Satellite Observation of Urban Metabolism

EarthzineArticles, Earth Observation, Original, Sections, Themed Articles, Urban Monitoring Theme

Map showing the location of Peddavagu basin, a tributary of Krishna River basin

Christopher D. Elvidge, NOAA National Geophysical Data Center (chris.elvidge@noaa.gov)

Paul C. Sutton, University of Denver

Sharolyn Anderson, CIRES University of Colorado

Kimberly E. Baugh, CIRES University of Colorado

Daniel Ziskin, CIRES University of Colorado

The term ‰ÛÏUrban Metabolism‰Û is a conceptual framework for analyzing the flow of energy and materials within cities. Efforts to assess the sustainability of cities via this conceptual framework have been proposed by many citizens, scholars, and philosophers including Abel Wolman (a sanitary engineer), Sir Patrick Geddes (a biologist), and Karl Marx (a political philosopher). Urban areas constitute roughly only 2 percent of the earth’s land surface yet house over 50 percent of the world’s population and an even greater percentage of the world’s economic activity. Understanding and controlling what happens in this ~2 percent of the world’s land surface is essential to any attempts at guiding humanity toward sustainable living.

Satellite remote sensing of nighttime lights provides the most straightforward approach to mapping and monitoring the human enterprise in the 21st century (Figure 1). There is a concept in urban design that ‰ÛÏform follows function.‰Û Outdoor lighting is used to illuminate the forms differentially, based on function, importance, and activity levels. A growing body of research indicates that satellite observation of lighting provides such a useful proxy for variables that would be difficult to measure and map by other means (Elvidge et al., 1997; Sutton et al., 2009; Elvidge et al., 2010a; Elvidge et al., 2010b). However, to date there has not been a satellite mission flown with an instrument optimized for observing global nighttime lights.

Figure 1.  Color digital camera image of Paris acquired from the International Space Station.  The spatial resolution is approximately 20 meters.The global human society can be viewed as a collection of interconnected organisms, each carrying on a standard set of processes collectively referred to as ‰ÛÏurban metabolism‰Û (Burgess, 2008). Global, national, and regional economies are often modeled with a circular flow diagram of households and firms exchanging capital, labor, goods, and services in a circulatory system. However, the human economy is oversimplified when modeled as a circulatory system without a digestive system (Daly and Farber, 2011). As with an organism, there is a digestive system which ingests and consumes resources (food, water, energy, materials) and releases wastes (gases such as CO2, solid and liquid wastes). The digestive system enables the economy, growth in development, and population expansion. The processes are standardized based on society’s drive to satisfy the basic requirements for human life plus the enhancements afforded by a common technological base. Through advances in technology, we have been so successful with this metabolism that our numbers have grown from ~300 million in the 1300s to 7 billion in 2011. In 2000, Crutzen and Stoermer (2000) coined the term ‰Û÷Anthropocene’ in recognition that humans had become a driving force in altering the global environment.

There are satellite-based systems for monitoring the metabolism of terrestrial ecosystems and near-surface aquatic ecosystems. The near-infrared versus red ‰ÛÏvegetation index‰Û in all its forms has served as a ubiquitous measurement in satellite remote sensing of terrestrial ecosystems since the mid-1970s (Tucker, 1979). Satellite instruments with red and NIR spectral bands are used for tracking spatial and temporal variations in the metabolism of terrestrial ecosystems. These instruments have flown continuously since the 1970s (Landsat, Advanced Very High Resolution Radiometer, MODIS, and others). Similarly, visible band indices have been developed to monitor the metabolism of surface waters via plant pigment concentrations, beginning with the Coastal Zone Color Scanner, SeaWIFS, OCTS and others (O’Reilly et al., 1988).

The success of these systems derives from the fact that they were designed around a unique spectral observable that integrates the spatial distribution and activity levels of photosynthetic organisms, the base of the food chain. Is there a corresponding observable for the human enterprise which could be observed from space? If one could be identified, sensors could be developed and missions could be flown to collect data for modeling ‰ÛÏurban metabolism.‰Û In this paper, we review evidence that satellite-observed nighttime lights can be used to map and monitor urban metabolism and review the observational requirements that should be considered in the design of a satellite instrument optimized for measuring nighttime lights.

