Evaluation of Luojia 1-01 nighttime light imagery for impervious surface detection: A comparison with NPP-VIIRS nighttime light data

Highlights

• This study is the first to use Luojia 1-01 data for detecting impervious surfaces.

• Luojia 1-01 data can produce more accurate impervious surface areas than NPP-VIIRS data.

• Luojia 1-01 data fails to provide reliable estimates of the imperviousness degree.

• Nightlight images with finer resolution do not always improve the estimation reliability.

Abstract

Impervious surface detection is significant to urban dynamic monitoring and environment management. One of the most effective approaches to evaluating the impervious surface is the use of nighttime light imagery. However, little work on this subject was carried out with the new generation nighttime light data from Luojia 1-01 satellite, which has a finer spatial resolution than the predecessors such as the nightlight data from the Visible Infrared Imaging Radiometer Suite (VIIRS) sensor on the Suomi National Polar-orbiting Partnership (NPP) Satellite. Therefore, this study conducted the first investigation of the capacity of Luojia 1-01 nighttime light data in detecting the extent and degree of impervious surfaces. Focusing on three cities of Beijing, Shanghai, and Guangzhou, several maps of the spatial extent of impervious surface areas were first extracted from two types of nighttime lights data (Luojia 1-01 and NPP-VIIRS data) by applying a dynamic threshold segmentation method. Meanwhile, a series of polynomial regression models were adopted to estimate the relation between imperiousness degree and light intensity. The results compared with the reference data derived from Landsat 8 Operational Land Imager (OLI) show that Luojia 1-01 data can produce a more precise map of the spatial extent of impervious surfaces than NPP-VIIRS data owing to the finer spatial resolution and the wider measurement range. Nevertheless, Luojia 1-01 data failed to provide reliable estimates of the imperviousness degree in comparison with NPP-VIIRS data as this nighttime light imagery with finer spatial resolution can better discriminate the surfaces that have the same imperviousness degree but are illuminated with different light intensities, consequently resulting in a weak correlation between imperviousness degree and light intensity. The over- and under-estimates of imperviousness degree suggested an increase in spatial resolution of nightlight imagery does not always improve the accuracy and reliability of nighttime light-based estimations. These study results confirmed that Luojia 1-01 nightlight imagery is a potential and promising data source for mapping the spatial extent of impervious surface areas, but difficult to accurately estimate the imperviousness degree. Future research may improve the accuracy of imperviousness degree estimation by integrating the Luojia 1-01 nightlight imagery with other useful data sources.

Introduction

Impervious surfaces are generally defined as artificial structures that impede the infiltration of water into the underlying soil, e.g. roads, rooftops, and parking lots (Weng, 2012; Wu, 2009; Yang et al., 2003). They are of great importance to human beings, not only being a significant indicator for the level of urbanization, but also playing a key role in the change of urban environment (Yang et al., 2012; Deng et al., 2012). With their construction, the impervious surfaces can affect hydrological systems through sealing the soil surface, avoiding rainwater infiltration and natural groundwater recharge (Brabec et al., 2002; Jacobson, 2011; Lu and Weng, 2006). Besides, the transformation of natural surfaces into impervious areas has an impact on the land surface energy balance, inducing an increase in the air temperature (e.g. urban heat islands) (Jr and Gibbons, 1996; Wang et al., 2016; Weng and Lu, 2008; Wilson et al., 2003). Given the close concern with human activity, the detection of impervious surfaces (including the extent and degree) is vital for monitoring urbanization dynamics as well as analyzing impacts on urban environment.

Currently, remote sensing technology is one of the most effective approaches to detecting the impervious surfaces since it can provide accurate information on the surfaces spatially and temporally (Lu et al., 2011; Parece and Campbell, 2013; Zhang et al., 2012). Many studies have been carried out to obtain the impervious surfaces at various spatial and temporal scales based on remote sensing satellite images. For example, Zhou and Wang (2008) used a high-resolution imagery (QuickBird-2) for the extraction of impervious surface areas. Sexton et al. (2013) focused on time series of Landsat images for retrieving long-term records of impervious surface cover based on an empirical method. Deng and Wu (2013) also developed a compositive approach of machine learning techniques and spectral mixture analysis to obtain the impervious surfaces using the single-date Moderate Resolution Imaging Spectroradiometer (MODIS) image. In another study using multispectral optical data and dual polarization synthetic aperture radar (SAR) data, Zhang et al. (2016) presented a comparative study to identify urban impervious surfaces in a study site. Moreover, based on the imagery collected with various satellites, several well-known datasets, e.g., the High Resolution Layer—Imperviousness (HRL-I) system produced for the European Union (Kuntz et al., 2014) and the Global Man-made Impervious Surface (GMIS) Dataset developed by NASA Socioeconomic Data and Applications Center (Brown De Colstoun et al., 2017) have been built and provided available estimates of impervious surfaces at large scale.

