| Forest is the valuable wealth and renewable natural resource. Developing and protecting these resources dose not only have a direct relationship with the country's economic construction, but also between the ecological environment construction and protection. It isn't only benefit for contemporary but also the descendants. Forest fire is the greatest enemy of forest safety.Hunan has forest land 7534670 hectares, accounting for 35.7%of the total area of Hunan, and Forest coverage rate was 34.3%, well above the national average. Forest land accounts for 57.4%of the land area in Hunan, it's one of the important timber production base in southern China. Hunan is also densely populated, more severe drought in summer and autumn and form more forest fuel and fragile ecological environment. The impact of global warming, high temperature, drought, wind and other weather disasters increased significantly in Hunan Province, resulting in continued high forest fire danger rating, significant increased number of forest fires, affected area and casualties. Hunan forest fire prevention work is now facing the most severe situation in 70 years, the task is arduous.The paper overviewed the present researching status and development trend of forest fire model. In the guide of the theory of ecology,forest fire science and spatial modeling theory, it deeply investigated social economics,natural geography and forest fire prevention status of Hunan province. By means of geography information system, self-organized criticality of system, spatial interpolation, spatial clustering, computer technique and mathematical methods, the paper studied forest spatial modeling in Hunan.The results are following:(1) The "frequency-area" distribution of forest fires in Hunan Province is found to well satisfy power-law relation., and the power-law relation of "frequency-area" distribution is steady and scale-invariant. In the similar external conditons, they have the same or similar parameters of the self-organized criticality (SOC). Therefor, the distribution of future forest fire can be forecasted with the detail history forest fire in a region. The fitting line of fire data from 2000 to 2009 is log F=-1.17* log A+3.41, the fitting line from 2000 to 2009 is logF=-1.17*logA+3.37.(2)Use spatial interpolation techniques to samples of forest fire data of Hunan and transform point data into surface data for analyzing the spatial distribution of forest fires. This paper use spatial deterministic interpolation—inverse distance weighted (IDW) interpolation method, radial basis function interpolation (RBF), global polynomial interpolation and geostatistical interpolation—kriging interpolation to construct spatial model of forest fire data in Hunan province. Compare the three interpolation methods using cross validation, the effect under different interpolation model come to the conclusion:Kriging interpolation is the best and IDW interpolation is better than RBF interpolation. In Inverse distance weighted interpolation method, the weight 2 is the best. The ring model in kriging interpolation method is better than the spherical model. The best model in local interpolation is 0.53688* Circular(1.2078)+0.44541* Nugget,0.53688 is the sill value, the range is 1.2078, the nugget is 0.44541.Establish trend surfaces model on forest fire of Hunan with trend surface analysis methods separately in power 1,2,3, inspect the precision and appropriate by sequence. The best is power 3 model:Explain the regularities of forest fire distribution according to the trend and residual value, thus verify the accuracy of the law.(3) Use spatial clustering technologies to study the spatial modeling of forest fire of Hunan. Spatial clustering technology can quantitatively determine the degree similarity between forest fire data, thus classify the forest land without the interference of man-made factors. The result is, the frequency of Hunan fire has high clustering, the significance level of Getis-Ord general G statistical is above 5%, the Moran's I is above 1%. The clustering result is,The forest fire frequency in the line from north-central to the Midwest of Hunan Province is always above 68 in the 10 years, and rising with the increased total frequency annually; Midwest, Southwest became the high forest fire frequency region since 2002, the average fire frequency is above 237; In 2008, the frequency is reached a peak and up to 5076, Shaoyang an area is up to 1093 times; Followed in 2005, the forest fire incidence reached a peak since 2000, the total frequency of forest fires is 3138 times.A large amount of information, timeliness and etc. of forest fire modeling continuously make higher demands on the quality of the model. With the maturing of forest fire prevention work, forest fire modeling approach is also developing. Computer technology, especially the rapid development of geographical information technologies provides strong support for the realization of forest fire space modeling. Firing prevention is a complicated systematic project needs long-time and difficult work in Hunan, which is influenced by some factors such as society, environment and economy. How to develop realistic and effective forest fire spatial model, which embraces the features of Hunan province is the key to rapid development of forest fire prevention in Hunan. |
PDF Full Text Request