Assessment of runoff simulation in the Yarlung Zangbo River Basin based on the multi-physics Noah-MP land surface model

Yarlung Zangbo River Basin (YZRB) is one of the highest basins in the world. Runoff in YZRB plays an important role in hydropower production and fresh water supply for millions of people in Asia. Land surface models (LSMs) are important tools for coupled hydrometeorological simulation and forecasting in this region. However, accurate runoff simulation over YZRB presents huge challenges due to the uncertainties in meteorological forcing and the highly complex runoff generation mechanisms in the basin (e.g., snowmelt, glacier melt, permafrost freeze/thaw, groundwater, and monsoon runoff). In this study, we systematically assess the runoff simulation skill of a physically-based land surface model named the multi-physics Noah land surface model (Noah-MP) over the YZRB. Runoff from 2000 to 2018 is simulated at a spatiotemporal resolution of 5 km and three hours forced by the Chinese Meteorological Forcing Dataset (CMFD). The simulated runoff is evaluated at four sub-basin outlets (i.e., Lazi, Nugesha, Yangcun and Nuxia) with respect to the Nash-Sutcliffe efficiency (NSE) and percentage bias (PBias) skill metrics. Our study features the assessment of ten physical parameterization schemes that are the most influential to runoff generation in the YZRB; they include: (1) Precipitation-related parameterization scheme (PPS) that concerns the precipitation partitioning into rainfall and snowfall, (2) evapotranspiration-related parameterization schemes (EPS) that describe the canopy stomatal resistance, soil moisture factor controlling stomatal resistance, surface drag, radiative transfer processes, and (3) runoff-related parameterization schemes (RPS) that describe snow albedo, frozen soil permeability, supercooled liquid water in frozen soil, glacier, runoff processes. Our results showed that the default parameterization scheme by Noah-MP only achieves a monthly NSE and PBias of 0.23 and -35.79% at Nuxia. Turning to the third option of precipitation partitioning (PTP_3), the modified two-stream radiative transfer parameterization scheme (RAD_1), and the BATS runoff parameterization scheme (RUN_4), the model performance is satisfactorily improved with PBias reduced to -23.36%, 3.85%, and -17.19%, and NSE increased to 0.37, 0.58, and 0.60 for Nugesha, Yangcun, and Nuxia, respectively. Based on the above results, runoff simulation was conducted using the combination of these schemes, and the monthly NSE can be improved to 0.89, 0.87, and 0.81 at Nuxia, Yangcun, and Nugesha, respectively, while Lazi showed a relatively low NSE (-0.06), which may be explained by the different runoff generation mechanisms there. Overall, our result demonstrated that optimizing the parameterization selection of Noah-MP can achieve satisfactory runoff simulation skills over YZRB without conducting model parameter calibration. Such good simulation skills in a calibration-free modeling experiment highlight the promising potential of a physically-based model such as Noah-MP. We anticipate this study to provide a useful reference to future Noah-MP users/developers in conducting coupled hydrometeorological simulations in this region.