Design And Implementation Of Prediction Model On Forestry Pest Biology Disaster

Forest pest biology disaster has serious effect and harm to the forest resource.In this paper,based on the analysis on the affecting factors of the occurrence of forest pest biology disaster,a ANN-CA Forest Pest Spatial prediction Model was build,combined with cellular automata and artificial neural networks,which was encapsulated using C#.Net. And this model was used to predict occurrence area of Dendrolimus punctatus(Walker) disaster and occurrence region of Hlyphantria cunea(Drury) disaster.The main research contents and results in this paper are as follows:(1)The main factors affecting the occurrence of forest pest biology disaster include: biological characteristics,stand condition,stand structure,climatic factors and the time. Biological characteristics include life history,life habit,morphological characteristics, distribution and damage characteristics;stand condition includes altitude,slope,slope position and aspect;stand structure includes tree species,stand layer structures,stand age, crown density and kinds of distance;climatic factors include temp,humidity,precipitation, sunshine hours and other combined climate factors.(2)An ANN-CA Forest Pest Model was build combined with cellular automata and artificial neural network.The core parts of cellular automata,including cell state,lattice, neighbor,rule and time,were extended.Combined with kinds of spatial influencing factors, the artificial neural network was applied in order to predict the states of stand after the damage.Through the predication of each patch,the spatial distribution was simulated.(3)Following the rule of Object Oriented Design(OOD),the ANN-CA Forest Pest Model was packaged,which was integrated by a developed geographic information system as the core.Tow core modules,the training module for the backpropagation algorithm and the Geo-CA module,were mainly designed and developed.(4)Following the base process of ANN-CA Forest Pest Model,two cases of the model were finished,which tested and improved the model.Occurrence area of Dendrolimus punctatus(Walker) disaster was predicted,with the rules including five climatic factors and sum of normalized difference vegetation index(NDVI) values according to the potential occurrence region;Occurrence region of Hlyphantria cunea(Drury) disaster was predicted, with the rules including plant condition,influence with distance from town and buffers surrounding the railway,main road and highway.(5)The predication accuracy of cases reached 80%,which shows the model can be used for predication.