Broad Scale Paddy Field Mapping Using Sentinel-1 and Sentinel-2 on Google Earth Engine

Abstract:

Paddy field is one of the major sources of methane (CH₄), the second influential greenhouse gas for global warming following CO₂. Therefore, high-accuracy broad scale paddy field maps are fundamentally important for methane budget estimations. However, only a few broad scale paddy field maps have been developed because of some difficulties. For example, in most case, rice cultivation periods overlap with rainy seasons. In our previous study, we proposed novel paddy field mapping method using Sentinel-1 SAR time series assisted by Sentinel-2 MSI on Google Earth Engine (GEE) and developed paddy field map for whole area of Japan in 2018 with 30 m spatial resolution. Our method was based on the seasonal variation of Sentinel-1 VH backscatter time series and improved the accuracy of extracting irrigated conditions by using an additional mask based on Sentinel-2 indexes (NDVI, EVI and LSWI) obtained during irrigated period. Our maps showed great potential for reproducing total paddy field areas at the prefecture-scale in Japan. However, rice cultivation method and cultivation environments vary greatly from region to region, it is important to confirm whether our method is applicable to mapping paddy fields in other areas for regional or global-scale paddy field maps. In this study, we developed a paddy field map for Monsoon Asia in 2018 with 30m spatial resolution by the method proposed in our previous study. By comparing with the statistics of paddy field area at country scale and existing paddy field maps, we examined the potential capability of our paddy field mapping method for the area including areas outside Japan.