Research On Prediction Model And Risk Zoning Of Forest Firess In Liangshan Prefecture,Sichuan

Forest fires is the main disturbing factor of the forest ecosystem,which will not only destroy the dynamic balance of the ecosystem,but also affect the living environment of human beings and cause huge economic losses to human society.Liangshan Prefecture is located in southwestern Sichuan with abundant forest resources and it′s forest coverage rate is at the forefront of the country.Liangshan Prefecture is an area where forest firess frequently occur in China and is a prevent area for forest firess.Large-scale forest firess occur every year,which have an important impact on forest ecosystem,ecological environment and social economy.Studying the spatiotemporal distribution characteristics and driving factors of forest firess in Liangshan Prefecture,predicting the occurrence of forest firess and stratifying forest fires risks are the keys to improve the level of prevent disaster in Liangshan Prefecture,provide the basis for decision-making for prevent fires,and predict forest firess,provide theoretical support for the management of the forest fires in Liangshan Prefecture,protecting forest resources and the safety of people.This paper takes Liangshan Prefecture as the research area,analyzes the spatiotemporal distribution characteristics of forest firess in Liangshan Prefecture,which using MOD14A1 data from 2005 to 2015,combines weather data,terrain data,vegetation data,and infrastructural data to analyze the impact of driving factors on forest fires,Quantitative analysis of the driving factors of forest fires,the use of logistic regression model to establish a forest fires prediction model in Liangshan Prefecture,and the risk zoning of forest fires in Liangshan Prefecture.The main research conclusions of this paper are as follows:(1)MOD14A1 data can identify firess and other thermal anomalies,and can be used to extract regional forest fires spots.Combined with land use types,non-forest fires spots can be eliminated,which significantly improves the accuracy of selecting fires spot.MOD14A1 data can extract the fires spots,which is of great significance to study the forest fires.(2)In this paper,the spatio-temporal distribution characteristics of forest firess in Liangshan Prefecture are studied in detail.The temporal variation mainly study the variation characteristics of the number of forest firess on the interannual,seasonal and monthly scales.The spatial distribution mainly analyze the distribution of forest firess in the administrative division of the study area,use Geo Da software to complete the spatial autocorrelation analysis of forest firess.The results show that the interannual changes of forest firess in Liangshan Prefecture from 2005 to 2015 showed fluctuations as a whole.Forest firess mainly occurred in March,April,May in spring and January and February in winter.Huili County,Yanyuan County and Dechang County is frequent places for forest firess,and forest firess are aggregative distributed in space.Dechang County is a hot spot for forest firess.(3)This paper uses the date of fires points,terrain data,meteorological data,vegetation data and infrastructural data in Liangshan Prefecture,to extract the driving factor of forest fires,analyzes the relationship between the number of forest firess and the driving factors,and use Geo Detector to analyze the influence of each driving factor on the forest fires in Liangshan Prefecture,research the interaction between different driving factors.This paper analyzes the driving factors affecting forest firess,and discusses the main factors.(4)Based on the logistic regression model,a forest fires prediction model in Liangshan Prefecture was established to realize the prediction of forest fires occurrence.Carry out multi-collinearity diagnosis and variable screening on the driving factors of forest firess,determine the driving factors that meet the requirements of model establishment,and establish a prediction model of forest firess in Liangshan Prefecture,it used for forest fires prediction.(5)Using the prediction results of the prediction mode,the forest fires risk grading of Liangshan Prefecture was realized.The results showed that the central and southern Liangshan Prefecture were high risk areas,and the northeast were low risk areas.According to occurrence of forest fires from 2005 to 2015,it has verified the reliability of the division of risk classification in Liangshan Prefecture,and provided a basis for decision-making for preventing forest firess in the region.