| Rainfall is the driving force of all the hydrological processes and its spatial and temporal variation affect the total flood volume, peak flow and the time of peak occurrence, and the uneven distribution of rainfall increases the uncertainty of the hydrologic model’s parameter estimation, so the accuracy of the rainfall data is a basis of quantitative hydrological analysis. As a kind of active remote sensor, radar can get real-time information of rainfall with a certain precision, a large area, high spatial and temporal resolutions, so the application of radar rainfall data combining with hydrological models to runoff simulation study has key research significance.In this paper, the HEC-HMS hydrologic model at the Xitiaoxi Catchment was built with two different methods to simulate the runoff and the relative error of peak flow, Nash-Sutcliffe coefficient and other indexes were used to evaluate the simulation results; then the rainfall data of the rain gauges at the Xitiaoxi Catchment were used to evaluate the data quality of radar one hour precipitation and the radar rainfall data was calibrated by multi-point average method; finally the simulation of HEC-HMS hydrologic model with three kinds of data entry programs such as the spatial interpolation of the observed rainfall data of rain gauges, the radar rainfall data and the radar rainfall data calibrated with the calibration factor for exploring the applicability of HEC-HMS hydrologic model at the Xitiaoxi Catchment was performed and the application effect of radar one hour precipitation product combining with the HEC-HMS model at the Xitiaoxi Catchment was evaluated. The main conclusions were as follows:(1) With one or two exceptions, both the actual measured discharge hydrograph and the simulated discharge hydrograph of two methods had fine consistency. In the first method, the qualified rates of the relative error of the flood volume, peak flow, and runoff depth were87.5%; the mean value of the correlation coefficient and Nash-Sutcliffe coefficient were0.95and0.83respectively. In the second method, the qualified rate of the relative error of the peak flow was87.5%and the qualified rates of the relative error of the flood volume and runoff depth were87.5%; the mean value of the correlation coefficient and Nash-Sutcliffe coefficient were0.94and0.77respectively. The time of peak occurrence of both methods was within3hours while the lagging peak appeared easily. The simulation results of events with multi peaks were worse than that of events with one peak.(2) The first method performed better than the second method in the qualified rate of the relative error of the flood volume, peak flow, runoff depth, but the second method performed better than the first method in the simulation of the time of peak occurrence. Taken together, the HEC-HMS hydrologic model had good applicability at the Xitiaoxi Catchment.(3) The MAR of the radar’s one hour accumulated precipitation decreased with the increasing of the hourly accumulated rainfall, and the CSI of the radar one hour precipitation increased with the increasing of the hourly accumulated rainfall. There existed overestimation when the hourly accumulated rainfall was less than or equal to1mm and existed underestimation when the hourly accumulated rainfall was more than1mm. The Pearson Coefficient of the radar’s one hour accumulated precipitation increased with the increasing of the accumulated time.(4) The calibration factor of each event was different because each rainfall event had different situations and the data quality of radar’s one hour accumulated precipitation was different. There existed a big error between the areal rainfall of the catchment calculated by the radar rainfall data not having been calibrated and that calculated by the rain gauges’ rainfall data with the Thiessen Polygon method, while the error between the area rainfall of the catchment calculated by the radar rainfall data having been calibrated and the area rainfall calculated by the rain gauges’rainfall data with the Thiessen Polygon method decreased.(5) When using the radar rainfall data calibrated to simulate the flood runoff, in both methods the relative error of the flood volume, peak flow, and runoff depth was within20%; in the first method the correlation coefficient was above0.9and the Nash-Sutcliffe coefficient was above0.7and the time of peak occurrence is less than3hours with easily lagging peak. In the second method the correlation coefficient was above0.8and the Nash-Sutcliffe coefficient was low and above0.5and the time to peak was less than3hours.(6) The actual observed discharge hydrograph and the discharge hydrograph simulated by the uncalibrated radar rainfall data in two methods had poor consistency and obviously the simulated discharge was less than the actual observed discharge, so the uncalibrated radar rainfall data couldn’t be used as a data entry of HEC-HMS hydrologic model. It was clear that the results became better after the radar rainfall data was calibrated, and the radar rainfall data calibrated could be used as a data entry of HEC-HMS hydrologic model. The HEC-HMS hydrologic model combining with the radar’s one hour calibrated rainfall data had good applicability at the Xitiaoxi Catchment. |
PDF Full Text Request