Census from Heaven: An Estimate of Global Electricity Demand
Nadejda VictorAssociateBooz-Allen Hamilton Pittsburgh, PAvictor_nadejda@bah.com Christopher NicholsSenior Analyst, National Energy Technology Laboratory,Morgantown, WV
How much electricity does the world really need? Economic theory suggests that electricity demand is based on a number of factors such as per capita income, economic output, population, supply and cost of electricity. GDP is a crucial indicator in many socio-economic studies and an important reference for political decision making. However, GDP is imperfectly measured all over the world (Feige & Urban, 2008). Many developing countries have only rudimentary economic statistics and the poor data quality has obstructed attempts to estimate economic growth, poverty, health and environmental quality in these countries (Nordhaus and Chen, 2014). The lack of good sub-regional data has been even more discouraging for researchers working at the sub-national level.
The wide distribution of global population has made it difficult to collect and synthesize consistent data on the human socio-economic conditions at the high-resolution levels (the levels that are higher than national and sub-national units are). In addition, many developing countries have no reliable censuses of population. Thus, formal electricity demand models may not capture the appropriate variables for responses at the high geographical resolution, and, as a result, may be quite wrong about prospective electricity demand. Therefore one of the most important issues in social, economic and environmental research has been how to improve the quality of socioeconomic data in developing countries (Chen and Nordhaus, 2011; Nordhaus and Chen, 2014).
Our aim is to explore the advantages offered by spatial resolution global nighttime light images in order to estimate the luminosity gap between the OECD nations and the rest of the world and use it as a proxy for electricity consumption gap between developed countries and other nations. The distinct advantage of nighttime lights is that they are a unique data set related to human activities that is available for most of the globe at a very high resolution.
Luminosity in Social and Economic Studies
Nighttime light intensity is affected by multiple factors, such as population density, economic activity, infrastructure, etc. Luminosity data are employed as social and economic indicators to study regional and sub-regional socioeconomic systems. Nighttime light data were used to map the distribution of economic activity (Elvidge et al. 1997, Doll et al. 2000, Ebener et al. 2005; Ghosh et al.2010), poverty levels (Elvidge et al.2009; Wang,W. et al 2012), electrification rates (Elvidge et al. 2011; Doll, C., Pachauri, S., 2010, Kiran et al. 2009), resource consumption (Sutton et al. 2011; Elvidge et al. 1997, Letu et al. 2010), density of constructed surfaces (Elvidge et al. 2007), the copper and steel stocks (Takahashi et al. 2009, Hattori et al. 2013).
Strong positive correlations between total radiance and sub-national GDP were observed for all the countries examined at the finest level of observation with a very strong relationship (R2 value of 0.98) between the nighttime lights of these countries and their nominal GDP. Sutton et al. further confirmed these results (Sutton et al. 2007). Using the stable light product for the year 2000 of the U.S., India, China and Turkey, Sutton found a log-linear relationship between the night light and the nominal GDP (with an R2 value of 0.74). Similar studies were conducted for China (Zhao, et al., 2011), India (Bhandari & Roychowdhury, 2011) and Mexico (Ghosh, et al., 2010).
More recently, Chen & Nordhaus (2011), Kulkarni et al. (2011) and Nordhaus & Chen (2014) linked the time evolution of luminosity to economic activity and used all available nighttime lights annual data. Nordhaus and Chen concluded that light can be used as a proxy for nominal GDP, but this approach adds value to the official statistics for countries only with poor reporting standards and low data quality (Chen & Nordhaus, 2011, Nordhaus & Chen, 2014). Henderson et al. furthermore confirmed this statement for GDP growth (Henderson et al., 2009, Henderson et al., 2012).
