Proceedings of the 30th Asia-Pacific Advanced Network Meeting
Disaggregation of national fossil fuel CO2 emissions using a global power plant database and DMSP nightlight data Tomohiro Oda 1,*, Shamil Maksyutov 1, and Christopher D. Elvidge 2 1 National Institute for Environmental Studies, Center for Global Environmental Studies / 16-2 Onogawa Tsukuba Ibaraki 305-8506 Japan. 2 National Oceanic and Atmosphere Administration, National Geophysical Data Center / Boulder, Colorado 80305 USA. E-Mails: [email protected]
; [email protected]
; [email protected]
* Author to whom correspondence should be addressed; Tel.: +81-(0)29-850-2968; Fax: +81(0)29-850-2219
Abstract: National fossil fuel CO2 emissions are often available in gridded form. Previous gridded inventories of fossil fuel CO2 emissions were made by using population statistics as proxies for their spatial distribution. Presumed in the use of such statistics is correlation between human population and their activities. That assumption is valid at national and state levels. The correlation, however, gets weaker beyond those spatial scales. A better approach was necessary for high-resolution emission mapping. We have proposed a two-step scheme to disaggregate national CO2 emissions and developed fossil fuel CO2 emission gridded inventory for 1980-2007 [Oda and Maksyutov, 2010]. Power plant emissions were allocated to their location based on a published power plant database. Emissions from other sources were distributed using DMSP satellite-observed nightlight dataset. The resulting inventory showed good agreement with a US high-resolution inventory, however, it was constructed using a single radiance calibrated lights product from 1996-97, which until recently was only available product of its kind. Here, we developed a gridded emission inventory for the year 2006 using the 2006 radiance lights product which was recently processed by the National Geophysical Data Center (NGDC) of the National Oceanic and Atmosphere Administration (NOAA). The 2006 radiance lights did not have saturated pixels and is well correlated with population data over major emitting countries. Due to the improvements in the 2006 radiance lights, source
regions in suburb areas were well depicted and were much larger compared to the inventory constructed using the 1996-97 radiance calibrated lights data. Keywords: 2006 radiance light; fossil fuel CO2 emission; proxy
1. Introduction Spatial inventories of atmospheric emissions such as carbon dioxide (CO2) are essential for atmospheric modeling. They are often available in gridded form and widely used in global and regional scale simulations using chemical transport models (CTMs). The US department of Energy Carbon Dioxide Information Analysis Center (CDIAC) maintains a continuous archive of global, regional and national fossil fuel CO2 emission data [Marland et al., 2008] and provides annual 1◦ ×1◦ emission gridded data (1751-2006) [e.g. Andres et al., 1999]. The statistics of population distribution has been used to approximate spatial distribution of national fossil fuel CO2 emissions, assuming high correlation between population with human activities (hence, CO2 emissions) [e.g. Andres et al., 1999]. Population data can be used as a good proxy at spatial scales of country and state levels where the high correlation is retained. Gridded fossil fuel CO2 emission datasets, which were constructed using population distribution, have been used in many atmospheric CO2 inversions as an input [e.g. Gurney et al., 2002]. Atmospheric CO2 inversion is an established method to estimate regional CO2 fluxes using atmospheric transport models and observations [e.g. Gurney et al., 2002]. Fossil fuel CO2 emissions are treated as known quantities in many regional flux inversions and the potential bias associated with flux estimates would be increased if spatial pattern of emissions was reduced [Gurney et al. 2004]. More recent studies suggested the use of satellite-observed CO2 data for flux inversions [e.g. Chevallier et al., 2007]. The Greenhouse gas Observing SATellite (GOSAT) [Yokota et al., 2009] was launched in 2009, and the GOSAT data product are now available. The space-borne CO2 monitoring can fill the gaps in the existing surface observation network and thus permits analysis of regional sources and sinks on much higher spatiotemporal scales. To fully utilize the satellite-based CO2 data, including those observed over polluted continental regions, for obtaining better inversion estimates, more spatiotemporally-detailed inventories are needed. The use of the conventional methods of disaggregating national emissions is valid at the country and state levels. However, the correlation between the population and human activities would become weak beyond those spatial scales (see Table 2 in Rayner et al. ). This is because intense CO2 sources such as power generations and highways are usually not co-located with human residence, for instance. Therefore, the development of high-resolution gridded 221
datasets of fossil fuel CO2 emissions is a key requirement for the emerging satellite-based inversions. Satellite-observations of nightlights have been identified as useful proxies for human activities and used for constructing CO2 emission maps [e.g., Doll et al., 2000; Rayner et al., 2010; Oda and Maksyutov, 2010]. Nightlight data obtained by the DMSP-OLS satellite can be used to specify the locations of human activities such as cities, human settlements, gas flaring, and fishing boats [e.g. Elvidge et al., 1997]. The DMSP nightlight data are available at the increment of 30 arc second and provided through the web page of NOAA/NGDC (http://www.ngdc.noaa.gov/dmsp/download.html, last access: Aug 27, 2010). One known difficulty in using regular nightlight data for emission mapping is that the light detected over major cities are often too bright for the instrument and thus the details are lost in saturated pixels. Rayner et al.  corrected those saturated pixels using population data based on a method described by Raupach et al. . We developed a high-resolution gridded inventory of global fossil fuel CO2 emissions using radiance calibrated data [Elvidge et al., 1999], which have less saturated pixels, together with a power plant database. The resulting gridded map was compared with several existing inventories using the US Vulcan emission map (10 km x 10 km resolution) as a reference, and was the closest to the reference data among the others at spatial scales which are useful to emerging satellite-based flux inversions [Oda and Maksyutov, 2010]. The new emission dataset, however, has several sources of errors and uncertainties. For example, 27-year data set was constructed using single nightlight dataset. Emission map for the year 2006 was based on 2006 emission estimates and 1996-97 nightlight data, as radiance calibrated data was only available for 1996-97. In this paper, we re-constructed a gridded inventory for year 2006 using 2006 radiance lights product which was recently processed by NOAA/NGDC.
