Research On Spatialization Methods Of Urban Population At Different Scales Based On Multi-source Data

When a destructive disaster occurs,the relevant departments need to obtain the macro judgment and loss assessment results of the disaster situation in a short period of time,and formulate corresponding emergency rescue plans.These tasks are inseparable from the support of the disaster preparedness geodatabase.However,the population data in the existing disaster preparedness geodatabase is usually the statistical data of the administrative division as the basic unit.It can only reflect the overall situation within the unit.The data has low refinement,low spatial and temporal resolution,and is inconsistent with the geographic unit.These issues make it difficult to integrate spatial analysis with other multi-source data.The spatialization of population data can realize the format conversion of administrative unit population data,and express the distribution and quantity of population in spatial position in an intuitive form.The current population spatialization distribution study does not fully cons ider the scale effect of population distribution,and most of the studies use land use/land cover data as the data source for population data spatialization,but the existing land use/land cover data are the results of land classification obtained through the interpretation of low-and medium-resolution remote sensing images,and the classification results are less accurate.In view of the above problems,I proposed a method for spatial distribution of population data from three different scales,and research on population distribution at three scales of macro scale,micro scale and more refined scale dynamic distribution.The main research works of this paper are as follows:(1)Spatial distribution of population at the macro scale.In this study,the correlation analysis between the land coverage type area index and the population density is firstly carried out to determine the modeling factor.On this basis,using the partition factor for principal component analysis to obtain the comprehensive impact factor,the feature area of the study area is consistently partitioned,and then the multivariate linear regression method combined with the modeling factor to Spatialization model of population data in kilometer grid scale,and the error analysis is compared with the overall modeling.Experimental results surface: The spatial data of the population modeled by partition modeling is higher than the accuracy of the population model data of the overall model.(2)Spatial distribution of population at the micro sca le.This study uses 2018 remote sensing image with high resolution of 0.8 meters to carry out construction land information extraction.Then introduce the POI data of the same year,study its correlation with population density,select relevant POI type da ta,analyze the spatial aggregation characteristics of each POI type data,and finally realize the population redistribution of 200-grid scale construction land based on POI data.(3)Time-space dynamic distribution of population with fine scale.This study uses distributed web crawling to obtain the Sina Weibo data of the Southwest Jiaotong University(Xipu Campus)library around 3 km.After data preprocessing,the vector map of the campus of Southwest Jiaotong University was combined with the experimental area.The time distribution and spatial distribution of Weibo data are comprehensively analyzed.Finally,the kernel density estimation method is used to time-segment the 24 hours a day to explore the spatial and temporal distribution characteristics of the student population,and to obtain the concentrated areas of students at different times and at different locations.The results show that the spatial distribution pattern of populations can be obtained from different scales,and population information w ith different levels of detail can be obtained.From the macro-scale analysis,the spatial distribution characteristics of the population within the counties of Sichuan Province can be obtained;from the micro-scale analysis,the spatial distribution chara cteristics of the population within the construction land of each county can be obtained;from the fine-scale analysis,the spatial-temporal dynamics of the experimental population can be obtained.Distribution characteristics.And macro-scale population spatial data and micro-scale population spatial data can be added to the disaster preparedness geodatabase to enrich the data sources of the disaster preparedness geodatabase,providing a reliable and more refined population data for the future disaster are as in Sichuan.The research on the spatio-temporal dynamic distribution of refined population based on micro-blog data can be applied to actual disasters,and get a more accurate population spatial data,and then efficient and reasonable rescue work can be carried out.