Abstract: The information of spatial distribution of population plays a significant role in the studies on resource environment, social economics, evaluation of the loss caused by a natural disaster, land use change, and other topics in geography and related disciplines. Traditional method hypothesizes that a population is distributed uniformly in a region, but the actual situation is not like that. Spatialization of census data for a population becomes rather important for a comprehensive analysis which combines social economics with natural environment. Therefore, the research on spatialization of census data has become a hot spot in geographic science and other social sciences. In this paper, Taips County, a typical region in the agro-pastoral zone of North China, was taken as a study case. This region has been heavily affected by the Grain-for-Green Project in China, and the population in the region has changed dramatically since the implementation of the project. Based on the characteristics of each of the 175 administrative villages in the region in 2000 and 2008, different scales were applied in the analysis. Through multi-variables regression analysis of the population's census data and various impacting factors, including land use indexes, topographical indices (mean elevations and mean slopes), and distance to main roads and rivers at the village's level in Taips County, using GIS software and SPSS statistical software as the tools, a model for the spatial distribution of population was established. In the meantime, the actual population density of each administrative village was used to validate the precision of the model. In this study, the number of independent variables was gradually increased to explore the model of the population's spatialization to achieve higher precision and make the model more suitable to the study area. It was found that there was a significant correlation between population density and land use type of each administrative village. The correlation ratio between actual administrative village's population density and the density calculated by spatialization model reached to 0.961 and 0.881 in 2000 and 2008, respectively, and the linear fitting slopes of both simulation results were close to 1. These results indicated that the spatialization model worked very well for simulation, and the accuracy satisfied the application of the model to the research on population's spatial distribution in small scales. And also, dividing spatial scales can improve the precision. In addition, the population in the region has changed dramatically during the 8 years. The population grew rapidly near the town center and in the suburban areas but dropped sharply in the other areas, indicating the trend that the population became concentrated near the town center and the surrounding areas. In conclusion, Grain-for-Green Project is one of the most important driving factors of the change of the spatial pattern of population in the regional scale.

Key words: agro-pastoral zone, GIS, Grain-for-Green Project, multi-variables regression, population spatialization, Taips County