Optimisation of AquaCrop backscatter simulations using Sentinel-1 observations

In preparation for active microwave-based data assimilation into a crop modeling system, the mapping of daily 1-km AquaCrop model (v6.1) biomass and surface soil moisture to backscatter was optimised, using two forward operators, i.e. Water Cloud Model (WCM) Support Vector Regression (SVR). Both operators were calibrated (2014–2018) with Sentinel-1 (γ°) observations in VV VH polarisation, three different study domains Europe. For validation period (2019–2021), γ° simulations showed reasonable performances around Czech Republic Iberian Peninsula, good over Belgium, but strong variations within each domain. The domain-averaged root mean square difference between remained below 2 dB both all domains, bias close 0 dB, 0.5 polarisation. WCM SVR performed better than overall slightly vegetation WCM. Additionally, assumed linear relationship holds VH. remaining differences or are mainly caused by errors.