The terminology of “urban shrinkage” originates from “Schrumpfende Städte”, referring to the phenomenon of population decline and economic recession caused by deindustrialization in Germany [ 1 ]. It is not a new phenomenon [ 2 ] and has been spreading worldwide [ 3 8 ]. Since the mid 20th century, with the adjustment of global economic structure and the ageing of the population, more than 25% of the world’s major cities have been considered to be shrinking cities [ 9 ]. Urban shrinkage has become a topic of major concern to scholars of geography and urban science. Due to the great heterogeneity of political, economic, geographic, and social backgrounds across different cities, the causes and spatial characteristics of urban shrinkage have varied by countries in different national contexts [ 10 ]. In many Western countries, urban shrinkage is mostly caused by deindustrialization [ 11 ] and suburbanization [ 10 ], often resulting in massive population loss and severe economic recessions [ 12 13 ]. In terms of space, such shrinkage is characterized by a comprehensive recession in the central area of the city, the emergence of a large number of vacant residential and office buildings, and the oversupply of urban service facilities [ 14 ]. Examples of cities where this has taken place include Detroit, Cleveland, and Youngstown in the United States, Leipzig in Germany, Manchester in the United Kingdom, and many mining cities in Australia [ 15 ]. As far as commonality is concerned, many of China’s resource-based cities are also facing a chain reaction process, such as the depletion of resources, industrial recession, reduction of jobs, population loss, and economic recession. In terms of specific characteristics, the shrinkage caused by de-industrialization, and by the growth brought about by urbanization and re-industrialization, overlap in China. Many scholars hold the view that China’s shrinking cities are mainly characterized by the scattered and fragmented shrinkage of partial areas [ 16 17 ], rather than an overall recession. As such, it is necessary to refine the current understanding of the internal shrinkage pattern of China’s shrinking cities.
At present, while the research on urban shrinkage both in China and internationally is on the rise, the definition of the shrinking city has been inconsistent [ 18 ], leading to a dilemma of identification. Generally, scholars believe that population decline is the main sign of urban shrinkage [ 19 24 ], thus the identification of shrinking cities mainly relies on population data [ 20 24 ]. However, traditional population data has the limitations of collection difficulties, inaccurate data statistics, statistical standard changes, long re-update cycles, and rough spatial expressions [ 25 26 ], among other limitations, thus leading to erroneous conclusions. Faced with the measurement dilemma, some scholars began to use nighttime light (NTL) composite data obtained by satellites to identify shrinking cities [ 27 30 ]. In urban space science research, the NTL data obtained by the OLS (Operational Linescan System) sensor under the Defense Meteorological Satellite Program (DMSP) is most commonly used. These data provide unified, spatially explicit, continuous, and timely annual, cloud-free, composited and stable light data. Therefore, they have been widely used in evaluating the process of urbanization [ 28 ], economic development dynamics [ 31 33 ], population distribution [ 34 35 ], power consumption [ 36 ], drawing poverty maps [ 37 ], and mapping out urban areas [ 38 40 ]. However, DMSP-OLS NTL data still have some defects, such as oversaturation, low spatial resolution (about 1 km at the equator), a lack of on-board correction and satellite correction, and a blooming effect on surrounding areas [ 41 44 ]. Moreover, the National Geophysical Data Center (NGDC) of the USA unfortunately stopped producing monthly composites of DMSP/OLS after February 2014. Its successor, the Visible Infrared Imaging Radiometer Suite (VIIRS) on board the Suomi National Polar Orbiting Partnership (NPP), launched in October 2011, and these data have been systematically updated for the period starting in April 2012. The NPP-VIIRS NTL data make up for the defect of the DMSP-OLS data, as they have a higher spatial resolution (375 m and 750 m at nadir), with a spectral range of 500–900 nm that is sensitive to very low visible light. In addition, an onboard calibration system is utilized to improve the quality of the NPP-VIIRS NTL data. Hillger et al. argue that the NTL data obtained from VIIRS provide more abundant information about human settlement and economic activities than DMSP NTL data [ 45 ]. A former study has, furthermore, demonstrated its superiority over DMSP-OLS with regard to urban characterization [ 45 46 ].
To date, urban studies, using NPP-VIIRS NTL data, have focused on the following areas: (1) the relationships between the NPP-VIIRS NTL data and socioeconomic indicators, such as population, GDP, and house vacancy rate [ 25 48 ]; (2) extracting built-up areas and dynamically monitoring urban expansion by using the NPP-VIIRS NTL data [ 36 50 ]; (3) exploring the spatial distribution of the population using NTL data [ 46 51 ]. These studies have provided insights and foundations for using NPP-VIIRS NTL data to identify urban shrinkage and growth patterns. Firstly, many studies have shown that NTL radiance values have a significant statistical relationship with gross domestic product (GDP) and population, which illustrates the comprehensiveness of NTL data and its strong relevance to human activities. In addition, the capacity of NTL time-series data to dynamically detect changes in urban landscapes demonstrates the objectivity and timeliness of these data. Finally, using NPP-VIIRS NTL data to spatialize socio-economic indicators can be seen as being more reliable than using traditional data. In order to test the superiority of NPP-VIIRS NTL data, some scholars have made meaningful attempts to contribute to research on city shrinkage. At present, there are two methods of processing the NTL data for urban shrinkage research. Firstly, Du et al. (2017) and Liu et al. (2018) employed a calculation of the difference of different years’ NTL radiance value in every grid to identify urban shrinkage and growth; however, this difference method cannot determine the continuity and trend of urban shrinkage and growth. Secondly, Li et al. (2019) used the NPP/VIIRS NTL data in the calculation of the change slope of each grid’s NTL radiance value to identify the urban shrinkage and growth pattern; however, the data here were not carefully processed through the implementation of noise removal. Moreover, the scope of the research focused on the whole prefecture-level city rather than the urban area, which differs from the western cities where there are urbanized areas within administrative boundaries.
Based on the above analysis, this paper takes Yichun, a typical resource-based city in China, as an example, using the NPP-VIIRS nighttime light (NTL) data of 56 months from 2012 to 2019 to identify the shrinkage and growth pattern. The research scope focuses on the built-up area in the urban area of Yichun, for which the area was extracted by the visual interpretation of a high-resolution Google satellite image. The key aim of the research was to calculate the slope of the change of the NTL radiance value of each grid in the built-up area in order to then identify the growth and shrinkage of the city. The main objectives of this paper are as follows: to implement an objective and effective method to understand the shrinkage and growth pattern of Yichun; to summarize the shrinking characteristics and types of Yichun City; and to explore the advantages of using NPP-VIIRS NTL data for urban shrinkage identification. Based on the body of research of typical cases in China, this paper hopes to provide a new perspective for objectively and meticulously identifying shrinkage and growth patterns within a single city, thus providing a basis for the governance of resource-based cities.