A novel fire danger rating model based on time fading precipitation model - A case study of Northeast China

With the increase of extreme climate and global warming, forest fires have become more frequent. Therefore, it is important to accurately predict whether fires will occur in forest in the future. Precipitation is an important factor that affects the probability of the occurrence of forest fires in the future. Previous models selected annual average precipitation, monthly average precipitation or drought days as the precipitation value, which the attenuation of precipitation is not considered. In this study, a time fading model is used to calculate the comprehensive precipitation index, which is an exponential weight decay model. The earlier the precipitation time, the smaller the weight. This method can better represent the effect of precipitation in predicting the occurrence of forest fires. Moreover, in this study, discrete fire points are converted into a continuous fire-point density. The structure of the prediction model is more reasonable, which is conducive to obtaining higher precision prediction results. Besides, the SVM regression model was used to construct a forest fire danger rating model. In the same area, considering the comprehensive precipitation index compared with the average precipitation value, the accuracy of the three forest areas in northeastern China in the test set has been improved by about 5%. The accuracy rates of 90.13%, 93.04% and 87.5% can be achieved respectively.