Simulation of Sentinel-1A observations and constraint of water cloud model at the regional scale using a discrete scattering model

Abstract

Effective calibration of microwave scattering model and accurate simulation of backscattering coefficient (σo) at the regional scale are essential for improving soil moisture retrieval based on active microwave remote sensing. This study investigates the potential of simulating the Sentinel-1A observations and constraining the coefficients of the water cloud model (WCM) at the regional scale based on a theoretical microwave scattering model named Tor Vergata model. The Tor Vergata discrete scattering model is firstly calibrated at both site- and regional-scale to find a single set of model parameters for capturing the Sentinel-1A observations at the regional scale. Subsequently, the coupled WCM and Oh model is calibrated at the regional scale based on three strategies, i.e., using either the calibrated Tor Vergata simulations or the Sentinel-1A observations, or with the assumption that the σo only comes from the vegetation contribution during the peak growth period. The investigation is conducted in a Tibetan meadow ecosystem configured with a long-term regional-scale soil moisture monitoring network.The results show the optimized parameter values obtained for the Tor Vergata model are consistent with each other based on both calibration methods, leading to comparable performances in simulating well the Sentinel-1A observations and scattering components. These results demonstrate the transferability of model parameters under similar land condition and the feasibility of simulating the regional-scale σo based on a single set of model parameters using the Tor Vergata model. The σo simulated by the coupled WCM and Oh model based on the three calibration strategies are comparable to each other and well agree with the calibrated Tor Vergata simulations as well as Sentinel-1A observations. However, the traditional calibration method using the Sentinel-1A observations tends to produce unreasonable WCM coefficients and thus unrealistic simulation of vegetation transmissivity. Above deficiencies are largely addressed by other two calibration strategies, whereby the first one preserves the physical mechanism of the Tor Vergata model and reduces the computation cost, and the assumption made by the other strategy may provide a new sight to solve the unknown WCM coefficients. These positive results warrant further investigation for developing new parameterization of WCM to improve soil moisture retrieval at the regional scale based on active microwave remote sensing.