Urban Population Spatialization Based On High-resolution Remotely Sensed Data

Urban population spatialization is one of most significant aspects of urban remote sensing,which is the index of the level of urbanization,capacity of environmental resources and development of urban economics.Besides,census data can characterize the spatial distribution of urban citizens,which is authoritative and consecutive to some extent.However,it is not time-effective since the period is always ten years with great load of field work and human interference.Besides,it cannot display the fine-scale population distribution to a large degree.Population distribution data in finer resolution could be used as an important basis for the assessment of natural emergency,dynamics of population,urban planning and social-economic development.According to research,there are four problems pertaining to urban population estimation: 1.Simple population statistical models,such as exponential jump-diffusion model based on spatial distance,are easy to conduct,but the errors are magnificent.2.Population distribution models based on statistics,such as the model based on the number of residential buildings,whose accuracy are almost dependent on the detail of statistics with strong subjectivity.3.Many researchers stagnate their study on land use types or building area,since they believe most people stay in the buildings,which causes the problem that few researchers study on three dimensional of living space and it is hard to estimate the real population distribution especially in the dense building area;4.Other research has taken the three-dimension information of buildings into consideration,however,most of them based on accurate statistical data.Therefore,the scope of research area is very limited,and it is difficult to obtain information of fine-scale population distribution in relatively large urban area.Meanwhile,other scholars did some research on fine-scale population distribution with satisfactory results to some extent,whereas there are three problems during the process of extracting building height.1.LiDAR data can be utilized to obtain the height information accurately in an area.Nevertheless,the cost is high with predefined research areas and no returning cycle or timeliness.2.Measuring building height with certain equipment is useful to compute building height in certain areas,but the process is complicated,which is inconvenient for collecting data in a relatively large area with an assured periodicity.3.Image fusion between remotely sensed images,such as optical imagery and radar imagery,and DEM to extract building height information is another way,which can get the relevant three-dimension information in building areas with great accuracy,but the precise geometrical registration is hard to guarantee.Overall,in this experiment,we choose two core areas i.e.Xicheng District,Dongcheng District,and four urban function extension areas,i.e.Chaoyang District,Haidian District,Shijingshan District and Fengtai District,as the main research areas and compare three different methods,namely photogrammetry based on ZY-3 stereo images,shadow length based on the geometrical relationship between the sun and the satellite and optimal combination of backscattering model based on SAR with Rardarsat-2 images,in calculating building height.By comparing the accuracy,feasibility and effectiveness of the three approaches above,the most suitable one is chosen to calculate the building height in research area when the specific characteristics of research area is taken into account.Then,morphological building index(MBI)could be computed based on high-resolution images to extract buildings with object-based methods,and final population distribution results can be gotten in research area with the help of deterministic three-dimension population model.This research explores the fine-scale population estimation within downtown areas preliminarily,which lays the foundation for the nation to make broad population policy and serves for management of urban land resources,distribution of resources and emergency response.