基于逆频率比采样方法的洪水易发性评价

打开文本图片集

关键词：洪水易发性;机器学习模型；采样方法；逆频率比；伊河流域中图分类号：P954;TV882.1 文献标志码：A doi：10.3969/j.issn.1000-1379.2025.08.016引用格式：，，.基于逆频率比采样方法的洪水易发性评价[J].人民黄河，2025，47（8）：96-101，108.

Assessment of Flood Susceptibility Based on Inverse Frequency Ratio Sampling Method

SUN Jiahui1，LIU Gengyuan²， ZHAO Xueqiang² （1.Henan Bureau Group Co.， Ltd. of China Chemical Geology，Zhengzhou 45Ooo0， China; 2.China Water Resources Pearl River Planning Surveying & Designing Co.， Ltd.， Guangzhou ， China）

Abstract：Idertomprovethuacyoffloodsusptibityodeling，tissudpropsedofdpntsampligmethodalld inversefrequencyratiosampling.Tismethodacountedforspatialdependenceinselectingon-flodpointsamples.TakingtheYieRier BasininCinaasaneample，wefistcomparedtheinversefrequencyatiosamplingmetodwithtotraditioalsmplingmethods（andom samplingandtratifedsampling）.Ten，thesemethodswereapliedintofoodsusceptibiltymodelingusingmacineleangmodelscludingRandomForest（RF）andMulti-LayerPereptron（MP）Teresusinicatetata）tespatialdistributioofon-floodpointsgeratedbytheinversefrequencyratioamplingmethodisferentfrothatgeneratedbyandomandstraifdsamplingexhbiingaivese gradientwithelevation.b）Teaccuracyofmachineleaingmodelsbasedonthinversefrequencyatiosamplingmethodisalhgherthan thatof stratifedsamplingandrandomsampling.FortheRFmodel，theAUCvaluesformodelsbiltbyusing inversefrquencyratiampling，stratifiedsamplingandrandomsamplingare0.97，O.94and0.90espectively.FortheMLPmodel，thevaluesareO.90，0.87and 0.86respectivelyTeproposedo-fodointsampligtodasimprovdteacuracyofloodsusceptiblitymodelingasdoachin learning to a certain extent.

KeyWords：flood susceptibility；machine learning models；sampling method；inverse frequency ratio；Yihe River Basin

0 引言

洪水是世界上最常见和最具灾难性的自然灾害之一，给全球各地造成严重的生命与财产损失[1-2]，如何减少洪水造成的损失是一个重要的研究课题。（剩余9837字）