删除或更新信息,请邮件至freekaoyan#163.com(#换成@)

人穷还是地穷?空间贫困陷阱的地统计学检验

本站小编 Free考研考试/2021-12-29

马振邦1,2,3,, 陈兴鹏1,2, 贾卓1,2, 吕鹏4
1. 兰州大学资源环境学院西部环境教育部重点实验室,兰州 730000
2. 兰州大学资源环境学院中国西部循环经济研究中心,兰州 730000
3. 兰州大学县域经济发展研究院,兰州 730000
4. 甘肃省扶贫开发办公室,兰州 730000

Poor people, or poor area? A geostatistical test for spatial poverty traps

MAZhenbang1,2,3,, CHENXingpeng1,2, JIAZhuo1,2, LVPeng4
1. Key Laboratory of Western China's Environmental Systems (Ministry of Education), College of Earth and Environmental Sciences, Lanzhou University, Lanzhou 730000, China
2. Research Institute for Circular Economy in Western China, College of Earth and Environmental Sciences, Lanzhou University, Lanzhou 730000, China
3. Institute for Studies in County Economy Development, Lanzhou University, Lanzhou 730000, China
4. Gansu Office of Poverty Alleviation and Development, Lanzhou 730000, China
收稿日期:2018-04-16

网络出版日期:2018-10-20
版权声明:2018《地理研究》编辑部《地理研究》编辑部 所有
基金资助:国家自然科学基金项目(41401204,41471462)中央高校基本科研业务费项目(lzujbky-2013-128)
作者简介:
-->作者简介:马振邦(1983- ),男,甘肃会宁人,讲师,研究方向为景观地理与区域可持续发展。E-mail: zbma@lzu.edu.cn



展开

摘要
引入地统计学的变异函数和交叉相关图方法,以甘肃省六盘山片区为案例区,通过分析村级贫困的空间格局及其与地理因子关系随空间尺度的变化,提供空间贫困陷阱检验关于尺度的深入理解。结果表明:地统计学方法兼具有效性和可靠性,可以反映地理因素—贫困状况关系随时空的变化,对“人地关系”视角下反贫困理论与实践研究具有积极意义。案例区空间贫困陷阱问题突出,村级贫困在一定空间范围内具有自相关性,空间总变异中自相关部分远高于随机性部分,这与不同尺度上地形、气候、区位等结构性因素的影响和控制有关,总体上到县城距离、海拔和总人口3个因子的影响范围和强度较大。

关键词:空间贫困陷阱;地统计学方法;尺度;六盘山片区
Abstract
The test for spatial poverty traps (SPTs) is a hot issue in the field of the geography of rural poverty. However, the main existing approaches cannot provide spatial scale-related information, which may be a restriction on gaining a deeper understanding of the mechanism of SPTs. Therefore, we conducted a case study in the Liupan Mountain Region by introducing geostatistical methods. The semivariogram and cross-correlogram were employed to quantitatively describe the spatial pattern of village-level poverty and its relationship with the selected geographical factors respectively, so that the scale-dependent spatial form and underlying reasons for SPTs can be explored. The village-level poor population (PP) and poverty rate (PR) were used as the poverty indicators. The results show that the geostatistical methods can provide satisfactory and reliable performance in the test for SPTs: (1) The semivariogram models can indicate both the spatial structure and the autocorrelation range of the two indicators, which can describe the extent and the range of the spatial form of SPTs (i.e. the spatial aggregation of poverty). The percentages of the random variance (nugget, C0) in the total variance (sill, C0 + C) are 34.4% and 11.5% for PP and PR, respectively. The range of autocorrelation is 9.3 km for PR, and 5 and 48 km for PP. (2) The cross-correlograms further show that the two indicators are significantly (P<0.05) correlated with the geographical factors within different spatial ranges. Generally, the poverty status of a village is mainly in response to three factors (i.e. the distance to the nearest county town, the elevation, and the total population) within a wide range. In conclusion, the evidence of SPTs from our work is consistent with the reality that the study area has suffered persistent poverty in the past three decades.

Keywords:spatial poverty traps;geostatistical methods;scale;Liupan Mountain region

-->0
PDF (1377KB)元数据多维度评价相关文章收藏文章
本文引用格式导出EndNoteRisBibtex收藏本文-->
马振邦, 陈兴鹏, 贾卓, 吕鹏. 人穷还是地穷?空间贫困陷阱的地统计学检验[J]. 地理研究, 2018, 37(10): 1997-2010 https://doi.org/10.11821/dlyj201810010
MA Zhenbang, CHEN Xingpeng, JIA Zhuo, LV Peng. Poor people, or poor area? A geostatistical test for spatial poverty traps[J]. Geographical Research, 2018, 37(10): 1997-2010 https://doi.org/10.11821/dlyj201810010
主旨聚焦:地统计学方法作为一种新途径,可实现空间贫困陷阱的定量检验并给出与尺度相关的认识,能为统筹协调面向“穷人”和“穷地”的脱贫攻坚政策提供科学依据。

1 引言

贫困问题是全球性的重大社会问题和现实难题,很大程度上是因为它是个多维复杂现象。在地理空间维度上,贫困往往表现出集聚分布特征[1,2,3,4],从而引起了国内外****浓厚的研究兴趣[5,6,7]。研究内容包括贫困空间分布特征刻画、地理因素与贫困的作用关系分析、区域贫困的测算及贫困地图的绘制、区域瞄准及效果评估等[6]。其中,具有开拓性意义的是空间贫困陷阱理论[7,8]。该理论认为包括气候、地形、交通等多因素在内的综合“地理资本”的相对不足且不易改变和缓和,会影响农户资本的收益率,使该地区农户相比其他地区农户更容易陷入持续性贫困,从而形成空间贫困陷阱。也即持续性贫困是农户资本和地理资本的共同函数,“人”穷的根本原因是“地”穷。它将贫困陷阱研究关注点从传统的农户层面拓展到地理层面,为在更广阔的“人地关系”视角下考察持续性贫困的地理分布与成因提供了理论支持,对基于地理的反贫困实践具有重要指导意义。
围绕空间贫困陷阱的检验与识别,****们基于农户和地理单元两个尺度进行了积极探索。农户尺度上,通常是将地理因子纳入基于农户动态调查数据的微观模型,来分析地理因子对农户消费增长的影响[8];或基于截面数据利用因素分解方法对地理因子引致的农户间福利差异进行分解[9],来检验空间贫困陷阱是否存在[6]。但是,由于抽样调查数据的样本量较小,微观模型或分解方法无法实现空间贫困陷阱的空间可视化表达。为此,研究者们尝试把农户调查数据与人口普查数据相结合,首先运用小区域估计方法得到基于地理单元的贫困截面数据,进而利用空间分析方法(如Moran's I指数)通过考察贫困是否存在空间自相关来判断空间贫困陷阱是否存在并实现其“全景式”描述[1,10-12],最后建立空间计量模型(如空间滞后或误差模型)来探究具有显著性影响的地理因子[10,11,13]。然而,部分****发现贫困的空间格局及其成因随着空间尺度的变化而变化[14,15]。Okwi等在肯尼亚的研究显示,当研究区域由全国缩小为省级时,贫困发生率在某些省份并未表现出空间集聚性,且具有显著性影响的地理因子数目及影响力省际间变化较大[14]。Ward等在尼日尔河流域的研究同样发现,只有少数地理因子与贫困状况的关系方向及强弱全局相对稳定,而大多数地理因子则仅在局部产生影响[15]。此时,空间计量模型无法定量描述贫困格局及其成因这种与尺度相关的变化。
地统计学方法可揭示区域化变量的空间相关、变异组成及其影响因素随尺度的变化,已在地质、土壤、生态、环境及人口等领域得到广泛应用[16,17,18,19],但在贫困地理方面的应用却十分少见。因此,以甘肃省六盘山片区为案例区,拟引入变异函数及交叉相关图等地统计学方法,通过考察行政村水平上贫困的空间格局及其与地理因子关系随尺度的变化特征,来提供空间贫困陷阱检验关于空间尺度方面的深入理解:① 借助变异函数方法能否实现空间贫困陷阱的有效检验,也即贫困陷阱成因中“人”“地”部分的定量分割?② 不同地理因子对村级贫困空间格局形成存在怎样的与尺度相关的影响?

