Fire Environment And Forecasting Models Of Summer Forest Fire In Daxing’an Mountain District

Although wildfires are recurrent natural phenomena and an important feature of many ecosystems,their occurrence is commonly perceived as a threat for human activities.Under the background of climate warming,the fires in the Daxing’an Mountains,a key stateowned forest zone in China,is increasing.In recent years,the number of summer fires has increased significantly.In this paper,the background of summer forest fires occurrence in Daxing’an Mountains and spatial-temporal distribution of summer forest fires were analyzed based on daily meteorological data and historical forest fires during the summer of 1967-2008.The quantitative of summer forest fire environment was made on meteorological elements,vegetation elements,terrain,slope,aspect and altitude combined with digital elevation model,vegetation distribution map and vegetation index datasets by using Arcgis and Python software.The models suitable for fire forecasting in Daxing’an Mountains,Logistic model and Zero-inflated negative binomial model,were selected through literature search to establish a prediction model for the occurrence of summer forest fire in Daxing’an Mountains.The results showed:The main reason for the increase in the summer forest fires is climate change,and the global warming has caused significant climate changes in this region.From 1967-2008,the average temperature and maximum temperature showed a significant increase with 0.23℃/10 a and 0.27℃/10 a,especially after 2000,the average and maximum temperature were significantly higher than annual average and maximum termperature.The average relative humidity and precipitation in the summer showed a decreasing trend in the whole.During 1989-1993,there was a lot of precipitation but no fire occurred.Moisture index was comprehensive index of dryness and humidity.The slip curve of summer moisture index in the past 42 years showed a downward trend,and it indicated the climate became drier.Lightning-caused fire was dominated in the summer forest fire of this study,accouted for 77.7%.There were obviously interannual changes in the total fire area and forest burned area.The fire dynamic indexs in summer of 200-2008 were higher than that in 1967-1988,while there was no fires in 1989-1999.For monthly scale,the number of fires and fores burned area were the highest in June.According to the Julian data of fire occurrence,around June 20 th was the time when most of summer forest fires occurred.The time for fire danger period in the summer extended back,and frequent fires began to appear in August after 2002,while summer forest fires only occurred in June and July before 2002,which was closed related to the increase of temperature,the decrease of humidity and the decline in precipitation after 2000.Summer forest fires were mainly concentrated in the northern and central part of the region,and the most fires occurred in Mohe county,and Huzhong district and Huma county were more severely affected regions in terms of total burned area and forest bunred area.Average emperature and maximum temperature had a significant effect on summer forest fire(P<0.01),which mianly occurred in the range of average temperature of 17-21 ℃and maximum temperature of 28-32℃;The wind speed that influenced summer fire was concentrated in 1.5-2.5m s-1,and there were significant differences in the numbers of fires in different wind speed ranges(P<0.05).The main ground temperature that affected summer fires occurrence was about 21-30℃;the average relative humidity of 60-70% and the minimum relative humidity of 20-30% were the main rage of summer fires occurrence,and there was a significant difference in the number of forest fires in different ranges,P<0.01 and P<0.05,respectively.The precipitaton had a significant negative correlation with the number of forest fires.When the precipitation was 0mm,the probability of forest fires was the highest,accounting for 75%,in which,the numbers of lightning-caused fires and human-caused fires were also the greatest,accouting for 71% and 52% in the total number of lightning fires and human-caused fires,respectively.The average pressure was mainly concentrated in the range of 945-965 hPa,and whether the summer forest fires occurred depended on certain conditions for average pressure on the day of fire and previous two days.The range of sunshine hours affecting fires was 12-14 hours,the probability of summer fires improved gradually with the increase of sunshine hours.But when sunshine hours were more than 14 hours,the probability deduced suddenly.The summer forest fires had the highest probability in the flat slope where the number of lightning fires and human-caused fires was also the highest,accounting for 53% and 77%,respectively in total number of lightning fires and human-caused fires.The probability tended to decreas as the slope increased.Lightning strikes had a slope ranging from 0° to 30°,while human-caused fires ranged from 0° to 17°,hence,the slopes where lightning fires occurred were generally greater than human-caused fires.The number of summer fires in sunny aspect is 3.3% higer than that in shady aspect,in which,lightning fires occurred on the sunny aspect were 25% higher than shady aspect,and 4% higher in semi-sunny aspect than semi-shady aspect.In the altitude of 550-650 m,the highest probability of summer fires occurred.The summer fire mainly occurred in the deciduous coniferous forest,accounting for 73%.In this vegetation type,the number of lightning strikes were also the most,accounting for 77% of the total number of lightning stikes,followed by meadows,accounting for 15%.For Logistic regression model,the area of ROC curve of building data was 0.886,P<0.001,and the correct rate of validation data was 79.0%,so the model had a better fitting and forecasting results.For Zero-inflation negative binomial model,the correct rate of validation data was 71.0%,and it had a good forecasting on overmuch zero,while poor forecasting on daily fires with more frequent occurrences.