资源环境科技发展态势分析平台

Water resources are impacted by water‐energy balance fluxes at the land surface, most notably evapotranspiration (ET ), the largest component of surface energy balance. While integrated hydrologic models show promise in quantifying the non‐linear dynamics at this interface, the model results are plagued by parametric uncertainties. Given the high computational demand of running multiple parameter spaces of these models, little is known about how these uncertainties propagate into land surface processes. In this work, we perform a global sensitivity analysis by computing Sobol and AMAE indices using a surrogate model constructed with a polynomial chaos expansion to assess the impacts of the subsurface physical properties on ET . We do so by modeling a semi‐synthetic test case. Our results show that the effects of vertical hydraulic conductivity, porosity, and the water retention curve parameter Van Genuchten α mainly control ET . However, we note that while evaporation (E ) shows behavior similar to the ET with high sensitivity to the parameters controlling the flow in the unsaturated zone, transpiration (T ) is very sensitive to the saturated zone parameters and groundwater flow, especially during periods without rain. Our results also show that heterogeneities in land cover impact the flow in the saturated zone. Furthermore, our work demonstrates that reduced‐order models can be developed to make integrated models more accessible for rigorous sensitivity analyses and calibration purposes.