A Study on the Spatial Hierarchical Effects for the Determinants and Characteristics of Agricultural Land-Use Change

• Main Authors: Chan, Chih-Chun, 詹智鈞
• Other Authors: Chan, Shih-Liang
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/9xbned

碩士 === 國立臺北大學 === 都市計劃研究所 === 106 === In recent years, under the influence of climate change, the deterioration of production environment has a direct impact on grain production, and it has also the public attention to the problem of food self-sufficiency and agricultural land use. Taiwan has congenital climate and geographical advantages on the production conditions. Since the 1950s and even accounted for 30% of gross domestic product (GDP), but under the guidance of the government's economic development policy, Taiwan's industrial structure began to shift from agriculture to industry, until today, agricultural production The proportion of gross domestic production in the remaining less than 2%, while agricultural use of land has also been subject to the threat of conversion of land usage from farmland to industrial / commercial development. In the study of land use change, the driving force is one of the most frequently discussed topics. The research foundation is mostly regional research, and most of them are discussed by spatial heterogeneity. For the land use change factors at different geographical scales, the hierarchical differences and the overall interaction are neglected. In view of this, ordinary least squares is the basis of comparison, and then the hierarchical linear model and geography weighted regression as a research tool, through the study design to explore the two research methods in the land use change research on the application of what will be different effects. According to the null model analysis in hierarchical linear model, it shows that the total variation of land-use change, the difference between the townships accounted 33%, and the estimation result of the intra-class coefficient shows that the inter-group variation has been neglected. It suggests that we continue to analyze by full model in the hierarchical linear model. In the comparison of the goodness-of-fit, the direct impact on dependent variable in hierarchical linear model is slightly better than ordinary least squares analysis, and the geographically weighted regression has a significant increase after adding the spatial weight. Although the direct impact on dependent variable in hierarchical linear model is not very different from ordinary least squares analysis, but the indirect impact of the interaction between the variables of different level shows that the variables witch are not significant impact on the dependent variable in ordinary least squares and geographically weighted regression analysis, will be significant impact on the dependent variable after the interact with other level of variables. That shows the ordinary least squares analysis and geographical weighting regression will still have the potential impact of neglect.