The impact of land model structural, parameter, and forcing errors on the characterization of soil moisture uncertainty
Abstract. A sensitivity analysis is conducted to investigate the contribution of rainfall forcing relative to the model uncertainty in the prediction of soil moisture by integrating the NASA Catchment Land Surface Model (CLSM), forced with hydro-meteorological data, in the Oklahoma region. This study depicts different sources of uncertainty, namely, errors in the model input (i.e., rainfall estimates from satellite remote sensing observations) and errors in the land surface model itself. Specifically, rainfall-forcing uncertainty is introduced using a stochastic error model that generates reference-like ensemble rainfall fields from satellite rainfall products. The ensemble satellite rain fields are propagated through CLSM to produce soil moisture ensembles. Errors in CLSM are modeled with two different approaches: either by perturbing model parameters using the generalized likelihood uncertainty estimation (GLUE) technique or by adding randomly generated noise to the model prognostic variables. While the first method only addresses parametric uncertainty, the second one addresses both structural and parametric uncertainty. Despite this, a reasonable spread in soil moisture is achieved with relatively few parameter perturbations through GLUE, whereas the same ensemble width requires stronger prognostic perturbations with the standard random perturbation method. The probability of encapsulating the reference soil moisture simulation increases when the rainfall forcing uncertainty and the model uncertainty approaches are combined (compared with using only rainfall uncertainty). This improvement is more significant when using the GLUE technique to perturb CLSM parameters as opposed to perturbing the CLSM prognostic variables.