2. Method and data used in this study In this section, the method we employed to construct a gridded inventory for the year 2006 is briefly described. Further details of the method are described in Oda and Maksyutov .
2.1. Estimation of national emissions Numbers for total emissions for 65 countries and 6 geographical regions for the remaining countries and regions were taken from Oda and Maksyutov . National and regional 222
emissions were estimated using the statistical review of world energy prepared by BP p.l.c. (http://www.bp.com/productlanding.do?categoryId=6929&contentId=7044622, last access: Aug 23, 2010). Total amount of consumed coal, oil and gas were converted into CO2 emission by multiplying emission factors. In addition to estimation of national and regional total emissions, national and regional total emissions attributable to power plants were estimated using a global power plant database CARMA (CARbon Monitoring and Action, http://carma.org/, last access: Aug 20, 2010). CARMA includes power plant attributes such as name, company, annual total CO2 emission and locations. Total power plant emission for a country was calculated as a sum of power plant emissions attributable to a country. By subtracting the total power plant emission from a national total emission, emissions from other sources (non-point sources) can be approximated.
2.2. Disaggregation of national emissions The different mapping methods were used to determine the spatial distribution of power plant emissions and emissions from other sources: 1) Power plant emissions were placed to exact locations using latitude and longitude CARMA indicates. 2) Emissions from other sources were distributed according to the spatial distribution of nightlights. To distribute non-point emissions, we used 2006 radiance lights prepared by NOAA/NGDC. 2006 radiance lights is a merged product of the stable lights data and data obtained in fixed gain mode and has no saturated pixels. By including the stable lights, 2006 radiance lights can be used to depict the dimmer lighting which can not seen in previous radiance calibrated lights. National (or regional) total emissions from non-point sources were distributed to 30 arc second (1 km × 1 km) pixels according to the distribution of nightlights. The distribution was formed by superimposing the nightlight data and national boundary data, which were used to identify the national attributes of pixels. The national boundary data was prepared in 2.5 arc minute (approximately 5 km) (year 2000) by the Center for International Earth Science Information Network (CIESIN) at Columbia University, NY, USA (Gridded Population of the World version 3 (GPWv3), http://sedac.ciesin.columbia.edu/gpw/global.jsp, last access: Aug 23, 2010). Radiance light quantities across all pixels attributed to a nation (or a region) were 223
summed, and the original quantity at each pixel was normalized by the national (or regional) sum. The CO2 emission intensity at a pixel was obtained by multiplying the normalized radiance with the annual total non-point emission (national total emission minus total power plant emission) of a country or a region. Boundaries between land and ocean, river, and water bodies (e.g. coastline) were defined using IGBP land-cover classification data of the NASA TERRA/MODIS HDF-EOS MOD12Q1 V004 product (http://duckwater.bu.edu/lc/mod12q1.html, last access; Aug 27, 2010). As the same manner as Oda and Maksyutov , pixels attributable to gas flaring were removed. Emissions from gas flaring will be supplemented by exiting global gas flare estimates by NOAA/NGDC (http://www.ngdc.noaa.gov/dmsp/interest/gas_flares.html, last access Aug 20, 2010) and thus, gas flaring emissions could be distinguished from other mixed land emissions. Gas flaring pixles were identified using shapefiles of gas flaring provided by NOAA/NGDC.
