Towards a dual-polarisation radar vegetation index for Sentinel-1 for grassland monitoring

Optical vegetation indices (VI) have long been used for grassland monitoring in remote sensing. Studies with radar VIs have value for vegetation monitoring as well. Nevertheless, there are no studies available for the specific configurations of the Sentinel-1 radar satellite. Active radar remote sensing is independent of light and cloud cover, and thus has a higher temporal resolution than optical sensors. This is an advantage for intra-annual monitoring of grassland vegetation. Our current study investigates the potential of converting Sentinel-1 radar data to indices to represent vegetation development on grassland sites. For this purpose, we deploy different optical VIs from Sentinel-2 as a proxy for on-site biomass development. Overall, 8,420 grassland parcels in Lower Saxony are used in this study. Dates with simultaneous Sentinel-1 and Sentinel-2 pass-overs in 2018 with no cloud cover were considered. We calculate correlations between all radar and optical VIs to get knowledge of the possibility to use solely radar VIs for vegetation monitoring. Further, we include additional geo-data like soil information and a digital elevation model in a multiple regression and a random forest regression model. First results show R² between 0.17 and 0.52 for variations of radar to optical VIs depending on the method.