A Two-Step Method for Monitoring Annual Paddy Rice Planted Area in Tropical Region by Integrating Rice Identification and Cropping Intensity Estimation

Accurate assessment of paddy rice planted area is challenging in tropical regions where the availability of optical satellite data and samples is limited due to the high prevalence of cloudy and rainy periods. This study, carried out in Vietnam, presents a novel two-step method to estimate annual paddy rice planted area at the 10-m spatial resolution overcoming the limitations of scarce optical remote sensing data and ground-truth samples. First, an easy and effective visualization method was designed to collect rice and nonrice samples based on the significant color difference of consecutive three monthly VH composite data between paddy rice fields and other land covers during the flooding to transplanting stage. Second, rice spatial distribution information was extracted by taking full advantage of coarse optical imagery in a spatialtemporal continuum and Sentinel-1/2 images combined with rice potential zone (RPZ) generation by threshold and rice fine classification by random forest. Third, the cropping intensity of paddy rice fields was identified from monthly VH bands smoothed by harmonic analysis of time series (HANTS) using the quadratic difference method. The utilization of RPZ improved $R^{2}$ of the paddy rice fields’ identification accuracy from 0.78 to 0.97 and reduced the root mean square error (RMSE) from $69.28\times 10^{3}$ to $31.54\times 10^{3}$ ha. Results indicate that in 2020, the paddy rice planted area of Vietnam was estimated as $6264.9\times 10^{3}$ ha, strongly corresponding to the estimation in the statistical yearbook by province ( $R^{2} =0.97$ ). The study demonstrated the effectiveness of Sentinel-1 synthetic aperture radar (SAR) data in the estimation of annual paddy rice planted areas in frequently cloudy and rainy tropical regions where optical data are scarce.