Font Size: a A A

Flood And Drought Disaster Risk Assessment Based On Text Data

Posted on:2021-04-27Degree:MasterType:Thesis
Country:ChinaCandidate:J T HuangFull Text:PDF
GTID:2370330611968166Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the intensification of the global warming trend,changes in hydrology and meteorology and ecology have led to an increase in the frequency and intensity of flood and drought disasters worldwide,which have caused more serious negative impacts on human society and the ecological environment.In this paper,Henan Province is used as the research area.Using historical text data of flood and drought disasters,text analysis and natural language processing(NLP)are used to analyze the time scale of flood and drought disasters in Henan Province.The spatial scale reaches the spatial regularity of the county level in Henan Province,and the flood and drought disaster risk assessment model and index system are constructed to obtain the Henan province's prefecture-level flood and drought disaster risk level.The research content is as follows:(1)Determining the Regularity of Flood and Drought Disasters Based on the Latent Dirichlet Theme Model.First,we need to tokenize and semantically search the text data of flood and drought historical materials,and build various dictionaries.Including stop word dictionary,flood and drought disaster proper noun dictionary,ancient and modern place name correspondence dictionary and flood and drought disaster level description dictionary,etc.Secondly,the topic model is used as the main method,and word frequency analysis and co-word analysis are supplemented to determine the main disaster levels in each disaster year.A time series of disaster levels is constructed to analyze the time pattern of flood and drought disasters.Taking Henan Province as an example,the research results show that,from 1804 BC to 3804 BC in Henan Province,there were droughts of varying degrees in 939,and the frequency of droughts was about four to five years;there were different degrees of 690 years.Floods,which occur about once every five to six years.(2)Determining the Spatial Regularity of Flood and Drought Disasters Based on the Method of Semantic Description and Sliding Window.In order to analyze the spatial distribution of flood and drought disasters,a method combining semantic description and sliding window matching was designed to determine the frequency of flood and drought disasters at each county level,and a spatial distribution map of flood and drought disaster frequencies was formed.Taking Henan Province as an example,the statistical results show that the frequency of drought in the northwestern region of Henan Province is greater than that in the southeast region.Flood disasters are common in the province,and the frequency of flood disasters in the northern region is greater than in the southern region.(3)Constructing flood and drought disaster risk assessment models to determine flood and drought disaster risks.In order to assess the risk of flood and drought disasters in the next year,the probability of flood and drought disasters on the time scale is obtained by combining the moving average method and the normal distribution,and the probability of flood and drought disasters at the county level on the spatial scale is obtained by the approximate probability of the law of large numbers.Combined with the construction of flood and drought disaster risk assessment models.The results show that the highrisk areas in Henan Province are concentrated in the northwest during drought disasters,and the high-risk areas in Henan Province are concentrated in the southwest during flood disasters.Through the comparison between the actual disaster situation and the flood and drought disaster analysis results from other data sources,the research results are verified.The actual situation is that from April to June 2019,drought occurred in Henan Province,Luoyang and Sanmenxia,etc.,and flood disaster occurred in Xinyang City,which proves the correctness of the research results.
Keywords/Search Tags:Flood and drought, Text data, Text analysis, NLP, Risk assessment
PDF Full Text Request
Related items