Persistent nighttime light is a clear indicator of the presence of human settlements. An early effort to assess the population densities with luminosity data was done by Sutton (Sutton, 1997). Sutton compared data from the 1990 U.S. census with a binary image containing only the saturated pixels and found that these images could only explain 22-25% of the variation in the population density of the urban areas in the U.S. In a later study, Sutton, et al. estimated the global human population for the year 1997 as 6.3 billion people compared to 5.9 billion, which was the generally accepted estimate this year (Sutton, et al., 2001). Similar studies were done at the national levels, e.g. for China (Lo, 2001; Ma, T. et al., 2012) and Brazil (Amaral, et al., 2005). Thus, luminosity may not be a perfect proxy to measure population; however, in combination with other sources, it can substantially add value.
The present study examines luminosity, population and electricity consumption in 2010. We used only 2010 data because of lack of reliable electricity consumption and population data behind 2010 for non-OECD countries at the time of writing. We compared population density, luminosity at the 1° latitude × 1° longitude grid-cell resolution level, and electricity consumption at the country level. There are two primary data sources for this study: the Defence Meteorological Satellite Program (DMSP)’s nighttime lights and the gridded population data. The majority of studies to date use the coarse spatial resolution datasets from the DMSP and data on luminosity at night are collected by the DMSP-OLS satellite program, maintained, and processed by the National Oceanic and Atmospheric Administration (NOAA), (NOAA, 2014). Satellites orbit the Earth fourteen times a day with a nigttime overpass between 20:30 and 21:30 with sending images of every location spanning -180 to 180 degrees longitude and -65 to 75 degrees latitude at a resolution of 30 arc-seconds. The images are processed to remove cloud cover, snow and ephemeral lights to produce the final product publically available (see Figure 1). Figure 1 illustrates clear differences in the quantity of lighting around the world. Populations in the OECD countries generally have a surplus of lighting, yielding the white areas on Figure 1. Areas with high population count and modest lighting levels show up as grey (some parts of India and China). The black color on Figure 1 indicates areas where no lighting was detected by the DMSP sensor.
Figure 1. 2010 Nighttime Lights Composite (Source: NOAA,2014)
Yale University developed Geographically based Economic data set (GEcon) that is devoted to creating geophysically based data on economic activity at the global level (Nordhaus et al. 2006). The main effort of this research was to create data on gross cell product (GCP), but in addition to GCP data, the authors merged the economic data with other important demographic and geophysical data such as climate, physical attributes, location indicators and population.
To start with, we calibrated the NOAA’s luminosity data. Each pixel in the luminosity data set is assigned to a digital number (DN) representing its luminosity and the DNs are integers ranging from 0 to 63. These data can be converted to radiance by the equation: Radiance = (DN)3/2 * 10-10 Watts/cm2/sr/um (Chen and Nordhaus, 2010). We upscaled the NOAA’s luminosity data up to 1° x 1° grid cells to match GEcon data. We took each pixel’s DN and then summed these radiances over all pixels in the grid cell (each grid cell includes 120 x 120 pixels). After upscaling luminosity data we merged them with GEcon dataset. At the next step, we aggregated luminosity data by country using the GEcon’s RIG coefficient and combined these data with electricityr consumption data from the World Development Indicators (WDI) database (WB, 2014).
Under economic theory, GDP per capita represents economic welfare, but, in some cases, other indicators may better reflect the level of economic development: per capita electricity consumption, for instance, is considered as one of the most relevant alternatives (Joyeux and Ripple, 2007). The hypothesis on convergence in electricity per capita consumption could be an analog of the hypothesis on convergence in economics when the economies with low GDP per capita will tend to grow at faster per capita rates than richer economies. Maza and Villaverde find that a process of electricity consumption convergence has taken place and that the reduction of disparities is related to the rapid economic changes experienced by some developing countries and the energy conservation policies implemented by most developed countries (Maza and Villaverde, 2008).
Figure 2 shows electricity per capita consumption versus GDP per capita in 1960-2010 in logarithmic scale. Electricity consumption per capita varies significantly over time and across the countries, and, generally, countries with lower per capita income are associated with lower electricity per capita consumption. Figure 2 illustrates electricity use convergence: diversity of electricity consumption is higher with lower GDP per capita, diversity decreases with increases in income. How much electricity will non-OECD countries be likely to demand in the future under the growth paths of economic catch-up?
