Description and application of a distributed hydrological model based on soil–gravel structure in the Qinghai–Tibet Plateau

Abstract. The Qinghai–Tibet Plateau, known as the “Asian Water Tower”, has a thin soil layer with a thick gravel layer underneath. Its unique geological structure, combined with widespread snow and frozen soil in this area, profoundly affect the water circulation processes of the entire region. To thoroughly study the water cycle mechanism of the Qinghai–Tibet Plateau, this study considered the geological and climatic characteristics of this area and selected the Niyang River Basin as the study area. The Water and Energy transfer Processes in the Qinghai–Tibet Plateau (WEP-QTP) model was constructed based on the original Water and Energy transfer Processes in Cold Regions (WEP-COR) model. This model divides the single soil structure into two types of media: the soil layer and gravel layer. In the non-freeze–thaw period, two infiltration models based on the dualistic soil–gravel structure were developed based on the Richards equation in non-heavy rain periods and the multi-layer Green–Ampt model in heavy rain periods. During the freeze–thaw period, a hydrothermal coupling model based on the continuum of the snow–soil–gravel layer was constructed. This distributed hydrological model can dynamically simulate the changes in frozen soil and flow processes in this area. The addition of the gravel layer corrected the original model’s overestimation of the moisture content of the soil layer below the surface soil and reduced the moisture content relative error (RE) from 33.74 % to −12.11 %. The addition of the snow layer not only reduces the temperature fluctuation of the surface soil, but also works with the gravel layers to revise the original model’s overestimation of the freeze–thaw speed of the frozen soil. The temperature RE was reduced from −3.60 % to 0.08 %. In the non-freeze–thaw period, the dualistic soil–gravel structure improved the regulation effect of groundwater on flow, stabilizing the flow process. The maximum RE at the flow peak and valley decreased by 88.2 % and 21.3 %, respectively. In the freeze–thaw period, by considering the effect of the snow–soil–gravel layer continuum, the change in the frozen soil depth of WEP-QTP lags behind that of WEP-COR by approximately one month. There was more time for the river groundwater recharge, which better shows the “tailing” process after October. The flow simulated by the WEP-QTP model was more accurate and closer to the actual measurements, with Nash > 0.75 and |RE|