3. Results 3.1. Correlation between radiance lights and population Correlation between nightlights and population could be a measure for evaluating nightlights as a proxy for human activities (hence CO2 emissions). Correlations between the 2006 radiance lights and population were calculated for 19 major emitting countries (United States, China, India, Russian Federation, Japan, Germany, Canada, United Kingdom, South Korea, Iran, Mexico, Italy, South Africa, Saudi Arabia, Indonesia, Australia, France and Spain) and Belgium (See Figure 1). The 19 countries (excluding Belgium) are listed as top 20 emitting countries by CDIAC estimate for year 2007. National total emission estimate for Ukraine, which is the 20th emitting country in CDIAC list, is not available in Oda and Maksyutov . As the same in the analysis by Raupach et al. (2009), the analysis was performed by reducing the spatial resolution of data to 0.25 × 0.25 arc degree. Radiance values plotted in the figures were original numbers and conversion factor was not applied. Among the twenty countries, overall population and nightlights are well correlated. Also, apparent decrease in correlation due saturated pixels, which are presented in Raupach et al. , cannot be seen. The correlation coefficient over Belgium (including Luxemburg in this study) was found to be fairly high (0.95). On the other hand, correlation is relatively low in China (0.27). Iran (0.14) and Saudi Arabia (-0.12), that can be explained by rural populations and the low resolution of the CIESIN population dataset for those countries. High correlations over many emitting countries suggested that major portion of global total emission was distributed based on fair assumption. 224
Figure 1. Scatter plots of radiance lights intensity against population density. From top left to right, United States, China, India, Russian Federation, Japan, Germany, Canada, United Kingdom, South Korea, Iran, Mexico, Italy, South Africa, Saudi Arabia, Indonesia, Australia, France, Spain and Belgium. Conversion factor was not applied to the radiance numbers plotted. Data was reduced to 0.25 × 0.25 arc degree according to Raupach et al. .
3.2. Emission map for the year 2006 A global picture of CO2 emissions is shown in Figure 2. The figure was drawn using emission data reduced to 5 km × 5 km (2.5 arc min) resolution for the computational efficiency. As same as the map constructed by Oda and Makyutov , many lights can be found in the northern hemisphere especially over emitting countries and regions. On the other hands, massive 225
source regions cannot be found in southern hemisphere. In Southern Hemisphere, lights can often be found along the coast. Compared to the map of Oda and Maksyutov , source regions are much larger and look bright. Those features are coming from the nature of the 2006 radiance lights.
Figure 2. Global emission map for the year 2006. The map was drawn using a reduced 5 km × 5 km resolution dataset. CO2 emissions are expressed in the unit of 5km-2 year-1.
Figure 3. Comparison of emission maps over East Asia. (a) Emission map constructed using radiance calibrated lights 1996-97 and (b) Emission map constructed using 2006 radiance lights. Since national total emissions are the same, the same national and regional emission estimates were used, the difference can be seen in figures are sorely attributable to the difference in nightlight data. The plots were drawn using original 1 km × 1 km resolution emission dataset. Therefore, the actual size of each pixel is 1 km × 1 km. Note that the maps are drain in slightly different scales. In figure 3, emission maps over East Asia are shown. Compared to the previous map (a), the new map shows emissions more widely distributed across the land surface (b). This is, as mentioned above, due to the inclusion of stable lights for the detection of dim lighting in the 2006 radiance lights product. Note that the lack of dim lighting in the 1996-97 radiance lights concentrates the allocated emissions into the primary urban centers, with no representation of the emissions that are more widely spread across the populated land surface. On the other hand, emissions might be overestimated over areas depicted by the dimmest lights in the new map. The nightlight product is an annual composite and the dimmest lights do not always represent steady state lighting. The application of the combination of the mapping method by Oda and Maksyutov  and the 2006 radiance lights to such areas need to be evaluated.
4. Conclusions We constructed an emission map for the year 2006 using 2006 radiance lights data. Correlations between population and 2006 radiance lights over major emitting countries are 227
overall high. This suggests that large portion of global emission was distributed using fair proxy. Due to the inclusion of the stable lights in 2006 light data, emissions over suburb regions were well depicted. As a result, the source regions in new map are much larger than previous map. Here we note that emissions might be overestimated over areas depicted by the dimmest lights. The evaluation is necessary for the use of the 2006 radiance lights in the mapping method by Oda and Maksyutov  over such areas. As a future work, we will conduct global and regional evaluations using existing inventories. As radiance observation have been made since 1992, new radiance lights data after 1992 could be produced. When they become available, we will re-construct gridded emission inventories using the new dataset.
