| Alien Species have become one of the most important factors that threaten our country's biodiversity and ecological environment. Since Bursaphelenchus xylophilus intrudes into our country, areas of attacked forests over the country have reached 7.9×104 ha. This disease has some characteristics ,such as high spread speed, faster death rate, difficulty to control . At present there have not been effective methods against the disease. Therefore,the disastrous of invasive species-- Bursaphelenchus xylophilus is the forward and the key scientific problem.To solve this problem,precise prediction of disastrous areas and early-warning should be the focus.In this thesis ,remote sensing,GIS (Geographic Information System)and GPS technologies were used and combined with ground investigation and subcompartment data to extract basic geographic data as well as temporal series data .The four different prediction models were built, such as Grey theory , Markov chain and Regression analysis and ANN models. To improve the model's veracity and practicality,bi-variable analysis was used to analyze 10 factors including damage degree of Pinus Massoniana stand, altitude ,slope etc.The most related factors to damage degree with 95% significance were selected. Liner Regression equations have been established with as covariance and with the elements directly obtained from GIS and RS.Significance of each equation was tested and the result is useful.There is no significant deviation between data of actually measured and estimated by regression equation.Therefore, the regression equations can be extended and used in practice .Meanwhile ,the predictive model of BP neural network of damage degree of Pinus Massoniana stand by using the Neural Network Toolbos(NNT) based on MATLAB was made. The test that the data of historical occurrence level of pests in Fuyang County in Zhejing province was done,the result shows the fesibility of the predictive model.But in order to better effect,the model need improve further.
|
PDF Full Text Request