Soil moisture profile estimation under bare and vegetated soils using combined L-band and P-band radiometer observations: An incoherent modeling approach

Effective water management in agriculture requires a comprehensive understanding of the distribution of water content throughout the soil profile to the root zone. This knowledge empowers farmers and water managers to make informed decisions regarding irrigation timing and quantity for optimizing crop growth. To estimate the soil moisture profile, this study utilized combined L- and P-band radiometry with four incoherent radiative transfer models, including three multi-layer models based on a zero-order (IZ), first order (IF) and incoherent solution (IS) approximation, and a uniform model (UM) model, as well as the stratified coherent Njoku model (NM). The impact of vegetation was considered through the conventional tau-omega model. Linear (Li) and second-order polynomial (Pn2) functions were used to represent the shape of the soil moisture profile. Observations from a tower-based experiment under various land cover conditions, including bare, bare-weed, grass, wheat and corn, were used. The root mean square error (RMSE) was calculated between the observed and estimated soil moisture profiles. The results revealed comparable RMSE values for all five radiative transfer models, with the Pn2 function outperforming the Li function in estimating the soil moisture of deeper layers. Regardless of the employed radiative transfer model, utilizing combined L- and P-band radiometry and employing the Pn2 function yielded RMSEs of 0.03 m3/m3, 0.08 m3/m3, and 0.1 m3/m3 over depths of 0–5 cm, 0–30 cm, and 0–60 cm, respectively. A comparison between the incoherent and stratified coherent Njoku radiative transfer models indicated that the latter slightly outperformed the former under the dry bare soil conditions, exhibiting a 0.003 m3/m3 lower RMSE at the surface while nearly equal performance at the bottom of the profile. Furthermore, the multi-layer incoherent radiative transfer models provided only a slightly better estimate than the UM model, especially for shallow layers, with the average RMSE over the entire profile being 0.002 m3/m3 lower. Consequently, the complexity of the multi-layer incoherent and coherent radiative transfer models is not justified for this small gain in performance. The depth for which the UM model provided a reasonable soil moisture estimate ranged from 1 cm (under wet corn) to 39 cm (under dry bare), and depended on the soil moisture profile gradient and soil moisture content values in the shallow layers. These important findings pave the way for estimating soil moisture profile on a global scale using combined L- and P-band radiometry from future satellite missions.