县域粮食单产及其生产投入因素的空间异质性——以河南省为例

在当前耕地总量很难增加的总体形势下,粮食单产及其生产投入因素的关系研究对国家粮食安全具有重要意义。Moran’s I指数及散点图显示,粮食单产及其OLS(Ordinary Least Squares)估计的残差存在显著的空间自相关性,因而不满足经典线性回归分析的建模条件,而GWR(Geographically Weighted Regression)模型能够克服这一缺陷,并能进行参数的局部估计。河南省108个县的实践表明,GWR模型有效地降低了残差的空间自相关,各项检验指标均好于OLS估计。4个生产投入影响因素的空间分异规律存在显著差异。灌溉因素在OLS和GWR模型中均为正效应,单位播种面积机械动力、单位播种面积化肥施用量(折纯)、单位播种面积用电量3个变量的系数估计值在OLS模型中为负,而在GWR模型中则有正有负,说明存在空间异质性。黄河以北粮食高产区主要受机械动力、灌溉和用电量3个因素的正向影响,化肥投入的再增加反而引致单产降低。淮河沿岸和信阳的粮食高产区主要受化肥施用量和用电量的正向影响。为提高粮食单产,西部山地、丘陵县的粮食低产区应加强机械动力投入并提高用电效率,最西部的山地县还应适度增加化肥的用量。

Spatial Heterogeneity of Grain Yield per Hectare and Factors of Production Inputs in Counties: A Case Study of Henan Province

In the current situation that is difficult to increase the total arable land area,the relationship between grain yield and the factors of production inputs is important for national food security.Global Moran's I index of grain yield per hectare in Henan was 0.6921,which indicates a strong spatial autocorrelation.Seen from Moran scatter plot,grain yield per hectare in Henan was the high-high cluster pattern.Moreover,there was significant spatial autocorrelation on the residual of ordinary least squares esti...