Study Of Plant Diversity And Vegetation Landscape Model Of Pingtan

Landscape diversity has become a major concern of landscape ecology in biodiversity research. Relying on landscape ecology theory, this paper did field investigation and analysis about species diversity of main plant community in Pingtan County and combined with satellite remote sensing image, vegetation landscape diversity was discussed in the study area. Using BP neural network, artificial neural network model of vegetation landscape in the area was established and the relationship between vegetation landscape diversity and natural and socio-economic factors are fitted. On this basis, formation mechanism of vegetation landscape in Pingtan County was analyzed and solutions of vegetation construction optimization were put forward.Conclusions are as follows:(1) The overall species diversity of main plant community is lower in the research area characterized by that artificial plant community is less than semi-natural shrub and growth type species diversity shows tree layer< shrub layer< herb layer. Evenness index which is related to species diversity also reflects the differences between artificial forests and natural shrub land which have the same trend. (2) Casuarina equisetifolia forests and secondary Acacia spp are the main component parts of vegetation landscape in Pingtan County. The woodland is widely distributed along the coast. Landscape patches are mainly fine-grained, shape index is generally small, landscape heterogeneity is worse and various landscape components are not in harmonious proportions. (3) Based on Matlab platform, BP neural network model of vegetation landscape was set up. Number of settlements, wind scale and distance from the coast are chosen as the input variables and contagion (CONTAG) index, dimension and diversity index are output variables. The results indicate that the model has high precision with average error 7.44% and minimum error 0.18%. Then the well-trained model was used to fit the relationship between vegetation landscape diversity and natural and socio-economic factors. It demonstrates that the model might be useful in the quantitative analysis and prediction of vegetation landscape patterns.