DTM And Artificial Neural Network Based Research On The Meteorological Risk Early Warning Model Of Geological Disaster And Its Application In Hefei

China is one of the countries with most severe geological disaster in the world，andthe disaster have the characteristics of variety, widespread and great harm. There are12times geological disasters happened in Hefei from2007to2013. Since most of themhad happened in no man’s land, so they caused fewer casualties, but caused a directeconomic loss about2.5million yuan. In order to reduce the people’s life safety isthreatened and property damage, it is necessary to study the early warning models ofgeological disaster.This paper take the mechanism of geological disasters in Hefei as the researchobject. According to the quantitative and qualitative analysis, considering the principleof disaster trend at the same time, the chart of geological disaster-prone area in Hefeihas been drawnby use the known geological disaster point and point information. Basedon the DTM and artificial neural network model, we have established meteorologicalrisk early warning model of geological disaster in Hefei, considering the factors, suchas precipitation, the position of the geological disaster points and prone to partition level,the level of prone to partition, and so on. We compared the two prediction results, anddiscussed the advantages and disadvantages of the two models. Based on the forecastwarning model mentioned above, we designed the system of meteorological risk earlywarning model of geological disaster and its prediction process. Our study results wereoutlined as follows.(1) We collected daily precipitation data from2007to2013, the geology informationof disaster point, the original administrative map of Hefei, Chaohu and Lujiang, etc.,and we created all kinds of database with the data mentioned above and mapped thenew administration of Hefei. The database and the maps provide important data base forthis research and the risk meteorological forecast warning of geological hazard inHefei.(2) The comprehensive analysis of Hefei geological structure, topography, climateconditions and precipitation factors shows that precipitation is the main factor ofinducing geological disasters. According to the geological disaster-prone partition map,Hefei is divided into one high prone to geological disasters area, seven medium prone togeological disaster areas, one low prone to geological disasters area and three no proneto geological disasters areas. (3) Based on the precipitation data from1952to2000, and combined with thegeological disasters case in the history of Hefei, the thresholds of the current day and5days precipitation in different geological disaster areas of Hefei was determined. On thebasis of constructing the DTM geological hazard meteorological warning and riskforecast model of Hefei,we successfully applied it to the geological hazardmeteorological risk forecast and early warning of Hefei and engineering practice, whichhad good effect on forecast warning, and the forecasting accuracy is65%.(4) In order to build the meteorological risk early warning model of geologicaldisaster based on the BP neural network, we comprehensively consider the X, Ycoordinates of meteorological observatory, precipitation and geological disaster proneareas and other factors. Using relevant geological disasters data and precipitation data totraining the model, the accuracy of training is81.25%. We use the trained model toforecast, and the result of the forecast accuracy is more than40%.(5)The early warning model based on DTM can fully considers many factors suchas landform and physiognomy. But due to less number of precipitation monitoringstation and the previous geological disaster data with poor quality in Hefei, the emptyrate is big in the forecast result, the forecast accuracy is low. The early warning modelbased on artificial neural network has strong ability of learning and analysis. But,beacause of the sample size of geological disaster is small and poor quality, though themodel has high accuracy of training, the prediction result is not good.(6) Adopting the idea of modular design, we reasonably designed the systemarchitecture, the main functions, the interface of the Hefei GIS-based geological hazardmeteorological risk and early warning system architecture. And according to the actualneeds of geological disaster forecast warning of Hefei, we designed the geologicalhazard meteorological forecast warning process, and the process achieved good effect inpractical application.