Research on the Forecast of Cultivated Land Variation Trend in Rapidly Urbanization Area: A Case Study of Wujiang City

Abstract

The city of Wujiang, which locates at the center of the Yangtze River delta, is a relatively ideal case for study because it is a typical area that experiences the rapid urbanization. According to the statistical and survey data at county level during the past 20 years, this article establishes a predicting model by comprehensively using both PCA and BP neural networks. Principal component analysis is firstly used to preprocess input variables in order to raise the network’s operational efficiency. While establishing the model of BP neural networks, it uses the data from 1990-2004 as the learning samples and that from 2005-2007 as the testing samples. The results show that the relative errors between the predicted value and the actual value are all less than 1.15%, which indicates that the neural network technique has a big power in the study of forecasting of cultivated area and the train of thought is reasonable. Finally the model established is used to do the simulated predication of the cultivated area for the year of 2010 in Wujiang, and the result shows that under the guidance of current policy, the decreasing rate of the cultivated area in the city of Wujiang dramatically drops.