Figure 2. Electricity per capita versus GDP per capita by Countries, 1960-2010 (Source: WB, 2014)
Methodology and Results
We estimated future electricity demand in non-OECD countries in the future under the growth paths of economic catch-up using luminosity, population and electricity consumption data. First, we estimated luminosity gap between the OECD countries and the rest of the world using regression equation of the presence of satellite detected nighttime lighting and population density at the 1° latitude × 1° longitude grid-cell resolution level. Figure 3 is a scatter plot of log luminosity density and log population density (“log” always refers to natural logarithms) for all grid cells associated with OECD countries for 2010 (n = 2,559) (see equation ).
The regression equation of luminosity density as a function of population density for OECD countries shows a significant elasticity of 0.81 and R2 of 0.79. We used the coefficient estimates from the OECD regression equation to estimate prospective luminosity in non-OECD countries “if Everyone Lived Like in OECD” (see equation  and Figure 4). Subtractions of observed luminosity data from estimations for non-OECD countries provided the estimated luminosity gap between OECD and the rest of the world (see equation ).
In Figure 4 unlit areas are black, and lights appear with intensity increasing from dark-gray to white. Light intensity in the OECD areas is the DMSP-OLS 2010 data and lights in non-OECD areas reflect luminocity gap. In most regions, the higher concentration of lights in coastal areas mirrors the higher population densities. The comparison of lights in Europe and India reveals huge differences in population density.
Figure 3. Luminosity Density versus Population Density in OECD Countries, 2010.
Figure 4. Nighttime Lights Composite “if Everyone Lived Like in OECD”
We employed luminosity as a proxy for electricity consumption estimates as much light observable from space is from electric illumination. Lighting is normally the primary application of electricity in households (especially in developing countries), so electricity is primarily used for lighting and small appliances, rather than cooking, and represents an insignificant share of total household consumption in energy terms.
Luminosity data has an advantage over other proxies as night lights data are available over time and all space, but data on electricity consumption is unavailable for many lower income countries and is generally unavailable for most countries at sub-national levels. We used all available cells from merging the GEcon data and DMSP-OLS data and aggregated luminosities for all grid cells in a country to obtain that country’s total luminosity (see equation ). We estimate a regression of log electricity consumption as a function of log of luminosity by country and got a highly significant elasticity of 1.08, and a R2 of 0.87 . As is seen in Figure 5, countries such as Japan and China consume more electricity relative to their sum of nighttime lights value; on the other hand, Russia consumes less electricity relative to its sum of lights. This relationship provided a moderately strong R2 value of 0.87.
Figure 5. Electricity Consumption vs. Luminosity by Country, 2010
We applied the regression equation to estimate prospective electricity demand in non-OECD countries (equations -) and received estimations of electricity demand “if everyone lived like in OECD countries” at 1° x 1° grid cells resolution and by countries (equations ). Our approach allows to estimate electricity demand in the areas with low population density and without luminosity indication. Almost one-third of grid-cells with positive population and output were recorded with zero lights (Nordhaus and Chen, 2014) and while these grid-cells contain only a small fraction of economic output and population, it is a large part of the land area of the globe.
We evaluated how much electricity the world would likely demand if everyone lived like in OECD countries. The results show that prospective global electricity consumption is 2.4 times higher than observation in 2010. At the per capita level, the average prospective global electricity consumption is 7,224 kWh/cap or about the same as in Germany today -- a significant increase from the current level of about 2,500 kWh/cap.
