The Research Of Forest Fire Occurrence Prediction Model And Fire Trend Prediction In Lijiang Area

With the global climate warming,forest fire number has increased in recent years,it has severe trends about forest fire prediction.The hot spots for researchers at home and abroad are how to forecast the fire those caused by forest fire and decreased the number of forest fire greatly.Yunnan Province as one of key forest productions base,which has rich forest resources,meanwhile it is a high risk area of forest fires.Based on data of forest fire statistics in Lijiang area from 1984 to 2017,meteorological data,climate scenario data,some statistic and spatial software,for instance,ArcGIS,SPSS,R and so on.The temporal and spatial distribution pattern of the forest fires and impact factors in Lijiang area are analyzed.At the same time,the forest fire prediction models are studied by the binomial model,Poisson regression model and negative binomial regression model respectively from day and month scale.Finally,to predict forest fire occurrence by combing climate scenario data in the next 13 years in Lijiang area.The results of the paper are very meaningful and can provide a scientific thesis foundation for forecast of forest fire and a theoretical reference for the effective implementation of fire management in Lijiang area.The temporal and spatial distribution pattern of the forest fires were analyzed by statistical analysis.The results showed that annual number of forest fires and burned areas varied significantly from 1984 to 2017 in Lijiang area,but the general trend was gradually declining.Forest fire occurred mainly in spring,in which the fires accounted for 72%of the total number of fires.On the spatial distribution,every county town had fires in Lijiang area,the most number of forest fires was Ninglang county.The fires caused by haze and smoking were the key component.Forest fire were mainly distributed in high-risk area between 1600 to 2800 m above sea level.When the altitude is greater than 2500 m.72.5%of forest fires concentrated on slope from 0° to 15°.Sunny slope is prone to fire.It was found 48%and 38%of forest fires on both sides of the highway and the river within the range of 2000m through the buffer analysis.Based on the logistic model,the relationship model between the occurrence of forest fires and topographic factors,vegetation factors and meteorological factors was established.The results showed that there was a significant relationship between altitude,slope,aspect,daily average temperature,daily average relative humidity and forest fire occurrence.The prediction rate of forest fire prediction model is the highest with 1:1.5 proportion.The accuracy rate of the logistic model is 86.9%.According to forest danger class,Lijiang area forest fires occurred mainly concentrated in the north of Yulong county,the south of Ninglang county and the central of Huaping county regions.The Poisson,zero-inflated Poisson,negative binomial and zero-inflated negative binomial model were chosen to model and analysis the relationship between human-caused fire and meteorological factors in Lijiang area,meanwhile comping those four models by the use of AIC and Voung methods to find out the best model.Which showed that zero-inflated negative binomial model was the best model for fitting forest fires data and meteorological data compared with other models.The number of forest fire was predicted by the use of the zero-inflated negative binomial model and climate scenario data in the next 13 years in Lijiang area.Which showed that future forest fires during 2018-2030 will present a rising trend.And severe fires probably occurred in the year of 2028.