基于GWR模型的都市城郊村域农户生计资本空间差异分析——以上海市青浦区为例  

Spatial differentiation of rural household livelihood capital in metropolitan suburbs based on GWR model:A case study of Qingpu District in Shanghai

作　　者：任国平[1,2] 刘黎明[1] 付永虎[1] 袁承程[1] 

机构地区：[1]中国农业大学资源与环境学院,北京100193 [2]湖南城市学院城市管理学院,益阳413000

出　　处：《资源科学》2016年第8期1594-1608,共15页Resources Science

基　　金：国家自然科学基金项目(41471455)

摘　　要：农户生计资本的影响因素具有空间非平稳性,因此一般以"平稳性假设嵌入"为基础的普通线性回归模型对农户生计资本影响因素空间差异分析的精确性不高。本文应用地理加权(GWR)回归模型对上海市青浦区182个行政村进行农户生计资本总量的空间分析及其影响因素的空间分异特征研究。结果表明:2014年青浦区各行政村农户生计资本总量和类型分布呈现明显的空间分异规律,生计资本总量在空间上由西向东逐渐递增;其中,农业生计资本类型由西向东逐渐减少,非农类型则由东部向西逐渐减少;经济发展因素和社会发展因素对行政村农户生计资本总量影响强于自然地理空间因素,其中与生计资本变化呈正相关的因素依次为工业化程度、到城镇中心距离、地均GDP、城镇化率、农户可支配收入;以空间"非平稳性"为基础构建的地理加权回归模型(GWR)比较"平稳性假设嵌入"为基础的普通线性(OLS)回归模型,能更精确地揭示出村域农户生计资本和空间影响因子之间复杂的关系,且GWR模型参数估计的空间模式可视化特点更加直观。The factors affecting livelihood capital are non-stationary; therefore, a general linear regression model based on the assumption of＇ stationary hypothesis embedding＇ is not accurate for the analysis of these factors for farmers. Here, the method of aggregative indicator was used to calculate the total livelihood capital of administrative villages in Qingpu District, then to cluster. We adopted an ordinary least square model （OLS）and geographic weighted regression model （GWR） to examine the spatial differentiation of factors affecting the total livelihood capital of administrative villages. We found that the total livelihood capital and classification distribution of administrative villages has obvious spatial differentiation. Total livelihood capital gradually increased from west to east; the classification of agricultural livelihood capital gradually reduced from west to east;while, the classification of non-agricultural livelihood capital gradually increased from east to west. Results demonstrate that economic and social factors had a larger impact on the total livelihood capital of administrative villages in Qingpu District than geographical and spatial factors. Industrialization degree, GDP per area, distance to center of downtown, urbanization rate, and disposable income of farmer household were positively correlated to variation in livelihood capital. Compared with the traditional ordinary least square model （OLS）based on ＇stationary hypothesis embedding＇ , the geographical weighted regression model （GWR）was based on ＇non- stationary＇ reveals more profound spatial differentiation of factors. These results can provide scientific reference points for the sustainable development of farmer household livelihood and land spatial consolidation in metropolitan suburbs.

关 键 词：农户生计资本 空间分异 地理加权回归模型 都市城郊村域 青浦区 

分 类 号：F323.8[经济管理—产业经济]