Automated calibration of Noah-MP land surface model for improved irrigation representation in the North China Plain

Irrigation significantly impacts the terrestrial water and energy cycle. Yet, due to insufficient agriculture management data, omitting important processes in models, and suboptimal model configuration, realistically representing irrigation processes in land surface models remains challenging. In this study, we employed the Noah-MP land surface model with a more realistic irrigation map and irrigation parameterization to replicate irrigation consumption over the North China Plain. For irrigation map, three different irrigation area maps are evaluated using prefecture-level census data. The best-performing, the Global Map of Irrigation Areas (GMIA), was chosen as the input of irrigation area to Noah-MP. Based on a widely-used calibration method, the Shuffled Complex Evolution method developed by the University of Arizona (SCE-UA), we constructed an automatic calibration framework for Noah-MP. The census data of irrigation water amount at the prefectural level was used to calibrate three key parameters in the flood irrigation module, namely, irrigation triggering point, flood irrigation loss, and flood application rate factor. Results show the calibrated irrigation parameters help mitigate the underestimation of irrigation water amount in the default simulation in good agreement with census data (R² = 0.71, N = 624). The calibrated irrigation simulation increases latent heat flux (+20.8 %) and decreases sensible heat flux (−34.5 %), and cools the land surface (−0.89 K), roughly proportional to irrigation amount. Furthermore, the additional irrigation water in calibrated simulation is partitioned into evaporation (58.5 %), followed by total runoff (39.8 %) and soil moisture storage change (1.7 %), significantly impacting hydrology, land–atmosphere coupling, and regional climate.