‰ÛÄ

Background:

For nearly 40 years, the U.S. Air Force Defense Meteorological Satellite Program (DMSP) has flown polar orbiting satellites with a visible and thermal cloud imaging sensor known as the Operational Linescan System (OLS). The digital OLS archive begins in 1992 and extends to the present. OLS is unique for its low light imaging capability. Originally designed for the detection of moonlit clouds with a 3000 km swath and a 2.7 km ground sample distance, the OLS detects lights present at the Earth’s surface, including lights from cities and towns, rural development, gas flares, fires and heavily lit fishing boats. NGDC has developed algorithms for processing annual cloud and fire free composites of nighttime lights (Baugh et al., 2010).

Figure 2.  Aggregate relationship found between the DMSP sum-of-lights index with population, gross domestic product, total primary energy consumption, and fossil fuel CO2 emissions for China from 1992-2009..  Years generally increase from lower left (1992) to upper right (2009).

Figure 2. Aggregate relationship found between the DMSP sum-of-lights index with population, gross domestic product, total primary energy consumption, and fossil fuel CO2 emissions for China from 1992-2009. Years generally increase from lower left (1992) to upper right (2009).

For evidence that satellite-observed nighttime lights, even at coarse spatial resolution, can serve as a proxy for various aspects of urban metabolism, we present results from the version 4 time series of DMSP nighttime lights, spanning 1992- 2009. For each satellite year, we extracted the sum of lights for the countries of the world. The sum of lights, or lights for short, is the sum of the digital numbers associated with each pixel within a country’s boundary. This data was then paired with national level annual data on population, gross domestic product (GDP), total primary energy consumption, and fossil fuel CO2 emissions.

Figure 2 shows the relationships found for China. There is a logarithmic relationship between lights, and population, a linear relationship with GDP, an exponential relationship with total primary energy consumption and a linear relationship with CO2 emissions. There is no other satellite remote sensing system with such a clearly demonstrated ability to correlate with these primary variables of urban metabolism. Certainly, the DMSP is not making a direct observation of variables such as those shown in Figure 2. The strong correlation that exists because nocturnal lighting is one expression of urban metabolism and there is a strong tie between all elements of urban metabolism due to the standardization of technologies used in urban areas around the world. Nighttime lights can serve as a proxy for the spatial distribution and intensity of variables such as economic activity, ecological footprints, and energy consumption, which are much more difficult to map and measure.

Elvidge et al., 2007 identified the following shortcomings for the observation of nighttime lights by the OLS sensor: (a) coarse spatial resolution; (b) lack of on-board calibration; (c) lack of systematic recording of in-flight gain changes; (d) limited dynamic range; (e) six-bit quantification; (f) signal saturation in urban centers resulting from standard operation at the high gain setting; (g) lack of a spectral band suitable for fire detection; (h) limited data recording and download capabilities (most OLS data are averaged on-board to enable download of global coverage); and (i) lack of multiple spectral bands capable of discriminating lighting types.

Figure 3. DMSP-OLS annual cloud-free composite of nighttime lights of the Los Angeles, California, region.  The image contrast has been enhanced to reveal dim lighting detected by the OLS sensor.  Note the overglow surrounding Los Angeles, extending more than 50 km offshore. The overglow arises from the OLS detection of dim lighting scattered in the atmosphere. Overglow digital number (DN) values exceed DN values for lighting from Interstate 15 highway traffic between LA and Las Vegas and many small towns.

Figure 3. DMSP-OLS annual cloud-free composite of nighttime lights of the Los Angeles, California, region. The image contrast has been enhanced to reveal dim lighting detected by the OLS sensor. Note the overglow surrounding Los Angeles, extending more than 50 km offshore. The overglow arises from the OLS detection of dim lighting scattered in the atmosphere. Overglow digital number (DN) values exceed DN values for lighting from Interstate 15 highway traffic between LA and Las Vegas and many small towns.

While these results appear impressive, there are substantial shortcomings to DMSP when it comes to the observation of nighttime lights (Figure 3). The coarse spatial resolution [(2.7 km ground sample distance (GSD) and 5 km ground instantaneous field of view (GIFOV)] make it impossible to detect the internal structure of urban centers related to the spatial distribution of functional sectors such as residential, commercial, and industrial. The data are recorded with 6-bit resolution. There is no on-board calibration. The low-light imaging gain is programmed to track predicted solar and lunar illuminance, but the gain settings are not recorded in the data stream. The system is typically operated at high-gain settings, resulting in saturation in urban centers.

