Research On Models Of Fire Danger And Fire Behavior In Pinus Yunnanensis Forests

Pinus yunnanensis forests belongs to Southwest forests of China,its special climate and geographical environment combined common fire source make forest fire frequently and seriously occur in this area,so it is urgent to develop fast and accurate forest fire prediction in this area.This thesis took one of main tree species in Yunnan forests--Pinus yunnanensis as research objects,started from two kinds of models,fire weather danger forecasting model,which suit for quick and easy calculation,and fire behavior simulating model,which can provide high accuracy of prediction results,collected the mainstream models both at home and abroad,collated the historical forest fire data,historical weather records,vegetation data and topographic data in this area,calculated the prediction accuracy of each models in this area.Based on evaluating and screening the calculated results,this thesis analyzed and summarized the main reasons of influencing the accuracy,and the guiding ideology for improvement.At last specific improvement was put forward on the basis of relative decisions.Research results can be used to offer scientific proof for choosing forest fire prediction model in Pinus yunnanensis of Yunnan forests.In addition,the construction of fire danger prediction model and the development of fire telemetry instrument in small area can also be used to provide technical support in forest fire forecast for this area.Specifically,contents,methods and decisions are contained as following:(1)Fire weather danger forecasting models is the main technical method for forest fire prediction,and its calculation is quick and easy as purely from meteorological view,so it is very suitable for use in basic level units.Due to most of the popular domestic fire weather danger forecasting models are structured for northeastern forests,southwestern forests is in urgent need for the selection and building of this kind of models for itself.So this study used all of the popular domestic fire weather danger forecasting models in China,calculated them with the actual fire data of Qujing Forestry Bureau Qilin Fire Office from 2013 to 2015 and gave the evaluation finally.The results shows that,“National Forest Fire Danger Weather Rating Model” got the highest Kappa accuracy coefficient,which was 0.308,the second highest is “Multi-factor Method” of 0.233.Meanwhile,the EC accepted error coefficient of “National Forest Fire Danger Weather Rating Model” is 0.788,lower than “Double Index Method”,“801 Method” and “Temperature-humidity Method” of 1.000,and “3 Factors Method” of 0.981.This results indicated that,“National Forest Fire Danger Weather Rating Model” is the most accurate model and it suit for forest fire weather danger forecasting in Pinus yunnanensis forests.(2)According to the forecasting error of “National Forest Fire Danger Weather Rating Model”,step problem of results,inaccurate results because of the direct use of meteorological station data,lengthy calculation process not conducive to rapid calculation of fire danger,this thesis presented a modified model to overcome these issues,which is named “Continuous forest fire weather danger forecast model”.Compared with the traditional model,this model improves the prediction accuracy.Further,based on this new model,this thesis proposed the method of continuous forest fire weather danger forecasting with the analysis function of GIS software,and a portable forest fire danger forecasting instrument using the technique of Android and embedded systems.This forecasting instrument solves the problem that the former forest fire weather danger prediction cannot highlight the locally significant fire zone and field staffs in the past cannot detect and calculate the fire danger in real time.(3)As an important technical reference for efficient prevention and fighting on forest fire,forest fire behavior parameters are mainly obtained from fire behavior simulating models in some developed countries.Fire behavior simulating model requires a comprehensive meteorological,combustible,terrain and other factors to simulate fire behavior,compared with the results of forest fire weather danger forecasting,the accuracy is improved a lot.The prediction results of this kind of model reflect both fire danger and fire behavior,can provide decision support for the prediction of fire and fire after disposal.So fire behavior simulating model has better practical value than fire weather danger forecasting model.This thesis selected two famous simulating models,the Farsite from USA and the Prometheus from Canada,which are both widely used in local area.Through comparing simulated results and relative data of the “3.29 Fire” occurred in Anning city,Southwestern China,we tried to evaluate accuracy of the simulators in a quantitative way.The results indicated that the precision of Farsite under Scott fuel model was the highest,while Prometheus was the worst,but the difference was not significant.The different simulation area of fire spread between Farsite and Prometheus mainly concentrated in the distribution area of Pinus yunnanensis;in the simulation of spread rate,the mean ROS results of Farsite under both two fuel models were closed to the actual situation,while results of Prometheus were far away from the actual situation.The different simulation area of ROS between Farsite and Prometheus mainly concentrated in the distribution area of Pinus yunnanensis;in the simulation of fireline intensity,the mean FLI results of Farsite under both two fuel models were similar,and Prometheus obtained significantly different FLI results from Farsite,and the different simulation area mainly concentrated in the distribution area of Quercus.