Abstract:

Based on daily meteorological data and forest fire information from 2006 to 2014 in Yanbian county of Panzhihua, this paper analyzed the correlation between meteorological conditions and forest fires occurrence, and screend out the main meteorological factors that contributed to forest fires occurrence. By applying the principal component analysis method of SPSS software, a meteorological grade model of forest fire danger was constructed. Forest fire data from 2015 to 2019 were used to test the model, which was expected to provide a basis for local forest fire danger prediction and serve the ecological construction of Panzhihua city. The results showed that forest fires in Yanbian county of Panzhihua city, mainly occured from January to June, with February to May being the months with the highest rate of forest fires. The meteorological factors that had a correlation with the occurrence of forest fires of Yanbian county, in descending order of correlation degree, were the average 10min wind speed, average 2min wind speed, average temperature, maximum temperature, sunshine hours, maximum wind speed and daily precipitation. Using the above 7 meteorological factors to construct a meteorological grade evaluation model for forest fire danger in Yanbian. The inspection found that the new mode had a 94.0% hit rate in predicting level 5 when a fire occurred. The hit rate of the new mode increased by 11.9 percentage points compared with the original mode, and the false alarm rate decreased by 20.2 percentage points. In terms of the normal distribution of meteorological grade predictions of fire risks at all levels, the new model was closer to the local objective reality and the accuracy in the results was higher. It can provide services for local forest fire risk forecasting.   

Key words: Forest fire danger, Meteorological conditions, Yanbian county, Fire hazard grade, Statistical model