Figure 6. Electricity Generation per capita in 2035 (IEA, 2013) and Prospective Electricity Demand per capita based on Luminosity Data
PDC – Prospective Demand per Capita
NPC – New Policies Scenario
450 –limiting concentration of greenhouse gases in the atmosphere to around 450 parts per million of CO2
CPS – Current Policies Scenario
According to the IEA the global electricity consumption in 2010 was 21,408 TWh and by 2035 it is projected to increase by 86% in the Current Policy Scenario (CPS), by 51% in the 450 Scenario (450), and by 73% in the New Policy Scenario (NPS) (IEA, 2013). Electricity per capita consumption growth is 19%-50% by 2035, and it is projected that almost 280 million people will get connected to the electricity by 2030 in NPS scenario (IEA, 2012).
The IEA estimated the global electrification rate at 80.5% with 1.32 billion people living without electricity. Thus, according to the NPS scenario, almost one billion people will live without electricity access by 2030 (IEA, 2012). Both theoretical and most empirical studies have demonstrated a causal relationship between per capita electricity consumption and per capita GDP globally. Though the empirical results are varied and sometimes conflicting (Maza, Villaverde, 2008), the conclusions derived from most of these studies show that the causality is running from energy and electricity consumption to GDP (Ozturk and Acaravcia, 2010; Chontanawat et al., 2008). Thus, electricity is a limiting factor to economic growth and, therefore, lack of electricity supply or shocks/interruptions of supply will have a negative impact on economic growth.
In 1990-2008, almost two billion people got access to electricity globally (GEA, 2012) and global electricity consumption in 2008 was 1.7 times higher than in 1990. This provides a basis to believe that electrifying the remaining 1.3 billion people without electricity is feasible. However, if universal access to electricity is achievable by 2030 (GEA 2012), the challenge to increase of electricity usage up to OECD level is tremendous and the results presented in this paper show the magnitude of the challenge for developing countries in electricity convergence.
Note: This paper is based upon a presentation by its authors at the 17th Annual IAEE/ASSA Energy Economics Meeting, Boston, Massachusetts, January 3-5, 2015. "Economics of The Global Energy Transition", http://www.iaee.org/documents/2015/IAEE-AEA_Boston.pdf.
Amaral et al. (2005), Amaral, S., Câmara, G., Miguel, A., Monteiro, V., Quintanilha, J.A., Elvidgec, C.D., Estimating population and energy consumption in Brazilian Amazonia using DMSP nighttime satellite data, Computers, Environment and Urban Systems, Volume 29, Issue 2, March 2005, pp. 179-195.
Bhandari & Roychowdhury (2011), Night Lights and Economic Activity in India: A study using DMSP-OLS night time images, Proceedings of the Asia-Pacific Advanced Network 2011 v. 32, pp. 218-236.
Chen, Xi., Nordhaus, W. (2011), Using luminosity data as a proxy for economic statistics, The Proceedings of National Academy of Sciences, Volume 108, Issue 21, pp. 8589–8594.
Chontanawat et al. (2008), Chontanawat, J., Hunt, L., Richard Pierse, R., Does energy consumption cause economic growth?: Evidence from a systematic study of over 100 countries, Journal of Policy Modeling, Volume 30, Issue 2, March–April 2008, pp. 209–220.
Doll et al. ( 2000), Doll, C.N.H., Muller, R.J.P., Elvidge, C.D., Nighttime imagery as a tool for global mapping of socio-economic parameters and greenhouse gas emissions, Ambio, Volume 29, Issue 3, 2000, pp. 157–162.
Doll, C.N.H, Pachauri, S.(2010), Estimating rural populations without access to electricity in developing countries through nighttime light satellite imagery, Energy Policy, Volume 38, Issue 10, October 2010, pp. 5661–5670.
Ebener et al., (2005), Ebener, S., Murray, C., Tandon, A., Elvidge, C.D., From wealth to health: Modeling the distribution of income per capita at the sub-national level using nighttime light imagery, International Journal of Health Geographics, Volume 4, 2005, pp. 1–17.
Elvidge et al. (1997), Elvidge, C.D. , Baugh, K.E. , Kihn, E.A., Relation between satellite observed visible-near infrared emissions, population, economic activity and electric power consumption, Int. J. Remote Sensing, Volume 18, Issue 6,1997, pp. 1373–1379.