The follow-on for the OLS is the visible/infrared imager/radiometer suite (VIIRS), which will fly on the U.S.’s next generation polar-orbiting operational environmental satellite system. The first VIIRS is currently being built and represents an improved, but still imperfect, instrument to measure nocturnal lighting (Lee et al., 2004). The NPOESS VIIRS instrument will provide low-light imaging data with improved spatial resolution (0.742 km), wider dynamic range, higher quantization, on-board calibration, and simultaneous observation with a broader suite of bands for improved cloud and fire discrimination over the OLS. The VIIRS is not, however, designed with the objective of sensing nighttime lights. Rather, it has the objective of nighttime visible band imaging of moonlit clouds—the same mission objective of the OLS lowlight imaging. While the VIIRS will acquire improved nighttime lighting data, it is not optimal for this application. In particular, the VIIRS low-light imaging spatial resolution will be too coarse to permit the observation of key nighttime lighting features within human settlements and the low-light imaging is in a single spectral band, offering no ability to distinguish between lighting types.‰ÛÄ

The Nightsat Mission Concept

The DMSP low-light imaging was developed for the detection of moonlit clouds. If a sensor was developed specifically for observation of nighttime lights, what would the system capabilities be? The sensor would be in a polar orbit, making it possible to acquire data over the entire land surface. The overpass time would be at about 21:30 local time. Too early and there are problems with solar contamination. Too late and some lights will be turned off for energy conservation. The orbit should have a repeat cycle similar to Landsat (16 days), but avoid synchronization with the 29-day lunar cycle. The spatial resolution should be sufficient to observe the structural features found within urban centers. The detection limits should be low enough to detect dim lighting present in development at suburban edges and in rural areas. The saturation radiance would be about E-2 Watts/cm2/sr/um to minimize saturation on most brightly lit areas. Such a wide dynamic range calls for a high level of quantization, such as 14 bit.

Nocturnal images of region of Los Angeles, California, that include LAX International Airport and Marina Del Rey. The 50- and 742- meter resolution images were simulated from a photograph taken from the International Space Station.

Figure 4. Nocturnal images of region of Los Angeles, California, that include LAX International Airport and Marina Del Rey. The 50- and 742- meter resolution images were simulated from a photograph taken from the International Space Station.

To examine the spatial resolution requirements, we tested the ability of satellite-observed nighttime lights at a range of spatial resolutions for their ability to accurately map the locations of residential, commercial / industrial development, and vacant lands based on comparison to a detailed land use map available for the Los Angeles, California, region. The primary source data was a 20-meter resolution nighttime lights image of Los Angeles acquired in December 2010 from the ISS. This image was georeferenced into a UTM projection using image to map control points. Then the 20-meter UTM image was aggregated to 50, 60, 70, 80, 90, 100, 200, 300, 500, and 742-meter resolution by aggregation (averaging). At the coarse end, we included the 2010 radiance lights from DMSP.

Figure 4 has three images which characterize the range of the spatial resolutions tested. The nearest analog to this capability are the ‰ÛÏcities at night‰Û images acquired by astronauts from the International Space Station (Figure 1). While approximately 2,000 such images have been acquired, it is impossible to make a global collection of such images from the ISS due to the orbit. Also, we have found that the ISS camera detection limits are not low enough, leaving some urban and suburban residential areas undetected.

Conclusion:

Using a proxy measure means that when you cannot measure exactly what you want, you measure what you can. This is exactly the case for nighttime lights and urban metabolism. Many scholars and institutions have attempted to develop indices or metrics of sustainability at local, national, and regional scales (Ehrlich and Holdren, 1971; Sutton, 2003; Wackernagel and Rees, 1996). The ‰Û÷Ecological Footprint‘ (EF) developed by Mathis Wackernagel is a sophisticated and well-regarded index for measuring humanity’s demand on nature. The EF includes hundreds of nationally aggregated variables that vary in quality and availability from one nation to the next. The EF measures how much land and water area a human population requires to produce the resources it consumes and to absorb its carbon dioxide emissions. Indices of human impact in the form of ‰Û÷urban metabolism’ derived from nighttime satellite imagery show strong correlations with metrics such as the EF and they are measured objectively and uniformly across the globe (Sutton et al., 2011).

A Nightsat mission would answer questions akin to those being asked by the Global Footprint Network: ‰ÛÏHumanity needs what nature provides, but how do we know how much we’re using and how much we have to use?‰Û The suite of satellites in orbit now have focused on the last part of this question: ‰ÛHow much do we have to use?‰Û A Nightsat mission is an ideal vehicle for answering the question: ‰ÛHow much are we using?‰Û As resources get increasingly scarce and the world’s population lurches toward 10 billion, it will be increasingly hard to keep track of the material and energy flow (‰Û÷urban metabolism’) of the human enterprise that takes place throughout the world.

