中文摘要：

运用高分辨率遥感数据源,从聚落入手,采用聚落调查法和GIS空间分析方法,在小尺度上建立基于研究区普定后寨河地区的总体情况、地形地貌差异、聚落等级的岩溶山区乡村聚落空间规模与人口数之间的相关关系,并在此基础上对其4个时期独立聚落人口动态数量进行回归模拟。步骤为：1）建立研究区已获取人口数的聚落空间分布规模（面积）与人口数的一元线性相关关系,对研究区聚落与人口进行回归分析,推算出研究区各时期无法获取准确数据的独立聚落人口数;2）建立各地貌区聚落空间规模与人口的线性关系,推算基于地貌类型无法获取准确数据的聚落人口数;3）考虑聚落空间规模的等级效应,将研究区各时期聚落按空间分布规模划分为高、中、低3个规模等级,计算不同等级内已获取人口数聚落的人均聚落面积,然后推算未知人口数聚落的人口数;4）在上述三种推算方式的基础上,使用平均法计算各时期每个独立聚落人口数。通过对比县志、人口资料和实地调查,证明通过聚落与人口数之间的相关关系推算聚落人口数是基本有效的、准确的。

英文摘要：

Access to the data of population number is of great significance for in-depth analysis of man-land relationship, but conventional techniques to obtain the data of the spatial distribution of population demand a lot of manpower, material and financial resources. In this paper, the Houzhai River area is taken as an example to be studied, where the relationship between population number and spatial scale of the settlements in Karst mountains is established based on high resolution remote sensing data source and the method of settlement survey and GIS spatial analysis. According to different physiognomy types and hierarchy of the settlements, regressive analysis is made on the dynamic population quantity of independent settlements in four periods.The steps are： 1） establish linear relationship between the population and the size （area） of the settlements to calculate the population of independent settlements whose data can not be obtained accurately in the four periods; 2） establish linear relationship between the population and the size of the settlements in different physiognomy areas to obtain the data of the population spatial distribution; 3） classify the settlement space distribution scales of different periods into high, medium and low levels according to the hierarchical effect of the settlements, calculate per capita area of the settlements whose population data have been obtained, and then calculate the population number of the settlements whose data are lacking; 4） calculate the average population number of each individual settlement in each period on the basis of the three methods above. The results calculated in this paper are proved to be basically effective and accurate, as compared with the data from county record, population data and field investigation.