Research of Forest Fire Prediction Method

Transcription

1 Vol. (ICE ), pp Research of Forest Fire Prediction Method Dan Liu,Yanrong Zhang Information and Computer Engineering College, Northeast Forestr Universit, Harbin, China Computer and Information Engineering College, Harbin Universit of Commerce, Harbin, China Abstract. In order to achieve the predicted speed, high accurac, the use of simple purpose, forest fire prediction of the ke issues is to choose the main predictors. Currentl leading factor in the prediction of forest fire is often used in the fuel moisture, precipitation or dr das, relative humidit, temperature and wind five factors. In this paper, some of the data Yichun fire nearl a decade predict the forest fire meteorological data analsis, using multivariate linear regression to derive forest fire prediction method in the wireless sensor networks. Kewords: wireless sensor network, forest fire prediction, Multivariate Linear Regression Regional Overview Research Yichun is the northern of Heilongjiang Province in China. Geographic coordinates of longitude -, latitude 7-7, Area of, square kilometers. Territor is hill areas, thus extending the main vein Xiaoxing'anling to the southeast. It is the temperate continental monsoon climate, annual rainfall mm, with an average annual temperature of degree, in Januar average temperature of minus degrees, in Jul the average temperature of degrees above zero. Throughout the Yichun area, the area of the fire occurred Wuiling, Jiain and Hongxing, which happens to be in three places with frequent forest fire area. Table. Yichun forest fires occurrence Year Frequenc 9~9 7 97~ ~989 99~999 9 ~9 ISSN: 87- ASTL Copright SERSC

2 Vol. (ICE ) Analsis of Factors Affecting the Occurrence of Forest Fires Forest fires has more factors, such as the maximum temperature of the da, the da the humidit, Diurnal temperature range, wind speed, precipitation and their snthesis. This section discusses the contribution of meteorological factors on the occurrence of forest fires through a single evaluation model proposed b Dong Guangsheng.. The contribution of the dail maximum temperature on forest fire occurrence The dail maximum temperature on the impact of forest fires (formula ): r ( ( x )) x C x C r is determined b a single factor in determining the contribution of high fire risk value(%). x is the highest temperature of the da, where the temperature is : as the maximum temperature of the da. At : the temperature is low, especiall in the following 8 the risk of fire basicall no; temperature is between 8 and, the degree of risk of fire is graduall increasing; at between and, the rapid changes in the risk function, in the low to high transition phase fire should pa particular attention to possible fire; special attention will be above. (). The contribution of diurnal temperature range on forest fire occurrence The diurnal temperature range effects on forest fires (formula ): r ( ( ( x x ))) ( x x ) C ( x x ) C (x -x ) is the single factor reflected b the diurnal temperature high fire risk contribution value (%).x is the highest temperature of the da, and x is the lowest temperature of the da, where the temperature of : is the minimum temperature of the da. Under normal circumstances, when (x-x)<, cloud, rain and fogg weather phenomenon more, so difficult to fire; while (x-x) is between and, the great increase in the degree of risk of fire; when(x-x)>, weather controlled b high pressure situation, the performance of sunn, datime warming intense, afternoon the wind speed increases, the fire to maintain a higher state. (). The contribution of diurnal temperature range on forest fire occurrence The relative humidit affect the occurrence of forest fires (formula ): 8 Copright SERSC

3 Vol. (ICM ) r ( ( E )) E % E % r is the relative humidit reflects the contribution of a high fire risk value (%). E is the relative air humidit of :(%). When the relative humidit is greater than %, chance of a fire occurring is low. While between % and %, the ratio of high fire risk begins to increase. And when the relative humidit is below %, fires will occur at high risk. (). The average relative humidit of the air three das before on forest fire occurrence Three das before the average relative air humidit on the impact of forest fires (formula ): r ( ( EE )) E % E % r is average relative air humidit three das before a high fire risk reflects the contribution value (%). EE is three-da average relative air humidit at : (%). When the three-da average relative air humidit is greater than %, the lower the probabilit of occurrence of fire; when between % and %, the ratio of high fire begins to increase; relative humidit below %, the occurrence of the fire appear high risk. ().. The contribution of hours of precipitation on forest fire occurrence hours precipitation effects on forest fires (formula ): r x x mm x mm r is reflected in the hours of high fire risk precipitation contribution value (%). x is precipitation of hours. Changes in precipitation curve was smooth downward trend, with increasing precipitation, fire danger index declining. When rainfall is less than mm, ou are in a high fire danger and fire trend will not significantl decrease. At this time if precipitation decreased, it will again appear high fire condition. ().. The contribution of wind speed on forest fire occurrence The wind speed effects on forest fires (formula ): Copright SERSC 8

