Examining Spatial Patterns of Urban Distribution and Impacts of Physical Conditions on Urbanization in Coastal and Inland Metropoles

Abstract

Urban expansion has long been a research hotspot and is often based on individual cities, but rarely has research conducted a comprehensive comparison between coastal and inland metropoles for understanding different spatial patterns of urban expansions and driving forces. We selected coastal metropoles (Shanghai and Shenzhen in China, and Ho Chi Minh City in Vietnam) and inland metropoles (Ulaanbaatar in Mongolia, Lanzhou in China, and Vientiane in Laos) with various developing stages and physical conditions for examining the spatiotemporal patterns of urban expansions in the past 25 years . Multitemporal Landsat images with 30 m spatial resolution were used to develop urban impervious surface area (ISA) distributions and examine their dynamic changes. The impacts of elevation, slope, and rivers on spatial patterns of urban expansion were examined. This research indicates that ISA is an important variable for examining urban expansion. Coastal metropoles had much faster urbanization rates than inland metropoles. The spatial patterns of urban ISA distribution and expansion are greatly influenced by physical conditions; that is, ISA is mainly distributed in the areas with slopes of less than 10 degrees. Rivers are important geographical factors constraining urban expansion, especially in developing stages, while bridges across the rivers promote urban expansion patterns and rates. The relationships of spatial patterns of urban ISA distribution and dynamics with physical conditions provide scientific data for urban planning, management, and sustainability.

Tracking spatial–temporal landscape changes of impervious surface areas, bare lands, and inundation areas in China during 2001–2017

“…Postclassification comparison is one primary approach to derive this type of from‐to‐change information (Kuang et al, ; Liu et al, ; Ning et al, ; Pouliot et al, ). Great progresses have been made in the landscape mapping fields in recent two decades, particularly these significant contributions in monitoring and understanding the rapid urban expansion process in China (Guo, Lu, & Kuang, ; Kuang et al, ; Kuang et al, ; Liu et al, ; Liu et al, ; Lu et al, ; Ning et al, ; Pouliot et al, ). This study borrowed their ideas and learned from them in order to spatiotemporally tracking multiple landscape changes (Guo et al, ; Kuang et al, ; Kuang et al, ; Liu et al, ; Liu et al, ; Lu et al, ; Lu, Li, Kuang, & Moran, ; Ning et al, ; Pouliot et al, ).…”

Multitemporal Land Use and Land Cover Classification from Time-Series Landsat Datasets Using Harmonic Analysis with a Minimum Spectral Distance Algorithm

“…Gounaridis et al [18] applied an RF classifier with time-series Landsat datasets (1991-2016) to classify LULC maps in Attica, Greece and attained an overall accuracy varying from 90.5% to 93.5%. Lu et al [15] applied Landsat images from 1990 to 2015 at five-year intervals to classify impervious surface and non-impervious surface areas using a linear spectral mixture analysis of the six selected metropoles in the coastal and inland metropoles in 2015, achieving an overall accuracy varying from 94% to 95%. Mi et al [14] applied an RF classifier with time-series Landsat datasets (1987-2017) to detect the LULC changes in a mining area, achieving an average OA of about 84%.…”

“…Accelerated urban growth and LULC changes exert pressure on the natural environment and human welfare and have become a global concern [8]. Several studies on LULC changes and their impacts have been conducted worldwide from multiple dimensions using satellite remote sensing and GIS technology [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23]. All of these studies require time-series datasets that are mostly derived from Earth observation satellites to classify multitemporal LULC maps.…”

Methods for calculation the spatial distribution of the territory membership to the urbanized based on the hybrid neural network

“…Object-oriented classification (OOC) is used to extract urban land use patterns from high spatial resolution images through the physical features of ground objects such as spectral, shape and texture features (Deng, Zhu, He, & Tang, 2019;Lu et al, 2018). However, without considering spatial relationships among ground objects, OOC methods can only recognize land cover information with low-level semantic features.…”