Research On Moisture Prediction Of Forest-Floor Fuel

In order to select the monitoring objects effectively for wireless sensor network nodes and predict the moisture of forest-floor fuel precisely, and then provide theoretical basis for the establishment of forest fire real-time monitoring system based on wireless sensor network technology and accurate forecast of forest fire risk, experiments were carried out in this paper to measure the sample moisture of sumac leaves, pine needles, grasses, small twigs, arborvitaes and mixed fuels, soil moisture(SM)within the 10cm underground, air temperature (T0, T10, T30, T50) and relative humidity (RH0, RH10, RH30, RH50) 0,10,30 and 50cm above the floor. EXCEL and SPSS software were used to carry out correlation analysis and regression analysis. The main conclusions were summarized as follows:(1) The most significant factor correlated to forest-floor fuel moisture was relative humidity, followed by soil moisture, and finally air temperature; air temperature and relative humidity more closed to the floor had more significant effect on fuel moisture;(2) Based on soil moisture, air temperature and relative humidity (0cm), the moisture prediction model of mixed fuels in woodland was FM=0.031RH02-1.661RH0+0.091T02-5.299T0-0.279SM2+7.991 SM+62.843 (R=0.746). Prediction error of the model was 8.001% and accuracy was 78.615%;(3) In winter plot, the moisture of sumac leaves, arborvitaes and mixed fuels were also correlated to cumulative time (MJDt) positively and significantly, while the moisture of pine needles, grasses and small twigs were correlated to day hours (t) negatively and significantly;(4) The moisture prediction models of different fuels in summer and winter plot were established separately, and all models had passed through the significant F-test and were extremely significant; winter models (0.691% of prediction error,83.627% of accuracy) were more accurate than summer models (7.731% of prediction error,74.418% of accuracy); winter models can meet the accuracy requirement of fuel moisture forecast, while summer models need further improvement.