2 研究区概况与研究方法

2.1 地统计学方法的引入

地统计学方法是以区域化变量理论为基础发展起来的,主要用来研究那些在空间上既有随机性又有结构性的现象[20]。该理论认为呈现出空间分布的变量均可视为区域化变量,其值是空间位置的函数,并且同时具有随机性和结构性[16,17]:随机性是指区域化变量局部的、随机的、异常的性质;结构性是指一般的或平均的结构性质,也即变量在两点XX+hh为空间距离)处的值具有某种程度的相关性。事实上,根据空间贫困陷阱理论及相关实证研究结果[1-3,7],贫困在空间上呈不均衡分布,说明它在空间上兼具结构性和随机性,可用区域化变量来描述。因此,地统计学方法适用于分析贫困这一区域化现象。
2.1.1 变异函数方法 尝试运用变异函数进行空间贫困陷阱的定量检验,定义为[21]
γ(h)=12E[Z(x)-Z(x+h)]2(1)
式中:γ(h)为变异函数;Z(x)和Z(x+h)分别是区域化变量Z在空间位置xx+h处的值;E[Z(x)–Z(x+h)]2是取样间隔为h时的样本值方差的数学期望。
γ(h)与h的散点图进行拟合,可建立变异函数理论模型,从而获得变异函数的几个基本参数:基台值(C0+C)、块金值(C0)和变程(a)。a可理解为变量Z的空间自相关范围。C0+C为变量Zh由小变大产生的最大变异,也即haγ(h)达到的稳定值,可分解为两部分[20,22]:① 小于分辨率尺度上随机部分C0,表示由Z的属性或测量误差所决定的非连续性变异总和;② 变程a以内由结构性因素引起的空间自相关部分C(偏基台值)。此时,C0/(C0+C)反映随机部分在总的空间变异中所占比例。
需要说明的是,变异函数对空间变异的定量分割与分辨率尺度的选择密切相关[20,22]。本文选择行政村作为分辨率尺度。这是因为,行政村既是农村基层管理单位,也是长期以来乡村人口居住、生产和生活所形成的基本经济社会单元,可作为贫困成因中“人”“地”两部分因素分割的理想单元:村内农户层面上年龄、健康、技能等是与“人”密切相关的因素[23,24,25,26],它们的影响大多是局部的、随机的,因此主要带来随机部分;与之相对,行政村单元以上气候、地形、可达性等地理因子的影响通常表现出结构性且影响范围较大[4,27]。此时,块金值与基台值之比(C0/(C0+C))可用来描述“人”(随机部分)、“地”(自相关部分)两部分因素影响之间的定量关系。另外,以行政村作为基本分析单元,也便于与目前中国将贫困村作为贫困瞄准与政策实施基本地域单元的精准扶贫方略相对接[5]
2.1.2 交叉相关图方法 交叉相关图用来探讨空间贫困陷阱成因与空间尺度的关系,可由交叉相关系数得出。交叉相关系数为间隔距离为h时两个变量间的相关性[28]
ρ12(h)=1ni=1nZ1(xi)Z2(xi+h)-m1-hm2+hσ21-hσ22+h(2)
式中:ρ12(h)为交叉相关系数;Z1(xi)和Z2(xi+h)分别是变量Z1xi和变量Z2xi+h处的值;n为样本对数;m1-hm2+hσ21-hσ22+h分别是间隔为hZ1Z2的均值和方差;ρ12(h)-h关系图称为交叉相关图,可定量刻画Z1Z2之间空间相关的性质、程度及其随尺度的变化[28]
一般地,变量间空间相关性随距离的增大而减弱,也即交叉相关系数绝对值随距离增大而减小。基于交叉相关图可计算平均空间相关系数(AR),表示Z1Z2在空间上的平均相关性[18]AR是指ρ12(D)从初始的非零点到其第一次为零(即ρ12(D)=0)时,ρ12(h)-h曲线与横坐标轴之间的面积和与D之比。AR具有和Pearson相关系数相似的特点,其正负表示两个变量间具有正和负的空间相关性,大小反映Z2对Z1空间分布的平均影响强度。

2.2 案例区的选择

六盘山片区是中国14个连片特困地区之一,其脱贫解困关系到全面建成小康社会战略目标的实现[5]。其中,甘肃省内的46个县区中,就包括20世纪80年代国家级区域扶贫开发实验地“三西”地区的安定区等19个县区。可以发现,甘肃省六盘山片区表现出区域性贫困与持续性贫困双重叠加的特点。本文以定西市陇西、通渭两县为中心,选择一块正方形区域作为案例区(图1a)。
显示原图|下载原图ZIP|生成PPT
图1案例区位置及行政村、水系、道路分布图
-->Fig. 1Location of the study area and spatial distribution of the villages, the rivers and the roads
-->

该区域面积1.32×104 km2,地处黄土丘陵沟壑区,山地、丘陵、沟壑、梁峁、河谷纵横交错(图1c)。地形起伏大,海拔1231~3172 m,坡度15°以上面积占比27.3%。案例区以温带大陆性季风气候为主,年均降水量约350~550 mm。除渭河、关川河、散渡河等河流的台阶地区水热、土壤等农业生产条件较好以外,大多数地区由于地形陡峭、水资源匮乏等原因,农业生产发展状况不良。
案例区有行政村1602个,分属114个乡镇和12个县区,2013年农村人口230.9万人。经济社会发展水平较低,同年12县区人均GDP和非农人口比例均值约为9862元和13.5%,仅是全国平均水平的25%和40%。从交通条件看(图1d),陇海铁路、G30连霍高速沿渭河、关川河河谷川地呈反“S”型贯穿全区,G22青兰高速、国道312线穿北部静宁、会宁县城与G30相交于安定区。

2.3 数据来源与处理

村级贫困状况由贫困人口数(2013年2736元贫困线以下人口)和贫困发生率两项指标表征。贫困人口数、总人口数由相关市州和县区扶贫开发办公室提供。贫困发生率为贫困人口数与总人口数之比。需要指出的是,变异函数分析要求区域化变量呈正态或近似正态分布,否则可能存在比例效应[16,29]。通过K-S检验发现贫困人口数和贫困发生率都不符合正态分布,直方图显示呈左偏分布(图2a、图2c)。为此,利用Box-Cox转换对原始数据进行处理,转换后数据虽未通过K-S检验,但呈近似正态分布(图2b、图2d),此时比例效应不存在或可忽略[16,29]
显示原图|下载原图ZIP|生成PPT
图2贫困指标的直方图
-->Fig. 2Frequency distribution histograms for the poverty indicators
-->

根据相关****关于地理资本的阐述及分类[6-8,30],借鉴Okwi等、Benson等和刘小鹏等的研究[13,14,31],选择平均海拔等8个地理因子来表征行政村的地理资本状况(表1)。其中,地形数据来源于中国科学院计算机网络信息中心地理空间数据云平台,空间分辨率为30 m;气象数据来源于中国科学院资源环境科学数据中心,空间分辨率为500 m;路网、河网和行政区划数据(包括行政村、乡镇中心、县城地理位置等)来源于甘肃省2012年1 10万基础地理数据集,由地球系统科学数据共享平台—寒区旱区科学数据中心提供。
Tab. 1
表1
表1村级地理指标描述及类型
Tab. 1Description and type of the geographical indicators at the village level
地理指标指标描述及含义类型预期
平均海拔村域海拔均值,表征地形特征及其关联的温度等农业生产条件N+
平均坡度村域坡度均值,表征地形特征及其关联的土壤、侵蚀等农业生产条件N+
年降雨量村域年降水量均值,表征水分等农业生产条件N-
总人口数村总人口数,表征人口及社会关系状况H/S+/-
到河流距离村点到最近主要河流距离,表征土壤及可灌溉程度等农业生产条件P+
到县城距离村点到最近县区政府驻地时间,表征较高水平教育、医疗、市场、信息等服务和非农就业机会的可达性程度,以及较强非农社会关系状况P/S+
到乡镇距离村点到最近乡镇政府所在地时间,表征教育、医疗、市场、信息等服务和非农就业机会的可达性程度,以及非农社会关系状况P/S+
到道路距离村点到最近县级及以上道路时间,表征交通基础设施条件P+

注:N/P/H/S为地理资本类型:N为自然资本;P为物质资本;H为人力资本;S为社会资本。
新窗口打开
利用ArcGIS软件对相关数据进行空间化处理。依据行政村矢量点位数据:使用百度地图自动测距功能[32],在驾车系统下测算村点到最近乡镇、县城、县级以上道路的时间;生成Voronoi图作为行政村边界,计算村域范围内海拔、坡度、年降雨量的均值。上述指标的描述性统计结果见表2。最后,依据行政区划代码将贫困数据与地理因子数据关联至行政村矢量点位数据,生成空间数据集。
Tab. 2
表2
表2村级贫困指标及地理因子的描述性统计
Tab. 2Descriptive statistics of the indicators at the village level
具体指标平均值中位数标准差偏度峰度最小值最大值
贫困状况贫困人口数(人)445.1388307.41.433.28242248
贫困发生率(%)33.133.417.50.26-0.0851.287.7
贫困人口数a18.6918.885.260.04-0.215.6536.30
贫困发生率b-0.75-0.730.22-0.03-0.35-1.21-0.12
地理因子平均海拔(m)19161961281-0.33-0.2112252826
平均坡度(°)12.2112.613.50-0.611.321.2623.69
年降雨量(mm)459.7463.431.60.35-0.26396.9546.1
总人口数(人)1441.11246783.726.162636808
到河流距离(km)3.893.303.070.71-0.290.0214.15
到县城距离c(min)47.7844.0027.530.992.041.00236.00
到乡镇中心距离c(min)20.0514.0019.762.125.310.00114.00
到县级以上道路距离c(min)5.300.009.782.516.930.0067.20

注:a表示Box-Cox正态转化后结果(λ=1/3);b表示Box-Cox正态转化后结果(λ=0.8);c距离表示车行时间。
新窗口打开
利用GS+Version 7软件进行变异函数和交叉相关图分析。由于案例区行政村点位是不规则分布的,步长选取距离组方法。考虑到郭晓东在陇中黄土丘陵区的研究显示中心村劳作半径约为0.8~1 km[33],与本区内样本村间的平均最近邻距离1.78 km相符,所以选择2 km作为分组值,也即最小分辨率单元行政村所对应的空间尺度。由于最大样本距为159.97 km,距离组共设40组。

