Prediction model and change regularity of moisture of forest surface dead fuel in Beijing

Taking the dead fuel moisture content of different tree species as the research object，we researched the relationships between the fuel moisture of different types and weather conditions according to the daily observation of fuel moisture content. The prediction model of fuel moisture content for different fuel type of different species including Quercus variabilis， Q. aliena， Platycladus orientals and Pinus tabulaeformis was built by using non-linear regression. The results showed that there were obvious differences in fuel moisture content of different tree species and among different fuel types. Q. variabilis had higher dead fuel moisture content than other species. The moisture content of dead leaves was generally higher than that of dead branches. Fuel moisture content was negatively correlated with temperature and wind speed， and was positively correlated with relative humidity. The established non-linear regression model could explain more than 30% of variance for fuel moisture content of different species， and the prediction model performed well by the mean MRE and mean MRE test. Fuel moisture content showed a trend of firstly increasing and then declining and lastly increasing during a day. Fuel moisture content peaked at 7： 00-8： 00 pm， and reached the lowest value at 14： 00. During the period of fire prevention， fuel moisture content was relatively high from December to February， while it was relatively low from March to May and in November.