Quantifying the Interactions of Noah-MP Land Surface Processes on the Simulated Runoff Over the Tibetan Plateau

The quantification of uncertainties in runoff over the Tibetan Plateau (TP), simulated by land surface models (LSMs), is of paramount importance for effective water resources management within this region. However, the interactions of land surface processes on simulated runoff, where the effectiveness of one process depends on the chosen scheme for another, have rarely been studied. To address this gap, we conducted ensemble simulations with the Noah-MP (Noah with multiparameterization) LSM by varying the optional parameterization schemes of six land surface processes and quantified the sensitivities of the simulated runoff to these processes. Results showed that the simulated runoff over the TP was most sensitive to the RUN (runoff-groundwater) process. The interplay of RUN and FRO (frozen soil permeability) accounted for up to 30% of the variation in the annual mean surface runoff in the TP's permafrost regions. The interactions of RUN and VEG (dynamic vegetation) on summer and autumn subsurface runoff exceeded 10% in the southeast TP. In regions where these interactions among land surface processes significantly affected simulated runoff, we observed elevated model errors and reduced model controllability. Therefore, this study underscores the imperative need to categorize land regions based on the interactions of land surface processes as a foundational step toward enhancing the performance of LSMs. Prioritizing improvements in model physics should be particularly directed toward regions marked by high interactions. Land surface models (LSMs) are important tools for managing water resources on the Tibetan Plateau (TP), as they predict how much water will flow into rivers and streams. However, these predictions can be uncertainty because different LSMs use various methods to represent the complex land processes. The way one process affects runoff can depend on how another process is represented. However, these interactions between different processes are often unmeasured. We conducted experiments using the Noah-MP LSM, varying optional schemes for six land surface processes related to runoff. We applied a variance-based method to assess the interactions between these processes. The results showed that interactions between the RUN (runoff-groundwater) and FRO (frozen soil permeability) processes explained up to 30% of the variation in simulated runoff over the permafrost regions of the TP. Additionally, the VEG (dynamic vegetation) process significantly interacted with RUN and FRO in the grassland areas of eastern TP. In regions with significant interactions between these land processes, we observed increased model errors and reduced controllability, emphasizing the importance of understanding these interactions for accurate regional modeling. Sensitivities of simulated runoff in the Tibetan Plateau to the land surface processes are quanti?ed by a variance-based method The interplay of runoff-groundwater and frozen soil permeability accounts for up to 30% of the variation in annual mean surface runoff Improving model physics should receive precedence in regions where land surface processes are highly interacted