Mapping Vegetation Water Content over Agricultural Landscapes Using Satellite C- and X-Band Synthetic Aperture Radar

Monitoring vegetation water content (VWC) over large spatial extents has become an important part of healthy ecosystem management, drought & wild fire risk assessment and precision agriculture. In this work we use two currently operational satellite synthetical aperture radar (SAR) systems, X-band KOMPSAT 5 and C-band Sentinel 1, to retrieve radar vegetation index (RVI) values over a wheat cropping field located in the state of Victoria, Australia. One quad-polarized RVI (KOMPSAT 5) and two types of dual-polarized RVIs (Sentinel 1 and KOMPSAT 5) were compared with the Normalized Difference Water Index (NDWI) derived Sentinel 2 as an indicator of VWC. The efficacy of RVI values in estimating VWC over wheat cropping field was examined at four different growth stages in May (sowing), June (Tillering), July (Stem Extension), and September (Heading) in 2019. Our results show that, for both Sentinel-1 and KOMPSAT-5, RVIs tested have a moderate skill to predict VWC when the comparison and evaluation were performed over multiple crop growth stages, but they present poor skill to predict VWC for a single snapshot of NDWI maps. This is in part due to the small range of VWC variation within the experimental field for each snapshot map. Quad-pol RVI outperformed dual-pol RVIs in predicting NDWI in multi-temporal application. However, all examined RVIs exhibited saturation for high NDWI conditions (NDWI > 0.5). RVI values also presented widely scattered variation for all examined growth stages. Sub-footprint-scale heterogeneity of the surface roughness and biomass, along with inherent noise of SAR, may contribute to high scatter of RVIs at individual snapshot level mapping.