Paddy Rice Identification by Blending Time-series SAR and Optical Data

Abstract:

In Asia, rice is a staple cereal crop and Asian countries are responsible for approximately 90% of the world rice production and consumption. Asian space and agriculture related agencies launched an Asia-Rice Crop Estimation & Monitoring (Asia-RiCE) component for the GEOGLAM initiative. However, there are some difficulties to monitor rice crop by satellite imagery, which is different from other crops: 1) rice is mainly cultivated in rainy season and a large amount of cloud cover limits rice crop monitoring by the optical sensors; 2) adequate rainfall and temperature enable to cultivate rice two or three times a year in the tropics; 3) each field size is small compared to croplands in Western countries; and 4) water related agricultural disasters such as flood and drought are frequently occurred. To overcome these issues, C-band backscattering coefficient from RADARSAT-2 and reflectance data from MODIS were blended to estimate rice planted area over the West Java, Indonesia. By blending SAR and optical data, roughness and spectral information can be effectively used to differentiate paddy rice from other landcovers. The methodology using multi-wavelength data including optical (visible to thermal infrared) and SAR (X/C/L) would be a promising way for monitoring paddy rice in terms of the accuracy improvement and also the operational use to meet the requirement of observing the whole country with high-revisit frequency. In particular, the combining C-band SAR with other frequency SAR data such as L-band SAR onboard the ALOS-2 would be a challenge.