Elvidge et al. (2007), Elvidge C.D., Benjamin T. T., Sutton P.C, Kimberly E. B., Ara T. H., Milesi, C., Bhaduri, B.L., and Nemani R., Global Distribution and Density of Constructed Impervious Surfaces, Sensors, 2007, Volume 7, pp.1962-1979, online: http://ecocast.arc.nasa.gov/pubs/pdfs/2007/ISA_Sensors.pdf
Elvidge et al., (2009), Elvidge, C.D., Sutton, P.C., Ghosh, T. , Tuttle, B.T. , Baugh, K.E., Bhaduri B., A global poverty map derived from satellite data, Computers & Geosciences, Volume 35, Issue 8, August 2009, pp. 1652–1660.
Elivdge et al. (2011), Elvidge, C.D., Baugh, K.E., Sutton, P.C., Bhaduri, B., Tuttle, B.T., Ghosh, T., Ziskin, D.,Who's in the Dark—Satellite Based Estimates of Electrification Rates, in Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment, Yang, X. ed., Chapter 15, Published Online: 13 APR 2011, DOI: 10.1002/9780470979563.ch15; online: http://onlinelibrary.wiley.com/doi/10.1002/9780470979563.ch15/summary
Feige, E.L., Urban, I, (2008), Measuring underground (unobserved, non-observed, unrecorded)economies in transition countries: Can we trust GDP, Journal of Comparative Economics, Elsevier, vol. 36(2), pp. 287-306.
GEA (2012), Global Energy Assessment: Toward a Sustainable Future. IIASA, Laxenburg, Austria and Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Econ (2011), Geographically based Economic data, Yale University, online: http://gecon.yale.edu/
Ghosh et al. (2010), T. Ghosh, R.L. Powell, C. Elvidge, K. Baugh, P. Sutton, S. Anderson, Shedding light on the global distribution of economic activity, The Open Geography Journal, Volume 3, 2010, pp. 147–160.
Hattori, R. et al. (2013), Hattori, R., Sadataka Horie, S., Feng-Chi Hsu, F.-C., Chirstopher D. Elvidge, C.D., Yasunari Matsuno, Y., Estimation of in-use steel stock for civil engineering and building using nighttime light images, Resources, Conservation and Recycling, Available online 14 December 2013, In Press, Corrected Proof.
Henderson V, Storeygard A, Weil D (2009), Measuring Economic Growth from Outer Space, NBER Working Paper w15199, National Bureau of Economic Research, Cambridge MA, 2009. Online: http://www.nber.org/papers/w15199.pdf
Henderson et al. (2012), Henderson, V., Storeygard, A. & Weil, D., Measuring Economic Growth from Outer Space. American Economic Review, Volume 102, Issue 2, pp. 994-1028.
IEA (2012), World Energy Outlook 2012, International Energy Agency, OECD/IEA, Paris, 2012.
IEA (2013), World Energy Outlook 2013, International Energy Agency, OECD/IEA, Paris, 2013.
Joyeux and Ripple (2007), Joyeux, R. and Ripple, R.D., Household energy consumption versus income and relative standard of living: A panel approach, Energy Policy, Elsevier, Volume 35, Issue 1, pp.50-60.
Kiran et al. (2009), Kiran Chand, T.R, Badarinath, K. V. S., Elvidge, C.D and Tuttle, B.T, Spatial characterization of electrical power consumption patterns over India using temporal DMSP‐OLS night‐time satellite data, International Journal of Remote Sensing, Volume 30, Issue 3, 2009, pp. 647-661.
Kulkarni et al. (2011), Kulkarni, R., Haynes, K., Stough, R. and Riggle, J., Revisiting Night Lights as Proxy for Economic Growth: A Multi-year light based growth indicator (LGBI) for China, India and the U.S., GMU School of Public Policy, Research Paper, 03/2011, DOI: 10.2139/ssrn.1777546.
