Evaluation and optimization of rainfall station network rationality for flood forecasting
[Objective]As observation sites designed for precipitation data collection,rainfall stations are an important source for collecting water information.A reasonable deployment of rainfall stations can reflect timely and accurate precipitation in the basin,essential information for flood forecasting.[Methods]Taking the Mi River basin in Shandong Province,China,as the research object,the rationality of the existing rainfall station network is evaluated based on the cone gradient method and Thies-sen polygon theory.Combined with entropy theory and Maximum Information Minimum Redundancy(MIMR)criterion,an optimization model was constructed using hourly-scale rainfall sequences from 22 surface rainfall stations in the basin,and an optimization scheme for the rainfall station network was proposed.[Results]The result show that some of the rainfall stations in the basin have too large a control area and some of them are poorly correlated with each other;the pattern of change in informa-tion entropy between rainfall stations indicates:with the change of the iteration times,the joint entropy gradually became larger and eventually stabilized,and the mutual information firstly increased and then decreased,and the redundancy was getting lar-ger and larger,and the ratio of the information of the selected stations to the total amount of information of all the stations reached 98％after 17 iterations.Considering the requirements of flood forecasting,setting up 5 rainfall stations at a reasonable location is proposed and 1 rainfall station is improved.[Conclusion]The optimized rainfall station network can better reflect the rainfall information of the basin.These research result can provide a theoretical basis for optimizing basin rainfall station network in the future.
entropyMIMRrainfall stationflood forecastingprecipitationclimate changetemporal and spatial distributionmutual information
万方数据
维普
