Research On Multi-level Fine-scale Spatialization Of Population Based On Building Footprints

With the continuous improvement of urban management efficiency and manageme nt level,the role of population distribution data is constantly highlighted,which is of gr eat significance for the comprehensive governance of various sectors of society.At pres ent,fine-grained spatialization methods combine multi-source geographic data using ma chine learning algorithms,but obtaining multi-source geographic data is difficult,and m ultiple types of data are difficult to match each other;Machine learning methods have o verfitting and lack interpretability.This article proposes a population refinement spatialization method based on buildi ng contour data,which converts the existing constrained unit population into building u nit based population data.This method divides the geographical space and calculates the number of people in different spatial units based on existing constrained unit populatio n data,assigning them to buildings.Compared with previous methods,this method take s buildings as the only auxiliary data and explores their geometric relationships with var ious spatial units to obtain a refined spatial distribution of population.However,in this s cheme,the scale and shape of the spatial units directly affect the population reduction pr ocess of the constraint units.This article takes the main urban area of Wuhan as an example and conducts a mult i-scale fine spatial experiment.In order to determine the appropriate spatial scale,the ge ographical space is divided into five different squares: 100 meters,200 meters,and 2000 meters.In order to analyze the impact of spatial unit morphology,Voronoi diagrams w ere introduced for irregular partitioning.We have selected existing constrained unit pop ulation data to allocate and calculate the population of each spatial unit.By utilizing geo metric constraints,the population in the spatial units was allocated to various buildings.Obtained population distribution data including 7 spatial division results,7 spatial unit p opulation distributions,and 7 spatial units converted to buildings.Through comparative analysis,the following conclusions are drawn.(1)Establishing a volume weight model to allocate the constrained unit population to buildings improves the spatial resolution of population data;Introducing data from di fferent land types and comparing the proportion of population in different land types bef ore and after the experiment,it was found that Weighted v increased the proportion of p opulation in residential land from about 60% to 73%,which is better than other spatial u nits.(2)Comparing SDPD and H ’index,it is found that the H’ value of 100 m constraint unit is the largest,the SDPD index value of 200 m constraint unit is the largest,followe d by 100 m,indicating that they are more appropriate in expressing population spatial he terogeneity.Introducing CM and SIR indicators to compare the 100 and 200 m squares with Simple v and Weighted v,it was found that the Weighted v indicators in 200 m and Voronoi in the squares have larger values,making them more suitable for characterizing population distribution patterns.By comparing and analyzing the impact of different la nd types and population density zones on the results,it is concluded that the accuracy of experimental results is better in areas with dense central buildings and areas with high p opulation density in cities.In summary,the population fine spatialization method based on buildings proposed in this article can effectively perform fine spatialization on existing constrained unit population distribution data.We hope that this study can provide a certain basis for producing high-quality and refined population data.