3 结果分析

3.1 变异函数分析结果

两项贫困指标的变异函数曲线见图3表3是变异函数理论模型概述。单个模型下,贫困人口数和贫困发生率的变异函数都符合指数模型,决定系数R2分别为0.926和0.968。由图3a,贫困人口理论模型对30 km以内实测值的拟合效果不佳,这样会忽略小尺度范围内贫困人口的空间分布及相关过程。为此,利用不同尺度上的套合结构表现其变异函数,经过不断拟合,发现以单个球状模型为基础的二次套合结构模型拟合效果最佳(图3b),R2达到0.984。根据理论模型,贫困人口数和贫困发生率都具有较好的空间结构性,表现为空间集聚分布。
显示原图|下载原图ZIP|生成PPT
图3贫困指标的变异函数曲线
-->Fig. 3Semivariogram curves of the poverty indicators
-->

Tab. 3
表3
表3各向同性条件下贫困指标变异函数的理论模型及参数
Tab. 3Isotropic semivariogram models and parameters of the poverty indicators
贫困指标理论模型块金值偏基台值变程(km)基台值块金值/基台值(%)决定系数
C0CaC0+CC0/(C0+C)R2
贫困人口数指数模型14.314.331.828.6500.926
套合模型:球状I10.311.1529.934.40.984
球状II8.548
贫困发生率指数模型0.00580.04469.30.050411.50.968


新窗口打开
块金值与基台值之比C0/(C0+C)为空间总变异中随机部分所占比例,它与空间自相关部分所占比例呈反比,可作为研究因子空间自相关的分类依据[29,34]C0/(C0+C)<25%属于强的空间自相关,说明研究因子受结构性因素的强烈影响而具有很好的空间结构性;25%<C0/(C0+C)<75%属于中等程度的空间自相关;75%<C0/(C0+C)表明空间自相关很弱,此时分辨率尺度以下的随机性因素起重要作用。就案例区来说,贫困发生率的C0/(C0+C)为11.5%(表3),表明在贫困发生率的总空间变异中,约有11.5%的空间变异由村域或2 km以下的随机性因素引起,而由自相关部分引起的空间变异占88.5%。显然,贫困发生率表现出很强的空间自相关,表明贫困发生率受到村域以上结构性因素的强烈影响。对贫困人口而言,总的空间变异中随机部分和自相关部分分别为34.4%和65.6%,意味着贫困人口的空间分布主要由结构性因素决定,而随机性因素也不可忽视。
既然与“地”相关的结构性因素被认为是空间贫困陷阱的主因[7,9],C0/(C0+C)值也可作为空间贫困陷阱检验的有效依据:C0/(C0+C)>75%表明“人”穷“地”不穷,空间贫困陷阱不存在或微弱;C0/(C0+C)为25%~75%表明“地”穷对“人”穷产生中等程度影响,空间贫困陷阱存在;C0/(C0+C)<25%可认为“人”因“地”而穷,空间贫困陷阱问题突出。在此意义上,案例区空间贫困陷阱问题突出。
由变程a和变异函数曲线斜率可以看出(表3图3),两项贫困指标的空间变异是尺度的函数,说明它们的空间格局及关联过程均是尺度相关的[20,22]。贫困发生率的空间自相关范围为9.3 km,表明各种结构性因素的影响主要集中在9.3 km以内[20]。套合模型显示贫困人口存在5 km和48 km两个变程,也即在0~5 km和5~48 km范围内贫困人口表现出强度不同的自相关性,超过48 km后自相关性消失。在0~5 km的短程范围内空间变异主要来源于球状模型I,变异函数曲线斜率较大,贫困人口表现出较强的空间结构性;在5~48 km的长程范围内空间变异主要来源于球状模型II,变异函数曲线斜率较小,贫困人口的空间结构性相对较弱。这说明贫困人口具有等级系统结构,不同尺度上它受到几个重要结构性因素的影响和控制[22,34]。需要强调的是,无论是随机性因素还是结构性因素,它们同时对贫困指标的空间变异起作用,而2 km、5 km、9.3 km和48 km的尺度分割可为理解不同空间范围内致贫因素的作用方式、强度等提供有力支持。

3.2 交叉相关图分析结果

从贫困指标与地理因子的交叉相关图可发现,各地理因子对村级贫困的空间影响范围不尽相同(图4)。就贫困发生率而言,降雨量的影响范围最大为50 km,到县城距离接近20 km,其他因子约为9~11 km。而对贫困人口数来说,降雨量和总人口数空间影响范围明显大于坡度、到河流/县城/乡镇距离的影响,后4个因子均小于5 km。平均海拔比较特殊,它对贫困人口数的影响强度在5~50 km之间显著变大而后回落,到县城距离也表现出类似情况。Stein等指出,这种倒“N”型形态(图4a和图4f)意味着两种空间格局在较大空间范围内也存在相关性[28]。就案例区而言,这可能是贫困人口分布对较大尺度上地理环境条件结构性差异和县域经济社会状况整体差异的一种响应。例如,河流网络和道路网络将案例区分割成连绵山地区及丘陵区,它们的体量如按矩形计长宽多在5~40 km范围内(图1c和图1d),一般连绵山地区贫困发生率相对高但贫困人口数较小。此外,由于自然、历史、经济、行政、文化等多方面因素的综合作用,安定、通渭、陇西三区县(约占案例区60%的面积)如按正方形计边长为48.9~51.3 km,且村级贫困人口规模整体上梯次升高。
显示原图|下载原图ZIP|生成PPT
图4贫困指标与地理因子的交叉相关图
-->Fig. 4Cross-correlograms between the poverty indicators and the geographical factors
-->

从影响方向看,贫困发生率与年降雨量和总人口数的空间相关性为负,而与其他因子的空间相关性为正(图5b),表明贫困高发于海拔高、坡度大、降雨量小、总人口少、到河流/县城/乡镇/县级以上道路距离远的地方。由AR值反映的贫困人口与到河流/县城/乡镇距离的相关性方向符合一般预期。但是,贫困人口与平均坡度、年降雨量的AR值分别为-0.04和0.05(图5a),表明坡度小、降雨量大的地方贫困规模反而大。Minot等在越南的研究同样发现,大量贫困人口生活在地理条件相对较好的湄公河三角洲和红河三角洲地区,较大的人口基数和人口密度是主要原因[35]。案例区的结果与之相似,坡度小的渭河等河谷地区,以及降雨量较大的南部甘谷、武山等县,虽然贫困发生率较低,但较大的人口基数使得贫困人口规模相应也大。这一点也能从总人口的AR值最大得到验证(图5a)。
显示原图|下载原图ZIP|生成PPT
图5贫困指标与各地理因子之间的平均空间相关系数
-->Fig. 5Values of AR between the poverty indicators and the geographical factors
-->

从影响强度看,贫困人口数和贫困发生率与各地理因子的AR绝对值并不是很高,分别是0.003~0.077、0.069~0.181(图5),低于常见的Pearson相关系数值。但是,考虑到AR值是二者在不同尺度范围内空间显著相关的平均强度,加之贫困成因的多维复杂性,因此较小的AR值也具有可信度和可比性。总体上,到县城距离、海拔和总人口3个地理因子对两项贫困指标的影响强度均大。这反映了现有行政等级体系和“中心—边缘”经济结构对贫困格局的强烈影响。县城提供的较好公共服务和非农就业机会等资源,能减轻农户生计的脆弱性并增加其多样性,而这些资源的辐射力往往是随空间距离递减的。值得注意的是,降雨量对贫困状况的影响相对较小稍显意外。很大程度上是因为案例区降雨量普遍小,农业生产总体上以旱作农业为主,因而降雨量并非村级贫困空间分异的主要因素。从图5还可以发现,贫困人口数与各地理因子的AR值总体较小,说明全局上贫困人口的空间分布受地理因子的影响较小,这印证了贫困人口数的C0/(C0+C)值较大的结果(表3)。

4 讨论

4.1 地统计学方法的有效性

空间贫困陷阱之所以存在,内在原因是地理因素的负外部性会降低农户资本的收益率,使农户更容易陷入持续性贫困,而其外在形态则是贫困的地理集聚[7,8]。因此,检验空间贫困陷阱是否存在的关键,从其理论内涵出发是实现贫困成因中地理和非地理两类因素影响的有效分割[9],从其外在形态出发是探测贫困是否存在地理集中并分析地理因素所起作用。目前基于农户尺度的微观模型和基于地理单元尺度的空间计量方法各有侧重及优势,但它们共同存在的问题是无法给出空间贫困陷阱检验与尺度相关的定量信息。这里,尺度既与表征贫困的最小分辨率单元有关,也与贫困的空间格局及其成因相对应的空间尺度有关。
本文基于案例区行政村的研究表明,地统计学方法中变异函数与交叉相关图相结合可有效解决上述问题。两项贫困指标的变异函数理论模型能精确描述村级贫困的空间分布(图3表3)。根据理论模型:① 村级贫困表现出较强的空间自相关,两项贫困指标的C0/(C0+C)为11.5%和34.4%;② 贫困的空间自相关均是尺度关联的,贫困发生率的自相关范围为9.3 km,贫困人口数在0~5 km和5~48 km两个范围内表现出强弱不同的自相关。交叉相关图显示(图4图5),两项贫困指标在大尺度范围内( <50 km)与降雨量和海拔显著相关,在中尺度范围内( <30 km)与到县城距离和总人口显著相关,在小尺度范围内( <10 km)与坡度、到河流/乡镇距离等显著相关。上述结果综合表明,相比村域以内的随机因素,村级贫困受到不同尺度上地形、气候、区位等结构性因素更强的影响和控制。充分说明,案例区空间贫困陷阱问题突出,这与该区域近三十年表现出区域性和持续性贫困的实际相符。