4 Vol. (ICE ) r ( (7 x )) x 7 m / s x 7 m / s r is the single factor reflecting the contribution of high fire risk value (%). x is as measured ground m~m height average wind speed at :. In the north the wind and precipitation generall occur simultaneousl, so the wind speed measurements to eliminate the influence of its precipitation. When the wind speed m/s or less, have little effect on the occurrence of forest fires; wind between m/s and 7m/s, the impact of forest fires started its rapid growth; when the wind speed reaches 8m/s, once the forest fires difficult to control. The main contribution of meteorological factors from forest fires occur above analsis point of view, there are certain factors that affect ever single occurrence of forest fires, but not the causes of forest fires can be one factor in the decision, but a number of result of factors working together. () The Achievement of Integrated Meteorological Indicators to Predict Forest Fire Method A single meteorological factor has poor predictive accurac characteristics in affecting the occurrence of the forest fires. The forest fire prediction requires the combined effect of a number of meteorological factors. After analsis of the previous sections, integrated meteorological indicators include the maximum temperature at :, diurnal temperature range, the average of three das minimum humidit and wind speed, in order to determine the method for predicting the occurrence of forest fires, according to a large forest fire occurred before the meteorological data, using multiple linear regression method to establish the occurrence of forest fires of linear equations, the four meteorological factors as independent variables linear regression equation. According to the analsis, the meteorological factors in forecast of forest fires have included the dail maximum temperature at :, the dail diurnal temperature range, the average of three das minimum humidit and wind speed, thus using the data in Yichun and regression models to predict the establishment of forest fires based on the above four factors regression equation 7. i representatives of fire danger rating, x is the maximum t temperature at :, x is the dail diurnal temperature range, x is the average of three das minimum humidit and x is wind speed. ( x x ) ( x x ) ( x x ) ( x x ) (7) i i i i i i () Calculating the sum, the arithmetic mean, the cross product of the variables. () Calculating the coefficient matrix A, the constant matrix B, and the inverse matrix A -. n A 7.7. L Copright SERSC

5 Vol. (ICM ) A i l B l l l i () Calculating the regression coefficients and regression equations. b b b A b b.. B ˆ.. ( x. ). ( x.799 ).87 ( x 8.9 ).7 ( x.9 ) From the calculations, forecast fire rating with linear regression equation is shown in Equation 8: ˆ.. x. x.87 x. 7 x (8) The regression after significant inspection meets normal distribution F table. Therefore there is a linear relationship between and x,x,x,x. The occurrence of forest fires can predict the fire danger rating, fire danger rating table as shown in Table : Table. Regression equation to predict the fire danger rating Fire danger Regression rating equation Degree of risk Forest flammabilit Ⅰ < No danger General nonflammable Ⅱ ~ Rarel danger After the fire spread ver slowl Ⅲ ~ Medium danger Quick flammable Ⅳ ~ High danger Spread faster Ⅴ > Extremel danger Fire was fierce and difficult to save Conclusion Since the meteorological factors on forest fires are an important role. So in this paper, the contribution of the various meteorological factors on the occurrence of forest fires is compared to select the probabilit of an impact on the larger fires factors as regression factor. According to a large meteorological data in front of forest fire has occurred, the use of multiple linear regression equation forecast fire danger rating. Copright SERSC 87

6 Vol. (ICE ) In comparison with similar prediction methods, the method than the relative humidit, temperature and humidit prediction method, and so comprehensive index method has higher prediction usefulness and predictive reference, suitable for promotion method according to different regions, has a good practical application value. Acknowledgements. This paper is supported b the Fundamental Research Funds for the Central Universities CBQ and Ke Laborator of Database and Parallel Computing, Heilongjiang Province. References. Kirsten Thonicke, Wolfgang Cramer. Long-term Trends in Vegetation Dnamics and Forest Fires in Brandenburg (German) Under a Changing Climate[J]. Natural Hazards. (-). Liping Di. Earth Observation Sensor Web:An Overview. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING.. Fengjun Zhao, Mingu Wang, Lifu Shu, Chuni Wang. Effects of climate change on forest fire dnamics progress [J]. Advances in Climate Change Research. 9(). Xiaorui Tian, Lifu Shu, Mingu Wang, Fengjun Zhao. Progress in forest fire and climate chang [J]. World Forestr Research. (). Hongang Zheng. Meteorological factors driving the fire sstem scale research[d]. Universit of Science and Technolog of China. 88 Copright SERSC