Mapping paddy rice agriculture over China using AMSR-E time series data

This study investigates the potential of AMSR-E surface soil moisture (SSM) time series data to distinctively identify paddy rice pixels and map paddy rice cultivated area at national scales. Taking China as the test site, we first established and applied a paddy rice planting index (PRPI) to each AMSR-E 25 km pixel over the entire country during the period 2003–2011. A mathematical model was then constructed for the retrieval of paddy rice cultivated area at the AMSR-E pixel scale using data obtained from PRPI as input, and a high coefficient of determination (R2 = 0.86) was recorded at the national scale. Provincial scale validation of the modeled paddy rice fraction shows a high correlation coefficient (R) of up to 0.94. In a subsequent step, the inter-annual variations in AMSR-E retrieved paddy rice areas were evaluated with annual agricultural census data, and an R-value of 0.93 was recorded. A good performance of the proposed retrieval method was also recorded when validated against MODIS paddy rice distribution images, and in which, five out of the seven selected validation sub-regions, had R-values larger than 0.85, while R-values of up to 0.70–0.75 were recorded in the other two. The results demonstrate that the proposed method is effective in mapping paddy rice agriculture at national scales, and therefore, the applicability of the PRPI algorithm on mapping paddy rice at continental and global scales is worth investigating.