4.2 稳健性分析

基于截面数据,发现利用变异函数和交叉相关图方法的组合,可以形成空间贫困格局及其成因的良好诊断。研究方法和结果的可靠性可能受到两方面因素的影响。其一,应用地统计学方法时,参数设置不同引起的不确定性,主要体现在步长选取上。由于本文采用距离组方法,步长变化时样本对将归入的不同距离组,可能会影响变异函数及交叉相关图分析结果[19]。其二,空间贫困陷阱本质上是“人地关系”的一种反映,虽然地理条件客观上不易改变或缓和,但其影响随时空而变化[14,36]。这一意义上,时空变化下研究方法和结果的稳健性有待进一步考察。
首先,将初始步长分别设置为1.5 km和2.5 km后,发现结果与2 km时基本一致。贫困发生率变异函数理论模型皆为指数模型,决定系数R2在0.965以上,C0/(C0+C)和a的变化在±14%和±4%以内。由交叉相关图指明的各地理因子的显著性影响方向、趋势一致,范围和强度略有变化(图6)。
显示原图|下载原图ZIP|生成PPT
图6不同步长下贫困发生率与地理因子的交叉相关图
-->Fig. 6Cross-correlograms between the poverty rate and the geographical factors at different lag intervals
-->

其次,进行不同贫困程度次区域间的比较分析。将案例区等分为4个正方形的次区域,虽然各区域贫困发生率平均水平不尽相同(29.7%~37.5%),但相似之处有(表4图7):① 较小的C0/(C0+C)值(12.6%~24.7%)说明次区域贫困发生率皆有较强的空间结构性;② 变程a与各地理因子显著影响的空间范围基本吻合;③ 各地理因子影响方向普遍一致。
Tab. 4
表4
表4次区域贫困发生率变异函数的理论模型及参数
Tab. 4Isotropic semivariogram models and parameters of the poverty rate in different sub-regions
次区域贫困发生率
均值(%)
理论模型块金值偏基台值变程(km)基台值块金值/基台值(%)决定系数
C0CaC0+CC0/(C0+C)R2
分区129.7指数模型0.00210.014619.10.016712.60.88
分区235.1指数模型0.00540.03246.30.037814.30.36
分区337.5指数模型0.01190.06939.70.081214.70.96
分区431.5指数模型0.01380.042015.00.055824.70.95


新窗口打开
显示原图|下载原图ZIP|生成PPT
图7次区域贫困发生率与地理因子的交叉相关图
-->Fig. 7Cross-correlograms between the poverty rate and the geographical factors in different sub-regions
-->

最后,由于其他年份数据的获得性受到限制,设置对照情景进行比较分析。按照全面建成小康社会的目标要求,2020年案例区现行贫困标准下村级贫困发生率将降至3%以下,基本消除区域性的整体贫困,空间贫困陷阱问题不再突出。因此,对2020年村级贫困发生率(PR)进行5次随机模拟,策略为:如PR20133%,则0 PR2020PR2013;如3% <PR2013,则0 PR20203%。结果表明:5次模拟结果相近,变异函数值随尺度的变幅不大,村级贫困在空间上呈现随机分布。相应地,交叉相关系数在不同空间范围上不再显著(P>0.05),意味着各地理因子不再显著影响贫困分布。
综上,地统计学方法在空间贫困陷阱的检验上具有可靠性,不仅可以应用在不同区域上,而且能反映地理因素—贫困状况关系的时间变化,对基于“人地关系”视角的贫困地理研究及反贫困实践具有积极意义。例如,到2020年如果贫困的空间结构性“从有到无”,究竟经历了怎样的变化?这与各地理因子影响的变化之间是否存在关联?这种关联是经由怎样的反贫困措施建立的?基于多期多维数据(特别是经济社会发展数据),利用该方法可以从时间轴线上探讨贫困空间格局演变规律及其驱动因素,特别是与空间尺度相关的规律,为扶贫政策的效果评估与合理制定提供有力支撑。再如,2020年以后如果贫困线大幅提高,或采用相对标准(如家庭收入中位数的50%),贫困发生率相应会上升。当这种上升在空间上呈现不均衡,贫困集聚区的再现是否意味着“新”的空间贫困陷阱的产生?这和与“地”有关的因素又有怎样的关联?此时,地统计方法仍可为“新”的空间贫困陷阱检验提供方法支撑。

5 结论

中国农村区域性整体贫困突出的背景下,基于空间贫困陷阱理论考察“人”穷与“地”穷之间的关系,具有重要实践价值,也将丰富人地关系理论研究。鉴于现有方法不能提供空间贫困陷阱检验与空间尺度相关的理解,本研究尝试引入地统计学方法来克服这一局限,并在六盘山片区进行了实证研究。
结果发现,地统计学方法可为空间贫困陷阱的定量检验提供一种新的思路和方法,对“人地关系”视角下的反贫困理论及实践研究具有积极意义。基于区域化变量理论的变异函数符合空间贫困陷阱的理论内涵,能通过测度村级贫困空间集聚的程度与范围,描述贫困空间分异中结构性因素和随机性因素的构成及其随空间尺度的变化。结合交叉相关图分析可进一步探明显著性因子的影响方向及空间范围。时空变化下该方法具有可靠性,不但能应用在不同区域,而且可以反映地理因素—贫困状况关系的时间变化。
案例区村级贫困在一定空间范围内具有空间自相关性,空间总变异中自相关部分远高于随机性部分,这与不同尺度上地形、气候、区位等结构性因素的影响和控制有关,总体上到县城距离、海拔和总人口3个因子的影响范围和强度较大。说明案例区空间贫困陷阱问题突出,因此需要继续以行政村特别是较大尺度上小流域、连绵山地丘陵区等地理单元作为贫困瞄准的基本单位,把地理溢出效应明显的城镇辐射力和带动力增强、特色富民产业壮大、基础设施完善、制度及技术创新等作为脱贫攻坚的主要手段。
需要指出的是,本文尚存在一些局限。空间贫困陷阱强调贫困的空间集聚性、时间持续性和“人”“地”互动性。然而受限于数据资料的获得性,本文仅利用截面数据,考察了自然因素、地理区位等对贫困分布与尺度相关的潜在影响,未能将社会经济因素纳入来深入全面考察“人”“地”相互作用下贫困空间格局的演变规律及其驱动因素。此外,由于数据的最小分辨率单元与块金值直接相关,较大的块金值表明随机因素的影响不可忽视。本文目标于实现贫困的空间总变异中“人”“地”两部分因素影响的定量分割,理论上空间分辨率越高越好(如自然村),在此意义上行政村单元理想但并非最佳。一定程度上,地理资本的质量(如社会资本、教育医疗等公共服务)相比数量和获取成本更为重要,本文主要考虑了后两个方面,进一步考察地理资本质量差异对贫困分布的影响具有重要意义。
The authors have declared that no competing interests exist.

参考文献 原文顺序
文献年度倒序
文中引用次数倒序
被引期刊影响因子

[1]Amarasinghe U, Samad M, Anputhas M.Spatial clustering of rural poverty and food insecurity in Sri Lanka
. Food Policy, 2005, 30(5/6): 493-509.
https://doi.org/10.1016/j.foodpol.2005.09.006URL [本文引用: 3]摘要
We mapped poverty, with reference to a nutrition-based poverty line, to analyse its spatial clustering in Sri Lanka. We used the Divisional Secretariat poverty map, derived by combining the principal component analysis and the synthetic small area estimation technique, as the data source. Two statistically significant clusters appear. One cluster indicates that low poverty rural areas cluster around a few low poverty urban areas, where low agricultural employment and better access to roads are key characteristics. The other indicates a cluster of high poverty rural areas, where agriculture is the dominant economic activity, and where spatial clustering is associated with factors influencing agricultural production. Agricultural smallholdings are positively associated with spatial clustering of poor rural areas. In areas where water availability is low, better access to irrigation significantly reduces poverty. Finally, we discuss the use of poverty mapping for effective policy formulation and interventions for alleviating poverty and food insecurity.
[2]Kam S, Hossain M, Bose M L, et al.Spatial patterns of rural poverty and their relationship with welfare-influencing factors in Bangladesh
. Food Policy, 2005, 30(5-6): 551-567.
https://doi.org/10.1016/j.foodpol.2005.10.001URL [本文引用: 1]摘要
This study determines the spatial variation of rural poverty in Bangladesh and its relation to people livelihood assets affecting their ability to procure food. We estimated household income for over 1 million census households using a predictor model based on a nationally representative sample survey data set. We computed and mapped poverty indices for 415 rural subdistricts revealing distinct areas with high poverty incidence that correspond with ecologically depressed areas. However, other livelihood-influencing factors such as education, accessibility and services are significantly correlated with poverty. This indicates the need for continued focus on providing education and access to income-generating opportunities so that the poor can better meet their food needs. Geographically weighted regression analysis indicated spatial differences in the relative importance of various poverty-influencing factors. Multivariate clustering of the local parameter ( ) estimates of the determinant factors revealed distinct spatial relationships, which have implications on poverty alleviation interventions specific to the different regions.
[3]陈烨烽, 王艳慧, 王小林. 中国贫困村测度与空间分布特征分析
. 地理研究, 2016, 35(12): 2298-2308.
[本文引用: 2]

