Global sensitivity analysis and calibration of parameters for a physically-based agro-hydrological model

Efficient parameter identification is an important issue for mechanistic agro-hydrological models with a complex and nonlinear property. In this study, we presented an efficient global methodology of sensitivity analysis and parameter estimation for a physically-based agro-hydrological model (SWAP-EPIC). The LH-OAT based module and the modified-MGA based module were developed for parameter sensitivity analysis and inverse estimation, respectively. In addition, a new solute transport module with numerically stable schemes was developed for ensuring stability of SWAP-EPIC. This global method was tested and validated with a two-year dataset in a wheat growing field. Fourteen parameters out of the forty-nine total input parameters were identified as the sensitive parameters. These parameters were first inversely calibrated by using a numerical case, and then the inverse calibration was performed for the real field experimental case. Our research indicates that the proposed global method performs successfully to find and constrain the highly sensitive parameters efficiently that can facilitate application of the SWAP-EPIC model.