Abstract:

In order to achieve sustainable development and scientific management of Beijing wetland, the author analyzed Beijing remote sensing data for acquiring different types of wetland data, by using remote sensing and GIS technology. Because wetland area is typical of small sample data, support vector machine was introduced into the time series model fixed order method, and then applied K cross validation methods to seek optimum parameters, established the Beijing wetland changes prediction model. Through the historical data of Beijing wetland was simulated, and comparing with RBF neural network, the author verified the validity of the forecast model of SVM, and then used this model to predict Beijing wetland future changes. The forecast results showed that the series model had higher precision of prediction and strong generalization ability, Beijing wetland in the next few years, reservoir, rivers, canals and ditches area would gradually decline, rice paddies area would continue to fall, breeding area would be increased, the predicted results conformed to reality Beijing wetland trend. The findings had provided the basis for the sustainable development and the scientific management of Beijing’s wetlands.