[Chen Yefeng, Wang Yanhui, Wang Xiaolin.Measurement and spatial analysis of poverty-stricken villages in China
. Geographical Research, 2016, 35(12): 2298-2308.]
[本文引用: 2]
[4]罗庆, 樊新生, 高更和, . 秦巴山区贫困村的空间分布特征及其影响因素
. 经济地理, 2016, 36(4): 126-132.
https://doi.org/10.15957/j.cnki.jjdl.2016.04.018URL [本文引用: 2]摘要
以秦巴山区11县为研究区域,运用GIS技术探讨贫困空间分布格局及演变特征,并对其影响因素进行定量分析.结果表明,秦巴山区贫困村的空间集聚特征较为明显,呈现“大分散、小集中”的格局;与2004年相比,2014年贫困村的空间集聚规模较小,集聚中心数量较多;库区是贫困村的主要集聚地,随着时间推移贫困村的分布呈现远离库区和向乡镇中心附近集聚的趋势.在影响因素分析上,泊松回归的结果表明,自然地理特征、地理区位特征、公共服务设施的可达性和政策因素均对贫困村区位具有显著影响.但随时间推移,贫困村分布的具体影响因素呈现一些新的变化,同一因素的影响大小和影响方向也存在着显著差异.
[Luo Qing, Fan Xinsheng, Gao Genghe, et al.Spatial distribution of poverty village and influencing factors in Qinba Mountains
. Economic Geography, 2016, 36(4): 126-132.]
https://doi.org/10.15957/j.cnki.jjdl.2016.04.018URL [本文引用: 2]摘要
以秦巴山区11县为研究区域,运用GIS技术探讨贫困空间分布格局及演变特征,并对其影响因素进行定量分析.结果表明,秦巴山区贫困村的空间集聚特征较为明显,呈现“大分散、小集中”的格局;与2004年相比,2014年贫困村的空间集聚规模较小,集聚中心数量较多;库区是贫困村的主要集聚地,随着时间推移贫困村的分布呈现远离库区和向乡镇中心附近集聚的趋势.在影响因素分析上,泊松回归的结果表明,自然地理特征、地理区位特征、公共服务设施的可达性和政策因素均对贫困村区位具有显著影响.但随时间推移,贫困村分布的具体影响因素呈现一些新的变化,同一因素的影响大小和影响方向也存在着显著差异.
[5]刘彦随, 周扬, 刘继来. 中国农村贫困化地域分异特征及其精准扶贫策略
. 中国科学院院刊, 2016, 31(3): 269-278.
URL [本文引用: 3]摘要
长期以来,中国坚持政府主导推动减贫事业,在实践中不断推进扶贫开发的理论创新、组织创新和制度创新,走出了一条中国特色的扶贫开发道路,为全球减贫事业做出了巨大贡献。然而,目前中国仍有7 017万农村贫困人口,成为全面建成小康社会的最大短板。文章深入剖析了新时期中国农村贫困化基本特征,揭示了农村贫困化地域分异规律,探明了农村贫困化的主导因素,提出了科学推进精准扶贫的战略与对策。研究结果表明:贫困人口规模大、分布广、贫困程度深、脱贫难度逐渐加大,是当前中国农村贫困状况的基本特征,因病、因残、因学、因灾致贫或返贫现象突出;农村贫困人口逐渐向我国中西部深石山区、高寒区、民族地区和边境地区集聚,具有贫困户、贫困村、贫困县、贫困区(片)等多级并存的组织结构和空间分布格局;"胡焕庸线"西北部、东南部贫困人口的比重分别占16.4%、83.6%;自然环境恶劣、区位条件差、基础设施落后、区域发展不均衡及前期扶贫开发政策精准性不够等,是中国农村持续贫困的主要症结。如期实现2020年全面消除贫困,亟需扶贫工作体制机制的创新,科学推进精准扶贫战略。
[Liu Yansui, Zhou Yang, Liu Jilai.Regional differentiation characteristics of rural poverty and targeted poverty alleviation strategy in China
. Bulletin of Chinese Academy of Sciences, 2016, 31(3): 269-278.]
URL [本文引用: 3]摘要
长期以来,中国坚持政府主导推动减贫事业,在实践中不断推进扶贫开发的理论创新、组织创新和制度创新,走出了一条中国特色的扶贫开发道路,为全球减贫事业做出了巨大贡献。然而,目前中国仍有7 017万农村贫困人口,成为全面建成小康社会的最大短板。文章深入剖析了新时期中国农村贫困化基本特征,揭示了农村贫困化地域分异规律,探明了农村贫困化的主导因素,提出了科学推进精准扶贫的战略与对策。研究结果表明:贫困人口规模大、分布广、贫困程度深、脱贫难度逐渐加大,是当前中国农村贫困状况的基本特征,因病、因残、因学、因灾致贫或返贫现象突出;农村贫困人口逐渐向我国中西部深石山区、高寒区、民族地区和边境地区集聚,具有贫困户、贫困村、贫困县、贫困区(片)等多级并存的组织结构和空间分布格局;"胡焕庸线"西北部、东南部贫困人口的比重分别占16.4%、83.6%;自然环境恶劣、区位条件差、基础设施落后、区域发展不均衡及前期扶贫开发政策精准性不够等,是中国农村持续贫困的主要症结。如期实现2020年全面消除贫困,亟需扶贫工作体制机制的创新,科学推进精准扶贫战略。
[6]罗庆, 李小建. 国外农村贫困地理研究进展
. 经济地理, 2014, 36(6): 1-8.
URL [本文引用: 4]摘要
农村贫困地理作为农村贫困研究的一个重要分支,其研究成果对丰富农村贫困理论和制定农村扶贫政策都具有十分重要的意义。随着普查数据的完善、地理信息技术和遥感技术的应用以及统计分析方法的发展,农村贫困地理研究呈现出一些新趋势和新特征。通过梳理近20年来国外农村贫困研究文献,从地理学视角对空间贫困陷阱的概念及存在性检验、地理因素对农村贫困形成的作用机理、区域贫困的测算及贫困地图的绘制、区域瞄准及效果评估等方面进行系统的回顾和评述,总结出国外农村贫困地理研究的特点与发展方向,提出未来国内相关研究的若干主题。
[Luo Qing, Li Xiaojian.The research progress of foreign rural poverty geography
. Economic Geography, 2014, 36(6): 1-8.]
URL [本文引用: 4]摘要
农村贫困地理作为农村贫困研究的一个重要分支,其研究成果对丰富农村贫困理论和制定农村扶贫政策都具有十分重要的意义。随着普查数据的完善、地理信息技术和遥感技术的应用以及统计分析方法的发展,农村贫困地理研究呈现出一些新趋势和新特征。通过梳理近20年来国外农村贫困研究文献,从地理学视角对空间贫困陷阱的概念及存在性检验、地理因素对农村贫困形成的作用机理、区域贫困的测算及贫困地图的绘制、区域瞄准及效果评估等方面进行系统的回顾和评述,总结出国外农村贫困地理研究的特点与发展方向,提出未来国内相关研究的若干主题。
[7]Bird K, Shepherd A.Livelihoods and chronic poverty in semi-arid Zimbabwe
. World Development, 2003, 31(3): 591-610.
https://doi.org/10.1016/S0305-750X(02)00220-6URL [本文引用: 5]摘要
Remoteness and geographic (natural, physical, human and social) capital are contrasted with social and political exclusion in explaining persistent rural poverty. We found that persistent poverty was strongly associated with the structural poverty of Zimbabwe semi-arid communal areas. Relative urban proximity assisted income diversification and improvement in a very poor, socially and politically excluded area. Less excluded but remote areas remained poor but not as poor as the excluded population. Livelihoods changed and diversified more in the nonremote area, speeding poverty reduction as measured by an index of perceived change. We conclude with what policy options and sequence might support the inclusion of chronically poor people.
[8]Jalan J, Ravallion M.Spatial Poverty Traps?. Washington,
DC: The World Bank, 1997.
[本文引用: 4]
[9]Ravallion M, Wodon Q.Poor areas, or only poor people?
. Journal of Regional Science, 1999, 39(4): 689-711.
https://doi.org/10.1111/0022-4146.00156URL [本文引用: 3]摘要
Instead of targeting poor areas, should poverty programs target households with personal attributes that foster poverty, no matter where they live? Possibly not. There may be hidden constraints on mobility, or location may reveal otherwise hidden householdattributes. Using survey data for Bangladesh, the authors find significant and sizable geographic effects on living standards, after controlling for a wide range of nongeographic characteristics of households, as would typically be observable to policymakers. The geographic effects are reasonably stable over time, robust to testable sources of bias, and consistent with observed migration patterns. Poor areas are not poor just because households with readily observable attributes that foster poverty are geographically concentrated. There appear to be sizable spatial differences in the returns to given household characteristics. Their results reinforce the case for anti-poverty programs targeted to poor areas even in an economy with few obvious impediments to mobility.
[10]Rupasingha A, Goetz S J.Social and political forces as determinants of poverty: A spatial analysis. Journal of
Socio-Economics, 2007, 36(4): 650-671.
https://doi.org/10.1016/j.socec.2006.12.021URL [本文引用: 2]摘要
This study contributes to basic knowledge of the structural determinants of poverty in the US by analyzing an expanded set of determinants of poverty, namely factors related to economic, social, and political influence using spatial data analysis techniques. New data sets and creative use of existing data sets make it possible to measure some of these county-wide social and political factors that have previously been excluded from formal investigation. Social capital, ethnic and income inequality, local political competition, federal grants, foreign-born population, and spatial effects are found to be important determinants of poverty in US counties along with other conventional factors.
[11]Epprecht M, Müller D, Minot N.How remote are Vietnam's ethnic minorities?. An analysis of spatial patterns of poverty and inequality
. The Annals of Regional Science, 2011, 46(2): 349-368.
https://doi.org/10.1007/s00168-009-0330-7URL [本文引用: 1]摘要
AbstractThis paper investigates whether physical accessibility or ethnicity is a stronger determinant of poverty in Vietnam. Spatially disaggregated welfare indexes for population subgroups show that overall inequality is shaped by an urban ural welfare divide, closely followed in importance by sharp welfare differences between ethnic groups. Accessibility to urban areas is a weaker determinant of poverty. The findings have important implications for the targeting of rural development investments. Addressing the factors isolating ethnic minorities from the mainstream economy is likely to be a more useful strategy in reducing rural poverty and inequality than simple geographic targeting.
[12]王永明, 王美霞, 吴殿廷, . 贵州省乡村贫困空间格局与形成机制分析
. 地理科学, 2017, 37(2): 217-227.
URL [本文引用: 1]摘要
以贫困态势严峻、区域内部贫困差异大的贵州省为研究区,分析了贵州省区县层面乡村贫困的空间异质性和空间依赖性格局,定量测度了乡村贫困空间差异的影响因素和因素效应的空间差异性,进而归纳了贵州省乡村贫困的形成机制。结果发现,贵州省区县乡村贫困具有时空稳定性,呈现出东、南、西部高而中、北部低的“马蹄”形空间异质性格局。区县贫困存在较强的空间依赖性,“高-高”型贫困地域即空间贫困陷阱区域,集聚分布在贵州省的东南部、南部。定量模型发现,坡度、到所在市中心的距离、青少年人口占比、少数民族人口占比是导致贵州区县层面乡村贫困空间差异的显著因素,且这些因素的效应水平呈现出不同的空间模式。产业发展受限、劳动力流动性差、金融和人力资本积累不足是贵州贫困空间形成的主导机制。最后建议扶贫政策层面应将基于地方和基于人的政策相结合。
[Wang Yongming, Wang Meixia, Wu Dianting, et al.Spatial patterns and determinants of rural poverty: A case of Guizhou province, China
. Scientia Geographica Sinica, 2017, 37(2): 217-227.]
URL [本文引用: 1]摘要
以贫困态势严峻、区域内部贫困差异大的贵州省为研究区,分析了贵州省区县层面乡村贫困的空间异质性和空间依赖性格局,定量测度了乡村贫困空间差异的影响因素和因素效应的空间差异性,进而归纳了贵州省乡村贫困的形成机制。结果发现,贵州省区县乡村贫困具有时空稳定性,呈现出东、南、西部高而中、北部低的“马蹄”形空间异质性格局。区县贫困存在较强的空间依赖性,“高-高”型贫困地域即空间贫困陷阱区域,集聚分布在贵州省的东南部、南部。定量模型发现,坡度、到所在市中心的距离、青少年人口占比、少数民族人口占比是导致贵州区县层面乡村贫困空间差异的显著因素,且这些因素的效应水平呈现出不同的空间模式。产业发展受限、劳动力流动性差、金融和人力资本积累不足是贵州贫困空间形成的主导机制。最后建议扶贫政策层面应将基于地方和基于人的政策相结合。
[13]Benson T, Chamberlin J, Rhinehart I.An investigation of the spatial determinants of the local prevalence of poverty in rural Malawi
. Food Policy, 2005, 30(5-6): 532-550.
https://doi.org/10.1016/j.foodpol.2005.09.004URL [本文引用: 2]
[14]Okwi P, Ndeng'e G, Kristjanson P, et al. Spatial determinants of poverty in rural Kenya
. Proceedings of the National Academy of Sciences of the United States of America (PNAS), 2007, 104(43): 16769-16774.
https://doi.org/10.1073/pnas.0611107104URL [本文引用: 4]
[15]Ward J, Kaczan D.Challenging Hydrological Panaceas: Water poverty governance accounting for spatial scale in the Niger River Basin
. Journal of Hydrology, 2014, 519: 2501-2514.
https://doi.org/10.1016/j.jhydrol.2014.05.068URL [本文引用: 2]
[16]王政权. 地统计学及在生态学中的应用. 北京: 科学出版社, 1999. [本文引用: 4]