Letu et al.(2010), Letu, H., Hara, M., Yagi, H., Naoki, K., Tana, G., Nishio, F., Shuhei, O., Estimating energy consumption from nighttime DMSP/OLS imagery after correction for saturation effects, International Journal of Remote Sensing, Volume 31, Issue 16, May 2010, pp. 4443-4458.
Lo, C.P.(2001), Modeling the Population of China Using DMSP Operational Linescan System Nighttime Data, Photogrammetric Engineering & Remote Sensing, Volume 67, Issue 9, September 2001, pp. 1037-1047.
Ma et al. (2012), Ma, T., Zhou, C., Pei, T., Haynie, S., Fan, J., Quantitative estimation of urbanization dynamics using time series of DMSP/OLS nighttime light data: A comparative case study from China's cities, Remote Sensing of Environment, Volume 124, pp. 99-107.
Maza, A., Villaverde, J. (2008), The world per capita electricity consumption distribution: Signs of convergence?, Energy Policy, Volume 36, Issue 11, November 2008, pp. 4255–4261.
NOAA (2014), Version 4 DMSP-OLS Nighttime Lights Time Series, F18 2010 Nighttime Lights Composite, online: http://ngdc.noaa.gov/eog/dmsp/downloadV4composites.html
Nordhaus et al. (2006), Nordhaus, W., Azam, Q., Corderi, D., Hood, K., Victor, N., Mohammed, M., Miltner, A., and Weiss, J., The G-Econ Database on Gridded Output: Methods and Data, Yale University, May 12, 2006.
Nordhaus, W. and Chen, Xi., (2014), A sharper image? Estimates of the precision of nighttime lights as a proxy for economic statistics, Journal of Economic Geography, Advance Access published May 7, 2014, pp. 1–30.
Ozturk, I., and Acaravci, A., (2010),The causal relationship between energy consumption and GDP in Albania, Bulgaria, Hungary and Romania: Evidence from ARDL bound testing approach, Applied Energy, Volume 87, Issue 6, June 2010, pp. 1938-1943.
Sutton, P. (1997), Modeling Population Density with Nighttime Satellite Imagery and GIS, Computers, Environment and Urban Systems, Volume 21, Issue 3/4, pp. 227-244.
Sutton et al. (2007), Sutton, P.C., Ghosh, T., Elvidge, C.D., Estimation of Gross Domestic Product at Sub-National Scales using Nighttime Satellite Imagery International Journal of Ecological Economics & Statistics, Summer 2007, Volume 8, No. S07, pp.5-21.
Sutton et al. (2011), Sutton, P.C., Anderson, S., J., Elvidge, C.D., Tuttle, B. T., Ghosh, T., Paving the Planet: Impervious Surface as a Proxy Measure of the Human Ecological Footprint Progress in Physical Geography, Volume 33, Issue 4, pp. 1-188.
Takahashi et al. (2009), Takahashi, K.I. , Terakado, R., Nakamura, J., Daigo, I., Matsuno, Y. , Adachi, Y., In-use stock of copper analysis using satellite nighttime light observation data, Materials Transactions, Volume 50, 2009, pp. 1871–1874.
Wang et al. (2012), Wang, W., Cheng, H., Zhang, L., Poverty assessment using DMSP/OLS nighttime light satellite imagery at a provincial scale in China, Advances in Space Research, Volume 49, Issue 8, 15 April 2012, pp. 1253–1264.
WB (2014), World Development Indicators, World Bank, 2014, online: http://data.worldbank.org/indicator
Wu et al. (2013), Wu, J., Wang, Z., Li, W., Peng, J. Exploring factors affecting the relationship between light consumption and GDP based on DMSP/OLS nighttime satellite imagery, Remote Sensing of Environment, Volume 134, July 2013, pp. 111–119.
Zhao et al. (2011), N.Z. Zhao, N. Currit, E. Samson, Net primary production and gross domestic product in China derived from satellite imagery, Ecological Economics, 70 (2011), pp. 921–928