Evaluation of runoff simulations from Noah-MP land surface models across High Mountain Asia from 2002 to 2010.

This study evaluates modeled runoff in the High Mountain Asia region from 2002 to 2010 at a spatial resolution of 1-km. Specifically, we assess and compare monthly averaged runoff simulations against ground-based total runoff measurements across six gauged basins using 1) different Noah-MP land surface models, i.e., without or with a glacier representation; 2) different precipitation input forcings, i.e., from either Climate Hazards Group InfraRed Precipitation with Station Data, Version 2 (CHIRPS) or the Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2), and 3) different data assimilation schemes, i.e., without or with a snow cover assimilation framework. Goodness-of-fit statistics (i.e., bias, root-mean-squared error, and correlation coefficient) and statistics calculated from flow duration curves (e.g., relative biases in high-flow, mid-flow, and low-flow segments) are both used to assess the model performance with different configurations mentioned above. The inclusion of a glacier component or a snow cover assimilation scheme in the Noah-MP model is found to be beneficial to 1) improve peak runoff estimates, and 2) reduce the systematic errors in flow magnitudes within both high-flow and mid-flow segments for snow-fed (or snow-glacier-fed) basins. Further, the superiority of the Noah-MP model with a glacier modeling routine in simulating flow duration curves is more evident in the low-flow segments, which is possibly due to a more accurate representation of the base flow originating from snow and glacier melt. In addition, MERRA-2 precipitation forced runoff estimates outperform those forced by CHIRPS in terms of lower systematic errors and less temporal variability mismatch, while CHIRPS-forced total runoff estimates are generally with lower random errors. However, when compared with another independent regional precipitation product, namely, the Asian Precipitation - Highly-Resolved Observational Data Integration Towards Evaluation, MERRA-2 behaves more like an "outlier" (relative to CHIRPS) in terms of the monthly precipitation anomaly trends.