TY - JOUR
T1 - Monitoring finer-scale population density in urban functional zones
T2 - A remote sensing data fusion approach
AU - Song, Jinchao
AU - Tong, Xiaoye
AU - Wang, Lizhe
AU - Zhao, Chunli
AU - Prishchepov, Alexander V.
PY - 2019
Y1 - 2019
N2 - Spatial distribution information on population density is essential for understanding urban dynamics. In recent decades, remote sensing techniques have often been applied to assess population density, particularly night-time light data (NTL). However, such attempts have resulted in mapped population density at coarse/medium resolution, which often limits the applicability of such data for fine-scale territorial planning. The improved quality and availability of multi-source remote sensing imagery and location-based service data (LBS) (from mobile networks or social media) offers new potential for providing more accurate population information at the micro-scale level. In this paper, we developed a fine-scale population distribution mapping approach by combining the functional zones (FZ) mapped with high-resolution satellite images, NTL data, and LBS data. Considering the possible variations in the relationship between population distribution and nightlight brightness in functional zones, we tested and found spatial heterogeneity of the relationship between NTL and the population density of LBS samples. Geographically weighted regression (GWR) was thus implemented to test potential improvements to the mapping accuracy. The performance of the following four models was evaluated: only ordinary least squares regression (OLS), only GWR, OLS with functional zones (OLS&FZ) and GWR with functional zones (GWR&FZ). The results showed that NTL-based GWR&FZ was the most accurate and robust approach, with an accuracy of 0.71, while the mapped population density was at a unit of 30 m spatial resolution. The detailed population density maps developed in our approach can contribute to fine-scale urban planning, healthcare and emergency responses in many parts of the world.
AB - Spatial distribution information on population density is essential for understanding urban dynamics. In recent decades, remote sensing techniques have often been applied to assess population density, particularly night-time light data (NTL). However, such attempts have resulted in mapped population density at coarse/medium resolution, which often limits the applicability of such data for fine-scale territorial planning. The improved quality and availability of multi-source remote sensing imagery and location-based service data (LBS) (from mobile networks or social media) offers new potential for providing more accurate population information at the micro-scale level. In this paper, we developed a fine-scale population distribution mapping approach by combining the functional zones (FZ) mapped with high-resolution satellite images, NTL data, and LBS data. Considering the possible variations in the relationship between population distribution and nightlight brightness in functional zones, we tested and found spatial heterogeneity of the relationship between NTL and the population density of LBS samples. Geographically weighted regression (GWR) was thus implemented to test potential improvements to the mapping accuracy. The performance of the following four models was evaluated: only ordinary least squares regression (OLS), only GWR, OLS with functional zones (OLS&FZ) and GWR with functional zones (GWR&FZ). The results showed that NTL-based GWR&FZ was the most accurate and robust approach, with an accuracy of 0.71, while the mapped population density was at a unit of 30 m spatial resolution. The detailed population density maps developed in our approach can contribute to fine-scale urban planning, healthcare and emergency responses in many parts of the world.
KW - Geographically weighted regression
KW - Land use
KW - LBS
KW - Spatial heterogeneity
KW - Urban functional zone
U2 - 10.1016/j.landurbplan.2019.05.011
DO - 10.1016/j.landurbplan.2019.05.011
M3 - Journal article
AN - SCOPUS:85066312662
VL - 190
JO - Landscape and Urban Planning
JF - Landscape and Urban Planning
SN - 0169-2046
M1 - 103580
ER -