A new source for high spatial resolution night time images — The EROS-B commercial satellite

Highlights

• We used the EROS-B sensor for acquiring high spatial resolution night-time images.

• We compared EROS-B images of Brisbane (Australia) with aerial night-time images.

• Land use and land cover data explained 89% of the variability in night-time lights.

• Arterial roads and commercial and services areas were the brightest areas.

Abstract

City lights present one of humankind's most unique footprints on Earth as seen from space. Resulting light pollution from artificial lights obscures the night sky for astronomy and has negative impacts on biodiversity as well as on human health. However, remote sensing studies of night lights to date have been mostly limited to coarse spatial resolution sensors such as the DMSP-OLS. Here we present a new source for high spatial resolution mapping of night lights from space, derived from a commercial satellite. We tasked the Israeli EROS-B satellite to acquire two night-time light images (at a spatial resolution of 1 m) of Brisbane, Australia, and analyzed their radiometric quality and content with respect to land cover and land use. The spatial distribution of night lights as imaged by EROS-B corresponded with night-time images acquired by an airborne camera, although EROS-B was not as sensitive to low light levels. Using land cover and land use data at the statistical local area level, we could statistically explain 89% of the variability in night-time lights. Arterial roads and commercial and service areas were found to be some of the brightest land use types. Overall, we found that EROS-B imagery provides fine spatial resolution images of night lights, opening new avenues for studying light pollution in cities worldwide.

Introduction

One of humanity's unique global footprints is that of artificial night lights. Artificial lighting has led to two types of light pollution (Longcore & Rich, 2004): “astronomical light pollution”, obscuring the view of the night sky due to atmospheric scattering, and “ecological light pollution”, the alteration of natural light regimes in urban and non-urban ecosystems affecting the flora, fauna and human health (Longcore & Rich, 2004; Navara & Nelson, 2007). To fully understand the mechanisms of artificial light emission and the resulting light pollution, we need to know the spectral power distribution of different lamps used, the spatial distribution of lighting types, and to be able to model and predict the effects of artificial lighting on sky brightness (luminance) observed from a site at a distance from the city center (Elvidge et al., 2010, Garstang, 1986). Understanding the relations between outdoor lighting characteristics to sky glow and human health is an active area of research (Aubé, Roby, & Kocifaj, 2013), however night-time imagery providing data to such models is limited by available sensors, which are most often of coarse spatial resolution (Duriscoe, Luginbuhl, & Elvidge, 2014).

Night-time light images have been successfully used at global and regional (1000's km²) scales, for various applications, including demographic, political and conservation oriented studies, as well as for modeling light pollution. However, the widely used DMSP (Defense Military Satellite Program) night-time imagery is limited in its application to local and within city scales (10's km²) due to its coarse spatial resolution (3 km pixels), overglow (the “spilling” of light from built-up areas into non-lit areas), saturation in urban areas and intra-sensor calibration problems (Doll, 2008). At present the only other operational night-time light sensitive sensors on-board satellites include the Visible Infrared Imaging Radiometer Suite (VIIRS) on-board NASA'S Suomi NPP satellite launched in 2011, offering night light imagery at a spatial resolution of about 740 m; (Elvidge et al., 2013, Miller et al., 2012) and the SAC-C and SAC-D sensors (SAC — Satélite de Aplicaciones Científicas) launched in 2000 and in 2009, respectively (Colomb, Alonso, Hofmann, & Nollmann, 2004). Although offering night-time light images at a pixel size of about 300 m, the SAC-C and SAC-D sensors have hardly been used for local scale city applications (but Levin and Duke, 2012, Mazor et al., 2013). While at the global scale, night lights from DMSP were found as a useful surrogate for national gross domestic product as well as for mapping poverty (Doll et al., 2000, Elvidge et al., 2009), such analyses are yet to be undertaken at the municipality level. Finer spatial resolution night-time imagery is therefore needed to better understand local-scale urban dynamics and to examine within-city socio-economic differences at the neighborhood and street level, at spatial scales enabling the identification of individual buildings and lighting infrastructure. Astronaut color night-time photographs from the International Space Station at a pixel size of about 50 m (Doll, 2008, Levin and Duke, 2012), freely available since 2003, cover only selected areas globally, and are not collected in a regular pattern. Recently there have been several night-time aerial campaigns collecting high spatial resolution images of selected cities, e.g., in Berlin (Kuechly et al., 2012), and Southampton (Hale et al., 2013). Other airborne campaigns have used hyperspectral sensors for acquiring night-time images, e.g., AVIRIS (Airborne Visible/Infrared Imaging Spectrometer) over Las Vegas (Elvidge & Jansen, 1999), or CASI (Compact Airborne Spectographic Imager) over Barcelona (Tardà et al., 2011). However such airborne campaigns can be relatively more expensive to carry out compared with satellite imagery acquisition (Mumby, Green, Edwards, & Clark, 1999) and cannot be easily tasked to remote areas or to cover very large areas (> 10⁶ km²). The potential use of information from high spatial resolution night-time light sensors has been summarized in a proposal submitted to NASA, to launch a night-time light dedicated sensor to be termed NightSat (Elvidge et al., 2007, Hipskind et al., 2011).

In this paper we present a new source for high spatial resolution night-time imagery — derived from the commercial satellite of EROS-B (Poli & Toutin, 2012). Specifically, we tasked two high spatial resolution images of Brisbane (Australia) and acquired aerial night-time color photos to explore the information that EROS-B can provide in an urban setting. For many studies the acquisition of one nighttime light image is sufficient. In our case, having two images acquired with a spatial overlap between them, allowed us to examine the correspondence between night lights at two different dates, something which has not been done before, as multi-temporal assessment of night lights so far has been mostly using the annual global mosaics of stable lights acquired by the DMSP-OLS (e.g., Zhang & Seto, 2011). We also aimed to compare the spatial pattern of night lights with other indicators of human activity and of urban land cover, including land use, roads and vegetation cover, as well as to explore its possible uses for urban ecology issues (Dearborn and Kark, 2010, Sushinsky et al., 2013).

More specifically, we asked the following questions:

• What is the radiometric quality of high spatial resolution night-time light imagery collected by the EROS-B satellite, and how well does it correspond with aerial night-time imagery?

• Which land cover and land use classes best explain the spatial variability in night-time lights?

• What are the similarities and differences between night-time light brightness, and other proxies for human activity within a city (e.g., building type, city attractions)?