[Wang Zhengquan.The Application of Geostatistics in Ecology .Beijing: Science Press, 1999.] [本文引用: 4]
[17]Zhao K, Liu X, Xu J, et al.Heavy metal contaminations in a soil-rice system: Identification of spatial dependence in relation to soil properties of paddy fields
. Journal of Hazardous Materials, 2010, 181(1): 778-787.
https://doi.org/10.1016/j.jhazmat.2010.05.081URLPMID:20561748 [本文引用: 2]摘要
In order to identify spatial relationship of heavy metals in soil ice system at a regional scale, 96 pairs of rice and soil samples were collected from Wenling in Zhejiang province, China, which is one of the well-known electronic and electric waste recycling centers. The results indicated some studied areas had potential contaminations by heavy metals, especially by Cd. The spatial distribution of Cd, Cu, Pb and Zn illustrated that the highest concentrations were located in the northwest areas and the accumulation of these metals may be due to the industrialization, agricultural chemicals and other human activities. In contrast, the concentration of Ni decreased from east to west and the mean concentration was below the background value, indicating the distribution of Ni may be naturally controlled. Enrichment index (EI) was used to describe the availability of soil heavy metals to rice. The spatial distribution of EIs for Cd, Ni and Zn exhibited a west-east structure, which was similar with the spatial structures of pH, OM, sand and clay. Cross-correlograms further quantitatively illustrated the EIs were significantly correlated with most soil properties, among which; soil pH and OM had the strongest correlations with EIs. However, EI of Cu showed relative weak correlations with soil properties, especially soil pH and OM had no correlations with EI of Cu, indicating the availability of Cu may be influenced by other factors.
[18]周国法. 生物地理统计学. 北京: 科学出版社, 1998. [本文引用: 2]

[Zhou Guofa. Bio-GeographicalStatistics.Beijing: Science Press, 1998.] [本文引用: 2]
[19]张海霞, 牛叔文, 齐敬辉, . 基于乡镇尺度的河南省人口分布的地统计学分析
. 地理研究, 2016, 35(2): 325-336.
https://doi.org/10.11821/dlyj201602010URL [本文引用: 2]摘要
基于地统计学方法,以乡镇层面的第六次人口普查数据为基础,对河南省人口分布及其影响因素进行分析。结果表明:全省人口分布基本上可划分为山区低密度、平原中密度和城市高密度三种类型;变异函数在东西、南北、东北—西南和西北—东南四个方向上趋向相近,距离超过80 km后,各向异性增强;变异函数的指数模型拟合效果最好,插值结果直观地表现了人口疏密的梯度变化;洛阳至淮滨一线为全省较明显的人口分界线;山区和平原两种地形条件是影响该省人口空间格局的基本因素;在平原地区,区位条件和社会经济发展的互动作用又成为影响城镇人口分布的主要因素。在推进城镇化的进程中,应结合主体功能区划和新型城镇化战略,有序推进城乡人口的再分布。
[Zhang Haixia, Niu Shuwen, Qi Jinghui, et al.Geological statistics analysis of population distribution at township level in Henan province
. Geographical Research, 2016, 35(2): 325-336.]
https://doi.org/10.11821/dlyj201602010URL [本文引用: 2]摘要
基于地统计学方法,以乡镇层面的第六次人口普查数据为基础,对河南省人口分布及其影响因素进行分析。结果表明:全省人口分布基本上可划分为山区低密度、平原中密度和城市高密度三种类型;变异函数在东西、南北、东北—西南和西北—东南四个方向上趋向相近,距离超过80 km后,各向异性增强;变异函数的指数模型拟合效果最好,插值结果直观地表现了人口疏密的梯度变化;洛阳至淮滨一线为全省较明显的人口分界线;山区和平原两种地形条件是影响该省人口空间格局的基本因素;在平原地区,区位条件和社会经济发展的互动作用又成为影响城镇人口分布的主要因素。在推进城镇化的进程中,应结合主体功能区划和新型城镇化战略,有序推进城乡人口的再分布。
[20]Trangmar B B, Yost R S, Uehara G.Application of geostatistics to spatial studies of soil properties
. Advances in Agronomy, 1985, 38(1): 45-94.
https://doi.org/10.1016/S0065-2113(08)60673-2URL [本文引用: 5]摘要
This chapter reviews some of the traditional methods of describing soil variability, discusses geostatistical approaches to quantifying spatial dependence and their use for interpolation under different kinds of spatial variation, and identifies some future applications of geostatistics to spatial studies in soil and agronomic research. Recognition of the importance of spatial variability on land use has led to the study of soil heterogeneity, ranging from a global scale to changes in structural and chemical composition of soil minerals on a microscale. Soil classification and soil survey have traditionally been the most practical approaches to grouping similar and separating different soils on a regional scale. Variability of properties within soil mapping units and within smaller sampling units uch as fields, experimental plots, or pedons, is acknowledged and has been described by classical statistical methods. Developments in statistical theory enable spatial relationships among sample values to be quantified and used for interpolation of values at unsampled locations. These developments are based on the theory of regionalized variables. This theory takes into account both the structured and random characteristics of spatially distributed variables to provide quantitative tools for their description and optimal, unbiased estimation. Geostatistical analysis of spatial variability is applied to estimation of ore reserves in the mining industry, water resources research, soil science, and archaeology.
[21]Matheron G.Principles of geostatistics
. Economic Geology, 1963, 58(8): 1246-1266.
https://doi.org/10.2113/gsecongeo.58.8.1246URL [本文引用: 1]
[22]李哈滨, 王政权, 王庆成. 空间异质性定量研究理论与方法
. 应用生态学报, 1998, 9(6): 651-657.
URLMagsci [本文引用: 4]摘要
通过变异函数对空间异质性定量研究进行了讨论。结果表明,空间异质性定量研究应从空间特征和空间比较两方面去考虑。对空间特征,着重讨论怎样应用变异函数将空间异质性分解成各定量组分;确定空间异质性程度;探测空间异质性变化的尺度。对空间比较,怎样对同一变量和不同变量用变异函数比较空间异质性时的统计检验;采用标准化变异函数比较同一地点上的不同变量的空间异质性。最后通过阔叶红松景观中林型和土壤类型的空间异质定量研究实例验证了上述理论与方法。
[Li Habin, Wang Zhengquan, Wang Qingcheng.Theory and methodology of spatial heterogeneity quantification
. Chinese Journal of Applied Ecology, 1998, 9(6): 651-657.]
URLMagsci [本文引用: 4]摘要
通过变异函数对空间异质性定量研究进行了讨论。结果表明,空间异质性定量研究应从空间特征和空间比较两方面去考虑。对空间特征,着重讨论怎样应用变异函数将空间异质性分解成各定量组分;确定空间异质性程度;探测空间异质性变化的尺度。对空间比较,怎样对同一变量和不同变量用变异函数比较空间异质性时的统计检验;采用标准化变异函数比较同一地点上的不同变量的空间异质性。最后通过阔叶红松景观中林型和土壤类型的空间异质定量研究实例验证了上述理论与方法。
[23]Arpino B, Aassve A.The role of villages in households' poverty exit: evidence from a multilevel model for rural Vietnam
. Quality & Quantity, 2014, 48(4): 2175-2189.
https://doi.org/10.1007/s11135-013-9885-6URL [本文引用: 1]摘要
Vietnam experienced a dramatic drop in overall poverty during the 90s. However, the poverty reduction showed substantial variation across households, villages and regions. Using a multilevel model on panel data from the rural sample of the Vietnam Living Standard Measurement Survey we demonstrate the important role of villages in household poverty exit dynamics. We also show how an analysis of village-level random effects predictions can help targeting of policies to reduce poverty.
[24]李小建. 农户地理论. 北京: 科学出版社, 2009. [本文引用: 1]

