Research On Gridding Method Of Buildings’ Spatial Distribution

China is one of the most serious victioms of earthquake disasters in the world. Strong earthquakes cause great harm to people’s life and property security. Many devastating earthquakes have occurred since twenty-first century in our country, such as the Wenchuan earthquake, the Lushan earthquake, the Ludian earthquake and so on. According to some statistics, seismic casualties caused by the damage and collapse of buildings accounted for 95% of the whole number of earthquake casualties. At present, researchers usually adopt the uniform distribution method to generate building spatial distribution which has a great difference with the actual situation in the process of earthquake risk and post-earthquake assessment in our country. The building spatial distribution result processing by the old method cannot provide effective basis for earthquake disaster prevention and post-earthquake rescue. With the rapid development of Remote Sensing(RS) and Geographic Information System(GIS) technology, it is important to develop gridding method of building spatial distribution based on RS and GIS technology.Firstly, this paper summarized the research on gridding method of population spatial distribution which has been widely studied, and it evaluated advantages and disadvantages of each method. Then it overviewed the gridding meghod of building spatial distribution in the recent years, found the shortcomings of current method and put forward the solutions. On this basis, it introduced research idea and discusses the gridding method of building spatial distribution in detail.Secondly, with the support of statistical data, basic geographic information data and remote sensing data, six natural geographical and social economic factors were selected as the factors of building spatial distribution which include elevation, slope, aspect, river, road and land use. Then we integrated land use data into urban residential area, township residential area, rural residential area, other building area and non-residential area. Taking the GIS spatial analysis and statistical analysis as the main means, we studied these factors’ impact on building spatial distribution respectively in five regions and establish the building spatial distribution weight model. Then we distributed the building areas of statistical unit to all grids inside it according to their weights. At this point building spatial distribution models in Dongchuan and Tianshui have been completed. After that, we applied these two models to the surrounding areas and evaluate the effectiveness of each model based on the prediction results.Thirdly, we applied these two models to 100 counties in North-South seismic zone. The gridding model of building spatial distribution was selected by principal component analysis and clustering method. Based on principal component analysis method, we selected six indexes include reclamation index, elevation, road density, forest area ratio, grassland area ratio, residential area ratio from the whole ten indexes. Considering the spatial distance between county resident and study area, we choose minimum distance method which as important in cluster analysis to select models. On this basis, we predict the building spatial distribution of 100 counties according to quantitative parameters of Dongchuan and Tianshui model.Summaries and prospects were proposed at the end of this paper. Due to the studies about gridding method of building spatial distribution are few, we only had a preliminary study. So there are some deficiencies:(1) when we build the model, factors selection have certain subjectivity, but we did not take the correlation between factors into account.(2) The difference of building spatial distribution does not reflect in the inner city resident. These two aspects need to be further studied.