In addition to the daytime imagery, satellite-observations of nighttime lights have also been applied to impervious surface estimation (Liu et al., 2015; Zhuo et al., 2018). The nighttime light imagery, mainly derived from the Defense Meteorological Satellite Program’s Operational Linescan System (DMSP-OLS), can well detect the anthropogenic lights at night (Imhoff et al., 2012; Small and Elvidge, 2013). Since the nighttime light illumination mostly originates from artificial sources that are closely related to human activities, the DMSP-OLS nightlight imagery can be potentially useful for the measurements of impervious surfaces based on the location and relative intensity of light sources (Pok et al., 2017). For example, using DMSP-OLS nightlight imagery, previous studies by Imhoff et al. (1997); Elvidge et al. (1999); Small et al. (2005), and Ma et al. (2012) approximately mapped the spatial extents of urban areas in different ways. Elvidge et al. (2007) and Sutton et al. (2009) further found that there was a positive relationship between imperviousness degree and light intensity, and proved that nighttime light data was appropriate in detecting impervious surfaces. In spite of that, DMSP-OLS nightlight imagery has some well-known shortcomings (Ou et al., 2015), e.g., coarse spatial resolution (about 1 km), blooming effect (spatial overextension of lighted areas), and saturation in urban cores, which always result in the overestimates for the spatial extents of impervious surface areas (Letu et al., 2012).

As a successor to the DMSP-OLS sensor, the Visible Infrared Imaging Radiometer Suite (VIIRS) on the Suomi National Polar-orbiting Partnership (NPP) satellite has offered a series of high-quality imagery of Day/Night Band nighttime lights since 2012. The NPP-VIIRS nightlight imagery has significant improvements, including the wider measurement range, higher spatial resolution, and on-board calibration, which partially eliminate the limitations existing in DMSP-OLS data (Shi et al., 2014a; Zhang et al., 2017). For example, based on a study of 12 cities in China, Shi et al. (2014b) found that the urban areas derived from NPP-VIIRS nighttime light data displayed higher accuracies compared with those from DMSP-OLS data. In another study, Yu et al. (2018) proved that NPP-VIIRS data have better capability in urban built-up area mapping by using a logarithmic transformation method. Besides, NPP-VIIRS nightlight imagery was found to be integrated with other data such as MODIS normalized difference vegetation index (NDVI) for mapping the distributions of impervious surface area accurately (Guo et al., 2015). The literature review shows that NPP-VIIRS nightlight imagery is widely applied in impervious surface estimations, but mostly limited to be at a moderate spatial resolution (about 750 m).

Now, Luojia 1-01, a new generation of nighttime light remote sensing satellite developed by Wuhan University in China, was successfully launched on 2 June 2018. The nighttime light imagery generated from Luojia 1-01 satellite supplement the existing nightlight data with the image features in regard to fine spatial resolution (about 130 m) and high radiometric quantization (14 bits). Compared to the NPP-VIIRS data, the spatial resolution of Luojia 1-01 data has greatly improved with on-board calibration, which can show more spatial details of light sources (Zhang et al., 2019). Besides, the Luojia 1-01 data does not suffer the problems of saturation and blooming which exists in DMSP-OLS data (Li et al., 2019). The advantages of this new data can significantly enhance the detection capacity of artificial lightings, thus bringing new insights and possibilities to the researches on urban and environment. Currently, several studies have employed Luojia 1-01 data to estimate the artificial light pollution and urban extent mapping, and demonstrated that Luojia 1-01 nightlight imagery probably provide higher capacity in comparison with NPP-VIIRS nighttime light data. For example, through assessing the sources and patterns of artificial light pollution with nighttime light data, Jiang et al. (2018) confirmed that Luojia 1-01 data can be usefully applied for investigating urban light pollution. In another study using the Luojia 1-01 data, Li et al. (2018) compared several methods for mapping urban areas, and also found that Luojia 1-01 data can result in better extraction results than NPP-VIIRS data. Unfortunately, they only focused on the extraction of spatial extent and ignored to estimate the imperviousness degree in urban areas. To the best of our knowledge, there is still no work that has examined the potential of Luojia 1-01 nightlight data for detecting impervious surfaces, especially the imperviousness degree. To better understand the quality of Luojia 1-01 nighttime light data as well as support further analysis in related studies of urban dynamics and environment, a comprehensive investigation of this new data is essential for impervious surface detection.

Thus, this study aims to assess the capability of Luojia 1-01 data for detecting the degree and extent of impervious surfaces in three cities of China, such as Beijing, Shanghai, and Guangzhou. For comparison, the NPP-VIIRS data is also used to examine the difference between two kinds of nighttime light data. Based on a reference data derived from Landsat 8 Operational Land Imager (OLI), the accuracy assessment is finally conducted to quantitatively measure the reliability of nighttime light data in impervious surface detection. This study is the first time that Luojia 1-01 nighttime light data is applied for an investigation of impervious surfaces, which will not only fill the gaps in the field of nighttime light research, but also provide useful support for government decision-making to plan the urban development and environmental management.