Optimizing a backscatter forward operator using Sentinel-1 data over irrigated land

Abstract. Worldwide, the amount of water used for agricultural purposes is rising, and quantification irrigation becoming a crucial topic. Because limited availability in situ observations, an increasing number studies focusing on synergistic use models satellite data to detect quantify irrigation. The parameterization large-scale land surface (LSMs) improving, but it still hampered by lack information about dynamic crop rotations, or extent irrigated areas, mostly unknown timing On other hand, remote sensing observations offer opportunity fill this gap as they are directly affected by, hence potentially able detect, Therefore, combining LSMs through assimilation can optimal way This work represents first necessary step towards building reliable LSM system which, future analysis, will investigate potential high-resolution radar backscatter from Sentinel-1 improve quantification. Specifically, aim study couple Noah-MP running within NASA Land Information System (LIS), with observation operator simulating unbiased predictions over lands. In context, we tested how well modelled soil moisture (SSM) vegetation estimates, without simulation, capture signal aggregated 1 km Po Valley, important area northern Italy. Next, together simulated SSM leaf index (LAI), were optimize Water Cloud Model (WCM), which represent experiments. WCM was calibrated scheme considering two different cost functions. Results demonstrate that using provides better calibration WCM, even if estimates inaccurate. Bayesian optimization shown result best system, minimal chances having error cross-correlations between model observations. Our time series analysis further confirms track impact human activities cycle, highlighting its irrigation, moisture, via assimilation.