[Li Xiaojian.Geography of Rural Households. Beijing: Science Press, 2009.] [本文引用: 1]
[25]刘丽娜, 李俊杰. 基于村级尺度的湖北武陵民族地区贫困现状及影响因素研究
. 华中农业大学学报: 社会科学版, 2015, (2): 126-132.
[本文引用: 1]

[Liu Lina, Li Junjie.Research on present situation and affecting factors of poverty based on village scale in Wuling ethnic areas of Hubei province
. Journal of Huazhong Agricultural University: Social Sciences Edition, 2015, (2): 126-132.]
[本文引用: 1]
[26]Park A, Wang S.Community-based development and poverty alleviation: An evaluation of China's poor village investment program
. Journal of Public Economics, 2010, 94(9): 790-799.
https://doi.org/10.1016/j.jpubeco.2010.06.005URL [本文引用: 1]
[27]刘彦随, 李进涛. 中国县域农村贫困化分异机制的地理探测与优化决策
. 地理学报, 2017, 72(1): 161-173.
[本文引用: 1]

[Liu Yansui, Li Jintao.Geographic detection and optimizing decision of the differentiation mechanism of rural poverty in China
. Acta Geographica Sinica, 2017, 72(1): 161-173.]
[本文引用: 1]
[28]Stein A, Brouwer J, Bouma J.Methods for comparing spatial variability patterns of millet yield and soil data
. Soil Science, 1997, 61(3): 861-870.
https://doi.org/10.2136/sssaj1997.03615995006100030021xURL [本文引用: 3]摘要
This paper investigates methods to compare spatial patterns of pearl millet [Pennisetum glaucum (L.) R. Br.] yield with spatial patterns of soil variables in a farmer's 1-ha field on an undulating sand plain in Niger near ICRISAT-SC. Spatial pattern comparisons are important for precision farming applications. Methods included the correlation coefficient, linear regression, a distance measure to compare separate maps and the cross-correlation function. Millet grain yield varied from 0 to 2885 kg ha
[29]Rossi R E, Mulla D J, Journel A G, et al.Geostatistical tools for modeling and interpreting ecological spatial dependence
. Ecological Monographs, 1992, 62(2): 277-314.
https://doi.org/10.2307/2937096URL [本文引用: 3]摘要
Geostatistics brings to ecology novel tools for the interpretation of spatial patterns of organisms, of the numerous environmental components with which they interact, and of the joint spatial dependence between organisms and their environment. The purpose of this paper is to use data from the ecological literature as well as from original research to provide a comprehensive and easily understood analysis of geostatistics' manner of modeling and methods. The traditional geostatistical tool, the variogram, a tool that is beginning to be used in ecology, is shown to provide an incomplete and misleading summary of spatial pattern when local means and variances change. Use of the non-ergodic covariance and correlogram provides a more effective description of lag-to-lag spatial dependence because the changing local means and variances are accounted for. Indicator transformations capture the spatial patterns of nominal ecological variables like gene frequencies and the presence/absence of an organism and of subgroups of a population like large or small individuals. Robust variogram measures are shown to be useful in data sets that contain many data outliers. Appropriate removal of outliers reveals latent spatial dependence and patterns. Cross-variograms, cross-covariances, and cross-correlograms define the joint spatial dependence between co-occurring organisms. The results of all of these analyses bring new insights into the spatial relations of organisms in their environment.
[30]曹诗颂, 赵文吉, 段福洲. 秦巴特困连片区生态资产与经济贫困的耦合关系
. 地理研究, 2015, 34(7): 1295-1309.
https://doi.org/10.11821/dlyj201507009URL [本文引用: 1]摘要
The ecological asset refers to the sum of natural resource value and ecosystem services provided by ecosystem to human beings, which is an important indicator for ecological environment assessment and ecological construction level. Assessment ecological asset for Contiguous Destitute Areas is conducive to the poor areas of ecological environmental protection. Further, study coupling between ecological value of the assets and economic development in poor areas is the basic premise for co-ordinate ecological construction and poverty alleviation. This study took Qinling-Dabashan Region as study areas, and assessed the ecological value of the assets based on quantitative measurement remote sensing method. In addition, this study constructed suitable poor economic evaluation index system from the natural, social and economic development aspects, and evaluated the level of economic poverty of the counties in the study areas. Then, on this basis, this research also built ecological assets-regional economic poverty coordination coupling model, and calculated the coupling coordination degree of ecological assets and regional economic poverty. The results can basically show the actual situation of the extent of ecological assets and economic poverty in the study area. The results showed that the ecological value of the assets in study area was 50.42 billion yuan in 2010 and shrub had the highest ecological value (11.347 billion yuan). For poverty evaluation aspect, the characteristic of natural conditions and other natural factors in the study area are important factors to poverty, and the corresponding social and economic development has brought a significant effect on poverty alleviation. As a whole, the evaluating results of coupling coordination degree indicated that the lower level of ecological assets goes with the lower comprehensive development level of ecological assets and economic poverty as well as the higher level of economic poverty. The ecological assets and economic poverty have symbiotic relationship. This study concluded that improving ecological environment and strengthening management of ecological assets can achieve the goal to reducing poverty in study area and should be added in the policies of poverty reduction. Finally, this research also discussed the interactive mode between ecological protection and poverty reduction in Qinling-Dabashan Region and Support mechanism is put forward from several aspects, which includes Conversion of Cropland to Forest, eco-migration, eco-tourism and ecological construction of cities and towns. This research can provide the reliable guidance for strengthening management of ecological assets and poverty reduction of Contiguous Destitute Areas in Qinling-Dabashan Region.
[Cao Shisong, Zhao Wenji, Duan Fuzhou.Coupling relation analysis between ecological value and economic poverty of contiguous destitute areas in Qinling-Dabashan region
. Geographical Research, 2015, 34(7): 1295-1309.]
https://doi.org/10.11821/dlyj201507009URL [本文引用: 1]摘要
The ecological asset refers to the sum of natural resource value and ecosystem services provided by ecosystem to human beings, which is an important indicator for ecological environment assessment and ecological construction level. Assessment ecological asset for Contiguous Destitute Areas is conducive to the poor areas of ecological environmental protection. Further, study coupling between ecological value of the assets and economic development in poor areas is the basic premise for co-ordinate ecological construction and poverty alleviation. This study took Qinling-Dabashan Region as study areas, and assessed the ecological value of the assets based on quantitative measurement remote sensing method. In addition, this study constructed suitable poor economic evaluation index system from the natural, social and economic development aspects, and evaluated the level of economic poverty of the counties in the study areas. Then, on this basis, this research also built ecological assets-regional economic poverty coordination coupling model, and calculated the coupling coordination degree of ecological assets and regional economic poverty. The results can basically show the actual situation of the extent of ecological assets and economic poverty in the study area. The results showed that the ecological value of the assets in study area was 50.42 billion yuan in 2010 and shrub had the highest ecological value (11.347 billion yuan). For poverty evaluation aspect, the characteristic of natural conditions and other natural factors in the study area are important factors to poverty, and the corresponding social and economic development has brought a significant effect on poverty alleviation. As a whole, the evaluating results of coupling coordination degree indicated that the lower level of ecological assets goes with the lower comprehensive development level of ecological assets and economic poverty as well as the higher level of economic poverty. The ecological assets and economic poverty have symbiotic relationship. This study concluded that improving ecological environment and strengthening management of ecological assets can achieve the goal to reducing poverty in study area and should be added in the policies of poverty reduction. Finally, this research also discussed the interactive mode between ecological protection and poverty reduction in Qinling-Dabashan Region and Support mechanism is put forward from several aspects, which includes Conversion of Cropland to Forest, eco-migration, eco-tourism and ecological construction of cities and towns. This research can provide the reliable guidance for strengthening management of ecological assets and poverty reduction of Contiguous Destitute Areas in Qinling-Dabashan Region.
[31]刘小鹏, 苏胜亮, 王亚娟, . 集中连片特殊困难地区村域空间贫困测度指标体系研究
. 地理科学, 2014, 34(4): 447-453.
URLMagsci [本文引用: 1]摘要
<p>在阐述多维贫困和空间贫困概念内涵及其指标基础上,提出了集中连片特殊困难地区村域空间贫困测度指标体系构建的基本原则,即强调科学性和主导性原则、重视数据的可获得性和测度的可操作性、体现减贫与反贫困的新要求、突出区域性和空间刻画能力。据此,构建了包括经济、社会、环境和政策4 个维度,收入和消费、市场连通性、人口状况、学有所教、病有所医、老有所养、住有所居、劳有所得、地貌要素、自然灾害、生态安全、农业生态、粮食安全和政策的实效性共13 个指标组,27 个原始指标或生成指标构成的集中连片特殊困难地区村域空间贫困测度指标体系。进一步讨论了空间贫困测度指标的检验、获取方法和空间化等关键问题。</p>
[Liu Xiaopeng, Su Shengliang, Wang Yajuan, et al.The index system of spatial poverty of village level to monitor in concentrated contiguous areas with particular difficulties
. Scientia Geographica Sinica, 2014, 34(4): 447-453.]
URLMagsci [本文引用: 1]摘要
<p>在阐述多维贫困和空间贫困概念内涵及其指标基础上,提出了集中连片特殊困难地区村域空间贫困测度指标体系构建的基本原则,即强调科学性和主导性原则、重视数据的可获得性和测度的可操作性、体现减贫与反贫困的新要求、突出区域性和空间刻画能力。据此,构建了包括经济、社会、环境和政策4 个维度,收入和消费、市场连通性、人口状况、学有所教、病有所医、老有所养、住有所居、劳有所得、地貌要素、自然灾害、生态安全、农业生态、粮食安全和政策的实效性共13 个指标组,27 个原始指标或生成指标构成的集中连片特殊困难地区村域空间贫困测度指标体系。进一步讨论了空间贫困测度指标的检验、获取方法和空间化等关键问题。</p>
[32]史坤博, 杨永春, 白硕, . 成都市体验性网络团购市场发展的空间特征
. 地理研究, 2016, 30(1): 108-122.
https://doi.org/10.11821/dlyj201601010URL [本文引用: 1]摘要
体验性网络团购模式正在对城市商业空间产生显著影响.以团购商品信息量作为基础数据,采用矢量数据符号法和空间插值法对成都市团购市场发展的空间特征进行分析.结果表明:餐饮、娱乐、生活和旅游酒店类团购信息在各商圈分布的集中化程度与摄影类相比较低.成都市中心区域的团购市场发展水平较高,形成了以春熙路商圈为核心的综合商业型热点区和以建设路商圈和双楠商圈为核心的生活服务型热点区.市场原则下团购市场发展的空间公平性较好的区域主要集中在成都市中心区域.团购市场发展的空间格局具有明显的区位特征,其空间发展的驱动力主要包括实体商业空间布局、消费者规模空间格局和区域可达性等;“时间距离”对团购市场区位特征产生的摩擦阻力已经超过空间距离.
[Shi Kunbo, Yang Yongchun, Bai Shuo, et al.Spatial characteristics of the experiential online group-buying market in Chengdu
. Geographical Research, 2016, 30(1): 108-122.]
https://doi.org/10.11821/dlyj201601010URL [本文引用: 1]摘要
体验性网络团购模式正在对城市商业空间产生显著影响.以团购商品信息量作为基础数据,采用矢量数据符号法和空间插值法对成都市团购市场发展的空间特征进行分析.结果表明:餐饮、娱乐、生活和旅游酒店类团购信息在各商圈分布的集中化程度与摄影类相比较低.成都市中心区域的团购市场发展水平较高,形成了以春熙路商圈为核心的综合商业型热点区和以建设路商圈和双楠商圈为核心的生活服务型热点区.市场原则下团购市场发展的空间公平性较好的区域主要集中在成都市中心区域.团购市场发展的空间格局具有明显的区位特征,其空间发展的驱动力主要包括实体商业空间布局、消费者规模空间格局和区域可达性等;“时间距离”对团购市场区位特征产生的摩擦阻力已经超过空间距离.
[33]郭晓东. 乡村聚落发展与演变: 陇中黄土丘陵区乡村聚落发展研究. 北京: 科学出版社, 2013. [本文引用: 1]

