Combining Sentinel-1 and -3 Imagery for Retrievals of Regional Multitemporal Biophysical Parameters Under a Deep Learning Framework

Regions with excessive cloud cover lead to limited feasibility of applying optical images monitor crop growth. In this study, we built an upsampling moving window network for regional growth monitoring (UMRCGM) model estimate the two key biophysical parameters (BPs), leaf area index (LAI) and canopy chlorophyll content (CCC) during main period winter wheat by using Sentinel-1 Synthetic Aperture Radar (SAR) Sentinel-3 images. imagery is unaffected cloudy weather has a wide width short revisit period, organic combination will greatly improve ability at scale. The impact different types SAR information (intensity polarization) on estimation BPs was further analyzed. UMRCGM optimized correspondence between inputs outputs, it had more accurate LAI CCC estimates compared three classical machine learning models, highest accuracy green-up stage wheat, followed jointing heading-filling stage, lowest found milk maturity stage. accuracies were slightly higher than that first stages while lower This study proposes new method (especially CCC) combining large differences in spatial resolution under deep framework.