Journal of Geographical Sciences

Incorporation of intra-city human mobility into urban growth simulation: A case study in Beijing

WANG Siying, FEI Teng, LI Weifeng, ZHANG Anqi, GUO Huagui, DU Yunyan

地理学报(英文版), 2022, 32(5): 892-912. DOI: 10.1007/s11442-022-1977-6

Spatial variables	Category	Explanation	Data source
y	Binary	Urban growth 2013-2016	Landsat 8 Operational Land Imager images
DisToMainRoad	Float	Euclidean distance to main roads	Road network map
DisToSecondRoad	Float	Euclidean distance to secondary roads	Road network map
DisToWaterArea	Float	Euclidean distance to a water area	Map of water area
DisToUrbanArea	Float	Euclidean distance to an urban area	VIIRS Day/Night Band (DNB) Nighttime Imagery
DEM	Float	Transition suitability considering terrain conditions data	Global Digital Elevation Model (ASTGTM)
Slope	Float	Transition suitability considering terrain conditions data	Global Digital Elevation Model (ASTGTM)
Neighborhood	Float	Amount of urban cells in 5·5 neighborhood	Landsat 8 Operational Land Imager images

Table 1 Model inputs and their data source

本文的其它图/表

Figure 1 The framework of the ANN-CA model in the integration of human mobility factors
Figure 2 Illustration of the stay point detection process
Figure 3 An example of a directed graph formed by human trajectories
Figure 4 The architecture of the ANN-CA model
Figure 5 Administration divisions of Beijing
Figure 6 The spatial variables in Beijing (a) slope, (b) DEM, (c) distance to urban area, (d) distance to the main road, (e) distance to the secondary road, and (f) distance to water area
Figure 7 The flow map of trajectory data in Beijing
Figure 8 The inflow magnitude distribution (a) and the outflow magnitude distribution (b) of Beijing
Figure 9 The observed urban growth from 2013 to 2016 in Beijing (a) and the simulated pattern in 2016 based on the four proposed models: (b) ANN-CA_withflow, (c) ANN-CA_withoutflow, (d) Logistic-CA_withflow, and (e) Logistic-CA_withoutflow
Table 2 Assessment of the simulation results
Figure 10 Resultant improvements of Kappa coefficients and FoM values of the simulation results for districts of Beijing based on ANN-CA_withflow model
Figure 11 The plots of training accuracy and loss
Figure 12 Flow map and an enlarged view of Beijing
Table 3 A comparison of the elevation of the four selected districts of Beijing