Phenological object-based paddy rice mapping in Sungai Burong Selangor Malaysia using Sentinel-1 data

Abstract

Rice is a vital staple food in Malaysia. Consequently, accurate mapping of rice fields is essential to support food security goals and inform government policy on production and trade. Mapping rice areas in tropical regions is challenging due to frequent cloud cover during the transplanting phase. Additionally, high-resolution pixel-based mapping struggles in fragmented landscapes, leading to inaccuracies and salt-and-pepper noise in depicting actual land cover within specific parcels. This study aims to develop a phenological object-based method to collectively map paddy field extent in Sungai Burong, Integrated Agriculture Development Project (IADA) Barat Laut Selangor (BLS), using cloud-free time-series data from Sentinel-1 Synthetic Aperture Radar (SAR). The phenological object-based approach was applied to produce 10-meter resolution maps of rice field extent and seasonal land cover changes over a two-year period of 2021 and 2022. Validation was conducted using field survey data and very-high-resolution street view images from Google Earth and, achieving an overall accuracy of 91.82% a kappa coefficient of 0.79. The results demonstrate the proposed method's effectiveness in producing high-precision rice extent maps. This approach offers potential for broader application across Malaysia and other tropical regions, providing a valuable resource for addressing food security challenges.