Application of an improved distributed hydrological model based on the soil–gravel structure in the Niyang River basin, Qinghai–Tibet Plateau

Abstract

Abstract. Runoff formation and hydrologic regulation mechanisms in mountainous cold regions are the basis for investigating the response patterns of hydrological processes under climate change. Because of plate movements and climatic effects, the surface soils of bare lands and grasslands on the Qinghai–Tibet Plateau (QTP) are thin, and the soil below the surface contains abundant gravel. This characteristic geological structure, combined with snow and frozen soil, affects the water cycle in this region. To investigate the influence of the underlying surface structure on water–heat transport and water circulation processes on the QTP, a comprehensive study was performed combining water–heat transfer field experiments, and a water and energy transfer process model for the QTP (WEP-QTP) was developed based on the original water and energy transfer process model in cold regions (WEP-COR). The Niyang River basin, located on the QTP, was selected as the study area to evaluate the consistency between theoretical hypotheses, observations, and modeling results. The model divided the uniform soil profile into a dualistic soil–gravel structure. When no phase change was present in the ground, two infiltration models based on the dualistic soil–gravel structure were developed; these used the Richards equation to model a non-heavy rain scenario and the multilayer Green–Ampt model for a heavy rain scenario. During the freeze–thaw period, a water–heat coupling model based on the snow–soil–gravel layer structure was constructed. By considering gravel, the improved model corrected the overestimation of the moisture content below the surface soil predicted by the original model and reduced the moisture content relative error (RE) from 33.74 % to −12.11 %. The addition of the snow layer not only reduced the temperature fluctuation of the surface soil, but also revised the overestimation of the freeze–thaw speed predicted by the original model with the help of the gravel. The temperature root-mean-square error was reduced from 1.16 to 0.86 ∘C. In the fully thawed period, the dualistic soil–gravel structure improved the regulation effect of groundwater on flow, thus stabilizing the flow process. The maximum RE at the flow peak and trough decreased by 88.2 % and 21.3 %, respectively. In the freeze–thaw period, by considering the effect of the snow–soil–gravel layer structure, the freezing and thawing processes of WEP-QTP lagged behind those of WEP-COR by approximately 1 month. The groundwater simulated by WEP-QTP had more time to recharge the river, which better represented the observed “tailing” process from September onwards. The flow simulated by the WEP-QTP model was more accurate and closer to the actual measurements, with Nash–Sutcliffe efficiency > 0.75 and |RE| < 10 %. The improved model reflects the effects of the typical QTP environment on water–heat transport and water cycling and can thus be used for hydrological simulation on the QTP.