基于GWR和改进CLUE-S模型的区域土地利用格局模拟研究——以浙江省桐庐县为例

The Research on Land Use Pattern Simulation Using Geographically Weighted Regression and Improved CLUE-S

本文以浙江省桐庐县为研究区,在CLUE-S模型空间分析模块的回归建模中,考虑驱动因子对土地利用格局影响的空间不稳定性,用GWR Logistic模型取代全局最小二乘法(OLS)Logistic回归模型来建立土地利用格局与其驱动因子之间的回归模型,实现了对CLUE-S模型的改进,并进行土地利用格局模拟的实证研究与模型比较。研究结果表明,运用GWR Logistic对CLUE-S模型进行改进,不但可以获得更高的模拟准确率,而且可以获得各驱动因子对土地利用格局影响的空间分异特征。同时,运用改进的模型定量地分析影响区域土地利用变化的内在机制,进行土地利用格局的动态模拟,预测土地利用未来变化的趋势,可以为桐庐县及其类似地区的土地利用规划决策提供科学依据。

In this paper, Tonglu County in Zhejiang province is adopted as a case to build a geographically weighted regression analysis instead of OLS Logistic regression to improve the spatial analysis module of CLUE-S model for land use patterns simulation, in which the spatial non-stationary of driving factors on land use patterns change was taken into account. The simulation result is compared with those by OLS Logistic regression model. It is indicated that the improved CLUE-S model by GWR is not only able to increase the accuracy of regional land use patterns simulation, but also able to obtain spatial differentiation of the driving factors on land use patterns. Meanwhile, this study can also support the government decisions making on land use for Tonglu county and similar areas.