Research On Dynamic Changes And Model Prediction Of Moisture Content Of Three Surface Fuels After Rainfall

As a natural disaster,the occurrence of forest fires has an extremely important impact on the global ecosystem and social economy,and from the perspective of climate change trends,the number and extent of forest fires will increase.Effective management of forest fires requires technical support for accurate forecasting of forest fires.Fuel moisture content is one of the important factors in predicting forest fire risk level and affecting fire behavior change,and is affected by a variety of meteorological factors.Rainfall has a very significant impact on the moisture content of surface fuels,and the amount and duration of precipitation directly affect the amount of water stored on the surface of forests,resulting in changes in the moisture content of surface fuels.However,in the model research of the mainstream equilibrium moisture content method to predict the moisture content of fuels,temperature and humidity are mainly used as the driving factor.Therefore,in this study,Quercus mongolica、 Pinus sylvestris var.mongolica and mixed surface fuels in the interlaced zones of the two were selected as research objects,indoor simulated rainfall experiments were used to explore the influence of rainfall on the moisture content of surface fuels,random forests were used to rank the relative importance of the impact factors,and the modeling was carried out by combining the data.After comparing the accuracy of the model,the excellent model was selected,combined with the direct estimation method to correct the prediction model of the moisture content of fuels,and the modified direct estimation method was evaluated and compared with the unmodified direct estimation method,the convolutional neural network model and the meteorological element regression method using field data,and the conclusions were as follows:(1)In indoor simulated rainfall experiments,the moisture content of three fuels under rainfall conditions increases logarithmically over time;Rainfall and the moisture content of fuels before the last rainfall have a significant impact on the growth of moisture content of fuels;However,the compactness and initial moisture content are not significant;In the process of monitoring the moisture content of fuels in the field,the moisture content of surface fuels shows significant changes over time.(2)Temperature,humidity,rainfall,wind speed,and solar radiation are all important factors for the moisture content of surface fuels to a certain extent,but their performance varies among different types of surface fuels.In the correlation analysis between moisture content and meteorological factors,there is a significant positive correlation between humidity,rainfall,and combustible moisture content(p<0.01);There is a significant negative correlation between the air temperature on that day and the moisture content of surface fuels(p<0.05).(3)In the comparison of different prediction models of fuel moisture content in the field surface,the modified direct estimation method improves the accuracy of the fuel moisture content prediction model under rainfall conditions to a certain extent compared with the unmodified direct estimation method;At the same time,the comparison between different models shows that the modified direct estimation method has the best results,followed by the direct estimation method,convolutional neural network(CNN),and finally the meteorological element regression method.Therefore,the revised direct estimation method,direct estimation method and convolutional neural network model are suitable for the prediction of surface fuel moisture content in this region,and this study has certain reference value for different surface fuel moisture content prediction models,and is also of great significance for forest fire management in this area.