中文摘要：

为研究土地利用空间自相关分析中研究对象观测变量和衡量指标的选择问题,以求较优的空间自相关格局。以临夏县北塬地区的农村居民点用地为研究对象,首先提取地形因子来分析区域地形变化特征,预测空间自相关格局,以此为依据对空间自相关分析中选择的观测变量和衡量指标进行对比分析。结果显示：全局Moran指数优于全局G系数;局部Moran指数对高值聚集识别较优,可识别异常特征;全局G系数对低值聚集识别较优;居民点分散度反映的空间自相关格局与地形特征耦合较好。文中研究可为后续空间自相关研究提供参考。

英文摘要：

The purpose of studying the optimal selection of observed variables and measuring indicators was to get better spatial autocorrelation pattern during implementing land use spatial autocorrelation analysis. Taking the rural residential land of Beiyuan region in Linxia County as the research object,the terrain factors were extracted to analyze terrain characteristics and predict spatial autocorrelation pattern in the region,then those selected observed variables and measuring indicators were analyzed and compared according to the predicted spatial autocorrelation pattern. The results showed that global Moran＇s index was better than Getis- Ord general G. Local Moran＇s index identified high value gather preferably and could identify anomaly characteristics. Local Getis- Ord G identified low value gather preferably. The coupling between the characteristics of the spatial autocorrelation pattern reflected by residential dispersion degree variable and the terrain feature was well. This paper can provide reference for the follow- up study of spatial autocorrelation.