Dynamics And Prediction Models Of Moisture Content Of Forest Fuels In QingAn County,Heilongjiang Province

The interface between plain and mountain area in Heilongjiang Province suffers from long-term human disturbance, causing forest severely damaged. Frequent human activities also increases fire hazard and leads to frequent forest fires in the region. Enhancing study on fuel moisture dynamics and prediction can improve the accuracy of fire danger rating. In this paper, the dynamics of the moisture contents of the surface dead fuel and living fuel were researched in a typical area (QingAn County, Heilongjiang Province), and three forecast model types using the meteorological factors, FWI factor and their combination as predictive factors were compared. The results showed that for different fuel moisture content ranges, the impact factor was different. The moisture content was influenced by humidity when it was bellow35%, but beyond which it was influenced by rainfall. Fuel moisture forecast models were established using weather variables, FWI indexes, and their combination as predictive factors, respectively. FWI indexes, mainly FFMC litter and surface dead fuel full range of moisture content, but with litter the surface dead fuel the moisture content of35%or higher relative difference. The index can be used to predict the surface dead fuel full range of moisture content of litter, but the error is greater than the meteorological elements regression method, is not suitable for predicting water cut after the rain Vapor exchange models have lower errors than FWI models for full range fuel moisture prediction but not for Litter the surface dead fuel<35%moisture prediction. Model accuracies were not improved when FWI indexes were incorporated. Vapor exchange models should be used for moisture prediction for litter layer all fuels in the region except<35%moisture prediction of fuel in Korean pine stand, which accuracy is MAE2.0%-7.8%, averaged5.4%, and MRE10.6%-28.1%, averaged15.8%。For the prediction of Litter the surface dead fuel moisture<35%in Korean pine stand, mixed variable models is the best. For the surface dead fuel half humus layer of moisture content, and a half coniferous forest humus layer moisture content FWI model should be adopted, It’s MAE is11.9%-24.5%on average,18.2%; MRE is5.7%-15.6%on average,10.7%; Half deciduous broad-leaved forest humus moisture content of meteorological elements should be adopted for the regression model, the MAE is26.7%-33.9%on average,30.3%; MRE is17.3%-an average of17.718.1but it can only explain the moisture content of18.5%to19.3%. For the surface dead fuel humus layer water content, except for the larch reunite other surface dead fuel prediction to water cut of humus layer meteorological elements should be used regression model, The MAE is18.5%-47.2%on average,35.1%, MRE is8.2%-24.7%on average,18.2%; Larch forest surface dead fuel humus layer FWI model for predicting water cut in MAE is13.7%MRE is6.4%.For living fuel moisture content, birches, Mongolian oak forest and larch forests with FWI model, evergreen coniferous forest living moisture content had no obvious seasonal change, deciduous forest living fuel moisture content changes with the seasons.