| The dynamic change of the moisture content of forest fuels is mainly affected by meteorological factors as well as its physical and chemical properties,the value of which value is very important for forest fire forecast and prediction.The direct drying method has the most accurate moisture content value during the previous period,which cannot be applied in practice.However,in practical applications,we need to obtain accurate and real-time moisture content of fuels.Therefore,forest fire researchers attach great importance to high-precision models for predicting the moisture content of fuels around the world.The three moisture codes in the Canadian Forest Fire Danger Rating System(CFFDRS)can reflect the dryness of fuels,which are semi-physical models and have good extrapolation.Meanwhile,the model form should not be complicated,and could be applied in practice.However,due to the different types of fuels,forest sites and climatic zones,the applicability of the moisture code needs to be verified before application.If it is not applicable,it needs to be corrected.If a moisture content prediction model based on the moisture code can be established in the study area,it will be of great significance to improve the accuracy of moisture content prediction and fire risk prediction.This study selected typical dead small and live fuels on the surface of Chayuanshan Forest Farm in Nanchang,Jiangxi Province as the research objects.During the fire prevention period of 2015-2016,the moisture content of fuels was monitored at 12a.m.every day as well as the meteorological data were recorded in the study area simultaneously.With meteorological data,dialy FFMC,DMC,and DC were calculated.With the total or part of the data(non-rainfall/rainfall)and based on meteorological elements and moisture codes,moisture content prediction model was established.Furthermore,the parameters in the FFMC and DC scale models were corrected,and the prediction models of the moisture content of fine dead fuels and live fuels on the surface were re-established to calculate the model accuracy and analyze the errors.The main results are as follows:(1)During the study period,the average moisture content of fine dead fuels on the surface of Chayuan Mountain ranged from 61.6-139.2%,and the dynamic change of moisture content was only significantly positively correlated to the relative humidity and rainfall of the previous day;The average change range was between 118.3-181.5%.The air temperature,relative humidity,and rainfall on the previous day had an extremely significantly positive correlation with the dynamic change of its moisture content.The minimum value of FFMC in the Chayuanshan Forest Farm during the fire prevention period was 0,the maximum value was 89.1,and the average value was 33.5.The minimum value of DMC was 0.1,the maximum value was 41.0,and the average value was only 5.4.The range of DC during the study period was 1.2-104.3,and most of the time it was lower than 20.7.The dynamic change of the moisture content of fine dead fuels on the surface was extremely significantly negatively correlated with the three moisture codes,and the effects of FFMC,DMC,and DC on it gradually weakened.The dynamic change of the moisture content of live fuels had an extremely significant negative correlation with FFMC;(2)Use all the data to establish a regression model of meteorological elements.The moisture content model of the fine dead fuels on the surface only adopted the relative humidity of the previous day.The MAE range of the model was 24.52-48.58%.For live fuels,the air temperature and relative humidity of the previous day were included.The MAE range of the model was 11.86-25.89%.Use all the data to build a moisture code prediction model.For fine dead fuels on the surface,except for Moso bamboo in sample spot 5,only FFMC entered the equation.The minimum value of the prediction model R2 was 0.2528.The maximum value was 0.3404,and the model MAE range was between 26.58-49.84%.In the predictions models for live fuels,FFMC is essential,and DMC is optional.Only sample spot 6 of thatch included DMC,and all live fuels' moisture content prediction models had MAE of 11.87-26.45%.Neither for fine dead fuels nor for live fuels on the surface,the moisture code was not applicable to be used directly to establish the prediction model within the scope of the total data;(3)Under non-rainfall conditions and rainfall conditions,the moisture content of fine dead fuels on the surface only increased significantly with the relative humidity of the previous day,and under non-rainfall conditions for live fuels,the dynamic changes of moisture content were extremely positively correlated with air temperature of the previous day.Under rainfall conditions,in addition to the air temperature,it was also significantly positively correlated to the relative humidity of the previous day.Under non-rainfall conditions,the dynamic change of moisture content,in most of the fine dead fuels and live fuels on the surface was only extremely significantly negatively correlated with DMC and DC.(4)Under rainfall conditions,all fuel types were not correlated to moisture codes.For fine dead fuels on the surface,under non-rainfall conditions,the meteorological factor regression model only included the relative humidity of the previous day.The MAE range of the model was 22.07-50.10%,and the MAE range of the moisture code model was 29.41-66.47%.Under rainfall conditions,the meteorological element prediction model included the relative humidity and wind speed of the previous day,the model MAE was between 22.81-42.08%,and the moisture code model cannot be established.For live fuels,under non-rainfall conditions,the meteorological element regression model only included the air temperature of the previous day.The MAE range of all models was 11.32-22.00%.The moisture code model mainly included DMC and DC.The minimum MAE of the model was only 10.91%,and the maximum MAE was 22.69%;Under rainfall conditions,the meteorological element prediction models mainly included the relative humidity of the previous day,and some of them included air temperature.The MAE range of all models was 10.54-51.11%,and the moisture code model cannot be established.Distinction between rainfall and non-rainfall,the moisture code was still not suitable for prediction of the moisture content of fine dead fuels and live fuels on the surface of the study area;(5)Based on the adjusted FFMC and DC,respectively,to predict the moisture content of fine dead fuels and live fuels on the surface,the errors of the prediction models were significantly lower than the uncorrected moisture code and regression models of meteorological elements.The MAE range of the prediction models for the moisture content of all fine dead fuels and live fuels was 2.15-6.90%and 4.86-8.92%,respectively.The adjusted moisture code can be used to predict the moisture content of fine dead fuels and live fuels on the surface of the study area,and the model error was within an acceptable range. |
PDF Full Text Request