Comparative Evaluation Of Precipitation Forcing Data Over The Upper Heihe River Basin And Runoff Responses Using Noah LSM

The Heihe River basin (HRB) is a typical inland river basin in the arid and semiarid region of northwest China. With naturally limited water resources and irrational water utilization, water-related problems have become severe in HRB. To solve such problems, a number of studies on hydrology and water-heat process of HRB have been conducted in past years. However, no high-quality precipitation forcing data is available in HRB to support accurate modelling activities, and the data qualities of existing forcing datasets remain unclear. Meanwhile, long time scale, high temporal-spatial resolution simulations on the upper stream of HRB with a land surface model are rarely found in literature. Using Noah LSM coupled with a routing scheme to simulate2D water-heat processes in Heihe upper stream could be an effective way to detect the performance of simulation, and could also be a reference case of HRB upper stream water-heat simulation study. It is significant in land surface model and hydrological model research, and the model coupling method.In this paper, we use three typical land surface model forcing datasets to force the coupled Noah LSM-VIC routing model. These datasets are WRF, GLDAS and ITP. Before simulations, a comprehensive comparative evaluation approach, consisting of a set of statistical indicators from both temporal and spatial aspects and comparisons against in situ observations from four established meteorological sites in the upper HRB, was proposed in this study. Then land use/land cover data, soil type data and other parameters were used to run the coupled Noah LSM-VIC routing model. The simulation is between2004-2009, with time resolutions of1hour and3hours, and a spatial resolution of5km. Simulation results include water-heat variables and runoff at outlet. Main findings are concluded as follows:(1) Each dataset has its own advantages. GLDAS and ITP are recommended for long-term simulations. Except for that, GLDAS, ITP and WRF precipitation data should be carefully evaluated before simulation. In this study, ITP data have showed best hydrological simulation performance.(2)Using forcing data produced by WRF and precipitation data from GLDAS and ITP, best simulation of the water-heat process in HRB upper stream can be achieved, especially in simulation of snow cover change, soil temperature and moisture.(3) NSE of the runoff simulation could reach0.66, which show that the coupled approach is effective in evaluating the performance of simulation.(4) Precipitation data plays an important role in simulation. Many items of precipitation data including magnitude, timing, and spatial distribution pattern could impact the water-heat and runoff simulation results.