Study On Layout Optimization Method Of Flood Monitoring Rainfall Station Network In Small Watershed Of Mountainous Area

Short-term heavy rainfall is the main inducing factor of flash flood disasters.Accurately obtaining rainfall information is very important for flash flood disaster prediction and early warning.However,due to the strong spatial heterogeneity of rainfall in mountainous areas,whether the layout of the ground rainfall station network is reasonable or not is particularly critical to obtain accurate rainfall data and early warning information release.At present,the planning and optimization of rainfall station network mainly adopts the pumping station method and watershed hydrological model method in the density analysis method of rainfall station network,and most of them are aimed at large and medium-sized basins.However,there are often no observation data or few observation data in small and medium-sized basins.The layout of surface rainfall monitoring stations is determined by empirical methods,which makes the monitoring data have great uncertainty and brings difficulties to the formulation of flash flood warning information.In this paper,the Chabagou watershed is taken as the research object.Based on the measured rainfall data of ground rainfall stations and PERSIANN-CSS satellite grid rainfall data,the edge entropy and joint entropy between ground rainfall stations and satellite rainfall grids are analyzed by information entropy theory.Based on the maximum information minimum redundancy criterion(MIMR),the density and location of ground rainfall stations in the watershed are determined.Finally,the reliability of satellite data is verified by the ground measured station network data of Chabagou basin,which provides a theoretical basis for the layout of ground rainfall station network in the future implementation of non-engineering measures planning for mountain flood disasters in the ungauged areas of the Loess Plateau.Through the research,the following main conclusions are drawn:(1)Based on the rainfall-runoff data and HEC-HMS hydrological model of Chabagou watershed,the runoff process from 2014 to 2018 was simulated and analyzed.The results show that the relative error of flood peak flow is less than 14%,the relative error of flood volume is less than 15%,and the deterministic coefficients are 0.789,0.857and 0.920 respectively.The relative error of flood peak flow in the verification period is less than 6%,the relative error of flood volume is less than 12%,and the deterministic coefficients are 0.750 and 0.785 respectively.The HEC-HMS model has a good simulation effect in the Chabagou watershed and can be used for related research in the Chabagou watershed.(2)Based on the measured rainfall data of Chabagou watershed,16 rainfall events were selected to calculate the area rainfall deviation coefficient C_V and the rainfall non-uniformity coefficientη,and the spatial distribution of rainfall was studied.The average C_V of 16 rainfall events was 0.73,and the averageηwas 0.50.The rainfall distribution was extremely uneven.Then,by setting 8 different rainfall station network layout situations,the runoff process under different layout scenarios is simulated,and the flood peak flow,flood peak time and flood volume in flood season are sorted out,and the comparative analysis between each scenario is carried out.Firstly,the flood peak flow,flood peak time and flood volume of simulated runoff in flood season of uniform layout scenario and optimized station network layout are compared and analyzed.The results show that the peak time of the two scenarios is consistent,the error of flood volume is within 5%,and the error of flood peak flow is within 8%,and the error is very small.Then,the flood peak flow,flood peak time and flood volume of flood season simulated runoff in 2003-2010 of scenario 1-scenario 6 are compared with scenario 7 and scenario8,respectively.The results show that the peak time of all scenarios is consistent,the flood volume error is within 10%,and the flood peak flow error is within 18%.However,the flood volume error and flood peak flow error of uneven layout are generally greater than those of uniform layout.The error between the same density station network layout scenarios will also be due to the unreasonable layout position.The error is greater than the station network layout scenarios of different densities.(3)The 24 h and 6 h time scale rainfall information entropy of satellite grid rainfall data in Chabagou watershed is calculated,and the edge entropy,joint information entropy and mutual information of each station are calculated.Based on the maximum information minimum redundancy(MIMR)criterion,the central station and station sorting are determined.Through threshold calculation,the station set is determined as the optimal station network set.The results show that the 24-hour time scale of the Chabagou watershed is ranked as 11、1、15、4、5、9、7、13、10、2、8、14、3、12 and 6.After 9 iterations,the threshold reaches 95.06%,that is,the rainfall information of the 9 stations can reflect 95.06%of the overall rainfall information in the Chabagou watershed.Under the 24h time scale,the optimal rainfall station density in the Chabagou watershed is 22.8 km~2/station.The stations on the 6h time scale are ranked as 11、1、14、4、5、13、7、9、10、2、8、3、15、12 and 6.After 11 iterations,the threshold reaches95.42%,that is,the rainfall information of 11 stations can reflect 95.42%of the overall rainfall information in the Chabagou Basin.On the 6h time scale,the optimal rainfall station network density in the Chabagou Basin is 18.6 km~2/station.The reliability of the satellite grid rainfall product data is verified by comparing and analyzing the location and site ranking of the satellite rainfall station and the ground rainfall station on the 24-hour time scale after optimization.The satellite grid rainfall data before and after the optimization of the Chabagou watershed were input into the HEC-HMS hydrological model to simulate the runoff process of the rainfall station network before and after the optimization.The results show that the runoff simulation results before and after the optimization of satellite grid rainfall data rainfall station network in Chabagou basin are basically the same.Six runoff data were randomly selected for analysis.It was concluded that the flood volume error of the six simulated runoff before and after optimization was less than 5%,with an average error of 2.21%,and the flood peak flow error was less than7%,with an average error of 1.81%.