基于LH-OAT方法的SWAP-EPIC模型参数敏感性分析

Sensitivity analysis is of great importance to identify the key parameters and to increase the calibration efficiency of agro-hydrological model. In this study, a sensitivity analysis method, based on LH-OAT sampling approach, was developed and applied for sensitivity analysis of parameters in SWAP-EPIC model. A case modeling was conducted using SWAP-EPIC, with field experiment of spring maize in Yingke Irrigation District in the middle Heihe River basin in 2012. Twenty-six parameters in SWAP-EPIC were chosen for sensitivity analysis. The ranks of parameter sensitivity were obtained and classified as four different levels: highly sensitive, sensitive, less sensitive and insensitive. Results showed that the degree of sensitivity were obviously different relative to different output variables. Most of soil hydraulic parameters showed high sensitivity while the water stress parameters were less sensitive. Ten highly sensitive parameters were selected for calibration and the model was well calibrated. Overall, the proposed LH-OAT method for global parameter sensitivity analysis can greatly increase the efficiency of model calibration. It was also beneficial to the field application of agro-hydrological model.