‰ÛÄ

References:

Baugh, K., Elvidge, C., Ghosh, T., and Ziskin, D. (2010) Development of a 2009 Stable Lights Product using DMSP-OLS data. Proceedings of the 30th Asia-Pacific Advanced Network Meeting, 114-130.

Burgess, E.W. (2008) The Growth of the City: An Introduction to a Research Project. Urban Ecology, Section II, 71-78.

Crutzen, P.J., and Stoermer, E.F. (2000) The ‘Anthropocene’. Global Change Newsletter, 41, 17-18.

Daly, H.E., and Farber, J. (2011) Ecological Economics: Principles and Applications, 2nd ed., Island Press: Washington DC. ISBN-13: 978-1-59726-681-9.

Ehrlich, P.R., and Holdren, J.P. (1971) Impact of population growth. Science, 171, 1212-1217.

Elvidge, C.D., Baugh, K.E., Kihn, E.A., Kroehl, H.W., and Davis, E.R. (1997) Mapping of city lights using DMSP Operational Linescan System data. Photogrammetric Engineering and Remote Sensing, 63, 727-734.

Elvidge, C.D., Imhoff, M.L., Baugh, K.E., Hobson, V.R., Nelson, I., Safran, J., Dietz, J.B., and Tuttle, B.T. (2001) Night-time lights of the world: 1994-1995. ISPRS Journal of Photogrammetry and Remote Sensing, 56(2), 81-99.

Elvidge, C.D., Cinzano, P., Pettit, D.R., Arvesen, J., Sutton, P., Small, C., Nemani, R., Longcore, T., Rich, C., Safran, J., Weeks, J., and Ebener, S. (2007) The Nightsat mission concept. International Journal of Remote Sensing, 28(12), 2645-2670.

Elvidge, C., Baugh, K.E., Sutton, P.C., Bhaduri, B., Tuttle, B.T., Ghosh, T., Ziskin, D., and Erwin, E.H. (2010a) Who’s in the Dark: Satellite Based Estimates of Electrification Rates. In X. Yang (ed.) Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment. Chichester, UK: Wiley-Blackwell, 30 pp.

Elvidge, C.D., Keith, D.M., Tuttle, B.T., and Baugh, K.E. (2010b) Spectral identification of lighting type and character. Sensors, 10(4), 3961-3988.

Lee, T.F., Miller, S.D., Turk, F.J., Schueler, C., Julian, R., Elvidge, C., Deyo, S., Dills, P., and Wang, S. (2004) The Day/Night Visible Sensor Aboard NPOESS VIIRS. 13th Conference on Satellite Meteorology and Oceanography, Norfolk, VA, USA: American Meteorological Society, 12pp.

O’Reilly, J.E., Maritorena, S., Mitchell, B.G., Siegel, D.A., Carder, K.L., Garver, S.A., Kahru, M., and McClain, C. (1988) Ocean Color Chlorophyll Algorithms for SEAWIFS. Marine Science Faculty Publications. Paper 6. http://scholarcommons.usf.edu/msc_facpub/6

Sutton, P.C., Anderson, S.J., Elvidge, C.D., Tuttle, B.T., and Ghosh, T. (2009) Paving the Planet: Impervious Surface as a Proxy Measure of the Human Ecological Footprint. Progress in Physical Geography, 33(4), 1-18.

Sutton, P.C. (2003) An empirical environmental sustainability index derived solely from nighttime satellite imagery and ecosystem service valuation. Population and Environment, 24(4), 293-311.

Sutton, P., Anderson, S., Tuttle, B., and Morse, L. (2011) The real wealth of nations: Mapping and monetizing the human ecological footprint. Ecological Indicators. ISSN 1470-160X, DOI: 10.1016/j.ecolind.2011.03.008.

Tucker, C.J. (1979) Red and Photographic Infrared Linear Combinations for Monitoring Vegetation, Remote Sensing of Environment, 8(2), 127-150.

Wackernagel, M., and Rees, W. (1996) Our Ecological Footprint: Reducing Human Impact on the Earth. New Society Publishers, Gabriola Island, BC Canada. 176 p. ISBN: 9780865713123.