Acknowledgements This study was conducted as a part of the GOSAT project promoted by the Japan Aerospace Exploration Agency (JAXA), the National Institute for Environmental Studies (NIES), Japan, and Ministry of the Environment (MOE), Japan. Power plant data used in this study were taken from CARMA (CARbon Monitoring and Action, www.carma.org/). Population and national identifier data (GPWv3) were provided by the Trustees of Columbia University in the City of New York, the United Nations Food and Agriculture Programme (FAO), and the centro Internacional de Sgricultura Tropical (CIAT). Image and data processing were done by NOAA's National Geophysical Data Center. DMSP data were collected by US Air Force Weather Agency. We would like to thank three anonymous referees for their useful comments to improve the manuscript of this paper. References 1.
Marland, G.; Boden, T. A.; Andres, R. J. Global, Regional, and National Fossil Fuel CO2 Emissions, in: Trends: A Compendium of Data on Global Change, Carbon Dioxide Information Analysis Center, Oak Ridge National Laborator y, US Depar tment of Energy, Oak Ridge, Tenn., USA, available at: http://cdiac.ornl.gov/trends/emis/over view.html (last access: 30 June 2010) 2008. 2. Andres, R. J.; Marland, G.; Fung, I.; Matthews, E. A 1◦ ×1◦ distribution of carbon dioxide emissions from fossil fuel consumption and cement manufacture, 1950–1990, Global Biogeochem. Cy. 1996, 10, 419–429. 3. Gurney, K. R.; Rachel, L. M.; Denning, A. S.; Rayner, P. J.; Baker, D.; Bousquet, P.; Bruhwiler, L.; Chen, Y.-H.; Ciais, P.; Fan, S.; Fung, I. Y.; Gloor, M.; Heimann, M.; Higuchi, K.; John, J.; Maki, T.; Maksyutov, S.; Masarie, K.; Peylin, P.; Prather, M.; Pak, B. C.; 228
Randerson, J.; Sarmiento, J.; Taguchi, S.; Takahashi, T.; Yuen, C.-W. Towards robust regional estimates of CO2 sources and sinks using atmospheric transport models, Nature 2002, 415, 626– 630. 4. Gurney, K. R.; Chen, Y.-H.; Maki, T.; Kawa, S. R.; Andrews, A.; Zhu, Z. Sensitivity of atmospheric CO2 inversions to seasonal and interannual variations in fossil fuel emissions, J. Geophys. Res. 2005, 110, D10308, doi:10.1029/2004JD005373. 5. Chevallier, F.; Bréon, F.; Rayner, P. J. Contribution of the Orbiting Carbon Observatory to the estimation of CO2 sources and sinks: theoretical study in a variational data assimilation framework, J. Geophys. Res. 2007, 112, D09307, doi:10.1029/2006JD007375. 6. Yokota, T.; Yoshida, Y.; Eguchi, N.; Ota, Y.; Tanaka, T.; Watanabe, H.; Maksyutov, S. Global concentrations of CO2 and CH4 retrieved from GOSAT: First preliminary results, SOLA 2009, 5, 160–163, doi:10.2151/sola.2009-041. 7. Doll, C. N. H.; Muller, J.-P.; Elvidge, C. D. Night-time imagery as a tool for global mapping of socioeconomic parameters and greenhouse gas emissions, Ambio 2000, 29, 157–162. 8. Rayner, P. J.; Raupach, M. R.; Paget, M.; Peylin, P.; Koffi, E. A new global gridded dataset of CO2 emissions from fossil fuel combustion: 1: Methodology and evaluation, J. Geophys. Res. 2010, 115, D19306, doi:10.1029/2009JD013439.. 9. Oda, T.; Maksyutov, S. A very high-resolution global fossil fuel CO2 emission inventory derived using a point source database and satellite observations of nighttime lights, 1980– 2007, Atmos. Chem. Phys. Discuss. 2010, 10, 16307-16344, doi:10.5194/acpd-10-163072010, http://www.atmos-chem-phys-discuss.net/10/16307/2010/acpd-10-16307-2010.html 10. Elvidge, C. D.; Baugh, K. E.; Kihn, E. A.; Kroehl, H. W.; Davis, E. R. Mapping city lights with nighttime data from the DSMP operational linescan system, Photogramm. Eng. Rem. S. 1997, 63, 727–734. 11. Raupach, M. R.; Rayner, P. J.; Page, M. Regional variations in spatial structure of nightlights, population density and fossil-fuel CO2 emissions, Energ. Policy 2009, 38(9), 4756–4764, doi:10.1016/j.enpol.2009.08.021. 12. Elvidge, C. D.; Baugh, K. E.; Dietz, J. B.; Bland, T.; Sutton, P. C.; Kroehl, H. W. Radiance calibration of DMSP-OLS low-Light imaging data of human settlements – a new device for portraying the Earth’s surface entire, Remote Sens. Environ. 1999, 68, 77–88. Accepted October 14, 2010 Published December 31, 2010 © 2010 by the authors; licensee Asia Pacific Advanced Network. This article is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/). 229