[Guo Xiaodong.Development and Evolution of Rural Settlements: A Case in Loess Hilly Area of Gansu Province. Beijing: Science Press, 2013.] [本文引用: 1]
[34]Cambardella C A, Moorman T B, Parkin T B, et al.Field-scale variability of soil properties in central iowa soils
. Soil Science Society of America Journal, 1994, 58(5): 1501-1511.
https://doi.org/10.2136/sssaj1994.03615995005800050033xURL [本文引用: 2]摘要
DIRECT EXTRACTION METHOD; SPATIAL VARIABILITY; MICROBIAL BIOMASS; CHLOROFORM FUMIGATION; NITROGEN; CARBON; DENITRIFICATION; GEOSTATISTICS; RESPIRATION; RELEASE
[35]Minot N, Baulch B.Spatial patterns of poverty in Vietnam and their implications for policy
. Food Policy, 2005, 30(5/6): 461-475.
https://doi.org/10.1016/j.foodpol.2005.09.002URL [本文引用: 1]摘要
This study examines the geographic distribution of poverty in Vietnam by applying small area estimation methods to household budget data and population census data. The resulting district-level poverty estimates suggest that the incidence of poverty is highest in the remote northern and central highlands and lowest in the south-east and in large urban centres. However, mapping the density of poverty reveals that most poor people do not live in the poorest districts but in the two lowland deltas, where poverty incidence is intermediate. The policy implications of these findings present an important trade-off between targeting poor areas and poor people that can only be resolved with better information on the relative costs of delivering different programmes and their expected impact. Existing government estimates of poverty at the district level are not closely correlated with our poverty estimates, perhaps because of regional variation in their methods of collecting poverty data.
[36]李小建, 周雄飞, 郑纯辉. 河南农区经济发展差异地理影响的小尺度分析
. 地理学报, 2008, 63(2): 147-155.
https://doi.org/10.3321/j.issn:0375-5444.2008.02.004URL [本文引用: 1]摘要
对中国第一农业大省河南的乡镇数据分析表明。地理因素对农区经济发展具重要影响。在较低经济发展水平的乡镇,地形和农业资源条件具有显著的影响;而在相对较高经济发展水平的乡镇,地理位置以及与此相伴生的基础设施条件的影响更为显著。模型分析表明,地形显著影响乡镇的人均收入,而且随着收入的增加,其影响程度递减。人均土地面积显著影响非农产业发展水平较低的乡镇的人均收入。对非农产业比重低于20%的乡镇,人均土地每增加1hm^2,人均收入就增加约96元。地理位置影响着相关农区经济的发展,尤其是对收入较高的农区的影响十分明显。全省高收入乡镇的70%集中在河南中部地区,正好与中原城市群的空间范围相吻合。就单个乡镇而论,离县城的距离显著影响高收入乡镇的人均收入。地理因素对欠发达农区经济发展影响处于变动之中。随着经济水平的提高,传统地理因素(如地形、农业资源因素等)的影响逐渐被对经济积聚有明显影响的地理临近性等要素的影响所取代。因此,不能简单的认为地理因素对经济发展的影响在下降。
[Li Xiaojian, Zhou Xiongfei, Zheng Chunhui.Geography and economic development in rural China: A township level study in Henan province, China
. Acta Geographica Sinica, 2008, 63(2): 147-155.]
https://doi.org/10.3321/j.issn:0375-5444.2008.02.004URL [本文引用: 1]摘要
对中国第一农业大省河南的乡镇数据分析表明。地理因素对农区经济发展具重要影响。在较低经济发展水平的乡镇,地形和农业资源条件具有显著的影响;而在相对较高经济发展水平的乡镇,地理位置以及与此相伴生的基础设施条件的影响更为显著。模型分析表明,地形显著影响乡镇的人均收入,而且随着收入的增加,其影响程度递减。人均土地面积显著影响非农产业发展水平较低的乡镇的人均收入。对非农产业比重低于20%的乡镇,人均土地每增加1hm^2,人均收入就增加约96元。地理位置影响着相关农区经济的发展,尤其是对收入较高的农区的影响十分明显。全省高收入乡镇的70%集中在河南中部地区,正好与中原城市群的空间范围相吻合。就单个乡镇而论,离县城的距离显著影响高收入乡镇的人均收入。地理因素对欠发达农区经济发展影响处于变动之中。随着经济水平的提高,传统地理因素(如地形、农业资源因素等)的影响逐渐被对经济积聚有明显影响的地理临近性等要素的影响所取代。因此,不能简单的认为地理因素对经济发展的影响在下降。
相关话题/空间 地理 数据 指标 农村