| At present,the coupling of land and atmosphere is the main means to prolong the forecast period and improve the forecast precision in the hydrologic forecast.The rainfall data from numerical atmospheric model and the hydrologic model of watershed are all multi-scale and how to select the appropriate scale of land-atmosphere coupling is the key to obtain good simulation results.On the one hand,high-precision rainfall simulation data can be obtained by constructing a small scale land-atmosphere coupled model,but the simulation effect is affected by the accumulated error in the high-precision hydrological model calculation.On the other hand,although it is impossible to obtain high-resolution rainfall information by constructing large-scale land-atmosphere coupled models,when rainfall distribution is in uniformity,large-scale hydrologic models can better generalize watershed conditions to obtain good simulation results This study analyzes the simulation results of different scale models in different rainfall field by constructing the land-to-air coupling model at different scales and exposes the relationship between the coupling scale and rainfall type in the study area.Firstly,it selected four typical rainfall processes in Fuping and Zijingguan watersheds with different rainfall processes being simulated by WRF model and 12 sets of physical parameterization schemes being constructed.The most suitable physical parameterized combination scheme was selected for each rainfall process and the simulation results were analyzed in time and space according to the grid point scale and watershed surface scale.It is found that the performance of the physical parameterization schemes for different rainfall events was different.When the rainfall time and space distribution was in uniformity,the simulation effect of the model was better and when the rainfall time and space distribution was not in uniformity,there were some errors in the output results of each scale.The simulation effect of the rainfall data from WRF model was better than that from the end of the rainfall period at the early stage of the rainfall period.In general,WSM6 was superior to Lin in point evaluation and surface evaluation.In the cumulus convection scheme,GD scheme and KF scheme were similar in the effect of area rainfall evaluation and the GD scheme was more stable in the evaluation results of point rainfall.The evaluation results of YSU scheme of boundary layer scheme were better than that of MYJ.According to the result of comprehensive evaluation,the scheme was selected by 3.6.1.7 respectively as rainfall 1,2,3 and 4 to be the physical parameterized scheme combination.Secondly,the order of calculation between grids was obtained by encoding each grid by a proposed coding method,based on the rain flood model of Hebei Province,the grid model of loose grid structure was constructed.Considering the influence of the lateral flow of groundwater on the grid soil water content,the process of runoff generation between grids was generalized and the grids were divided into river grid and non-channel grid,.The relationship between soil topographic index,soil water storage capacity and infiltration capacity was studied and analyzed.The discretization of soil water storage capacity and infiltration capacity was realized by establishing the soil topography index and their calculation methods.The fitting degree of the two methods was similar by comparing the discrete results with the simulation results of lumped model,which shows that the discrete method is effective.Finally,it constructed three scales of coupled model of 1km × 1km,3km × 3km and 9km X 9km through the research on the simulation of WRF model of rainfall and grid based hydrological model and analyzed the results of lumped models and simulation results.The results show that for different rainfall,coupled model the performance of three scales were not the same.The simulated peak flow error was 7.60%in 9km × 9km grid in rainfall number 1 and its Nash efficiency coefficient was 0.8302 with the optimal simulation results.The rainfall volume error was-0.12%in the grid of 3km X 3km in rainfall number 2 and its Nash efficiency coefficient was 0.8353 with the optimal simulation results.The flood peak flow error was-3.69%in the 1km × 1km grid in rainfall 3 and the simulation result was the best.The volume error was-5.88%in 1km×1km grid in number 4 rainfall and its Nash efficiency coefficient was 0.8986 with the best simulation results.Through the analysis of the relationship between rainfall types and simulation results,it is found that the difference between the simulation results of lumped model and that based on grid model was increasing with the spatial distribution of rainfall becoming more and more uneven.For rainfall 1 and 2,the simulation results of lumped model and 1km × lkm,3km × 3km and 9km × 9km grids were not the same.The Nash efficiency index was better than that of raster model.As the spatial distribution of rainfall became more and more uneven,lumped model can only simulate the overall situation of runoff,but can not describe the runoff process in different periods.Overall,in the study area,when the rainfall spatial distribution was in uniformity,simulation results of 3km ×3km grid was the most stable and was superior to lkm × lkm and 9km × 9km grid simulation and results between different scale did not show higher accuracy or better simulation trend.When more and more uneven spatial distribution and simulation results of 9km × 9km grid began to decline,simulation of 1lkm × 1km and 3km × 3km grid results were significantly better than the 9km × 9km grid.When the space was not uniform,the grid three scale simulation results show that the higher accuracy of simulation,the better the trend. |
PDF Full Text Request