledge-based rice detection algorithm implemented in MAPscape-RICE®, an image processing software developed specifically for rice. A total of 48 images from six footprints covering 98% of the land area in Mindanao were used for rice mapping. Over 381,000 hectares of rice area were mapped and the overall accuracy in one region based on 159 ground observations was 89% (kappa = 0.79). Planting months ranged from October to March with the peaks in either November or January depending on the region. The usefulness of Sentinel-1A and other SAR data sources, and the challenges in developing a SAR-based monitoring system at national scale are discussed.