SPECIAL REPORT: Satellite data estimate worldwide flared gas volumes

Nov. 12, 2007
For the past several years, a study led by the National Oceanic and Atmospheric Administration (NOAA) has developed procedures for independent estimation of flared gas volumes worldwide using satellite remote sensing.

For the past several years, a study led by the National Oceanic and Atmospheric Administration (NOAA) has developed procedures for independent estimation of flared gas volumes worldwide using satellite remote sensing.

None of the currently available earth observation sensors is designed or flown specifically for observing gas flaring, although several satellite systems can detect gas flares based on the radiative emissions from the flames.

Oil producers often flare associated gas because an area has no infrastructure for marketing or using the gas. Venting the gas, without combustion, is another alternative used in these production areas without infrastructures.

While producers keep reasonable records on the quantities of oil and natural gas brought to market, the quantities of natural gas flared are much less known.

Gas flaring wastes an energy resource, is a source of local pollution due to trace chemicals in the gas, and adds to the carbon burden in the global atmosphere.

In recent years many companies, governments, and international groups have sought to reduce gas flaring. The World Bank’s global gas flaring reduction (GGFR) initiative has published gas flaring estimates for 20 countries and has worked with individual countries to encourage alternatives to gas flaring.

Because there are no requirements for reporting gas flaring volumes, it is difficult to know whether these efforts are succeeding in reducing gas flaring.

Satellite systems

Satellite systems vary substantially in terms of their spatial resolution, repeat cycle, and collection-processing cost.

High spatial-resolution imaging systems enables the detection and accounting of the number of individual gas flares at a site. The Google Earth (GE) system provides access to a substantial quantity of high spatial-resolution satellite imagery for sites around the world acquired by the Digital Globe Corp. If one knows where to look in GE, it is possible to find active gas flares.

The image on the left shows nine individual flares on Aug. 21, 2003, while in the image on the right only the ninth flare was active at the site on Sept. 24, 2007 (Fig. 1).
Click here to enlarge image

Fig. 1 shows a set of nine gas flares at three closely spaced sites located in GE. The Digital Globe imagery on the left was collected on Aug. 21, 2003. As a comparison, a newer image, on the right, collected by Digital Globe on Sept. 24, 2007, found only one of the nine flares active.

The loss of eight gas flares could indicate that gas flaring was reduced, or it might be chance that the other eight were not active at the time of the second image collection. The observations clearly require a larger number of observations to document a reduction in gas flaring.

The slow revisit rate and high cost of acquiring and analyzing high-resolution imagery makes it impractical to consider global monitoring of gas flares with these sources. The alternative is to monitor gas flaring with meteorological or environmental satellite data having coarser spatial resolution, global coverageand higher temporal frequency.

Global map

To find the gas flares in Fig. 1, we required prior knowledge of gas flaring locations and zoomed in on these gas flares with a global map of gas flares generated with coarse spatial resolution (>1 sq km) high-temporal-frequency satellite imagery acquired by the US Air Force defense meteorological satellite program (DMSP) operational line-scan system (OLS).

OLS collects global cloud imagery with a pair of broad spectral bands placed in the visible and thermal ranges. One benefit of working with data sources such as OLS is that it collects many observations every year.

The DMSP satellites are in polar orbits and each collects images from 14 orbits/day. With a 3,000 km swath width, each OLS can collect a complete set of global nighttime images in 24 hr. At night, a photomultiplier tube (PMT) intensifies the visible band signal, enabling OLS to detect moonlit clouds.

The boost in gain enables the detection of lights present at the earth’s surface. Most lights are from outdoor lighting in cities and towns. The OLS sensor can also detect wildfires and agricultural burning, but these are short-lived. It also can detect light from gas flares, which can be identified easily when they are offshore or in isolated areas not affected by urban lighting.

This image shows gas flaring regions of the world in a Mollweide 1-km equal area projection. The images are color composites made with the annual data from 1992 (blue), 2000 (green) and 2006 (red). Note that the colors of the flares indicate their activity patterns during the 3 years used in the color composite. Flares active in 2006 but not in 2000 or 1992 are red. Those active in 2006 and 2000 are yellow. Those active in 2000 but not in 1992 or 2006 are green. Those active in 1992 but not in 2000 or 2006 are blue (Fig. 2).
Click here to enlarge image

The study involved the processing of a time-series of annual cloud-free composites of OLS nighttime lights spanning 1992 through 2006. Gas flares were identified visually in the nighttime lights composites based on three characteristics:

  1. Because gas flares are bright point sources of light with no shielding to the sky, they tend to form circular lighting features with a bright center and wide rims.
  2. Most gas flares are active for a period of years, but a few gas flares persist with little change in intensity for a decade. Thus many gas flares exhibit color in color-composite images made with data from the beginning, middle, and end of the nighttime lights time series.
  3. Gas flares tend to be in remote locations, outside of urban centers. Offshore gas flares are easy to identify. For onshore gas flares, the study reviewed a US Department of Energy population density grid to evaluate lights identified as potential gas flares. Vector polygons were created to outline and identify gas flares for individual countries. The study found gas flaring in 60 countries.

By combining the polygons for all of the countries, the study extracted a global map of gas flaring as observed by the OLS for each year from 1992 to 2006. Fig. 2 is a colorized version of the global map of gas flaring made with flaring from 1992 as blue, 2000 as green and 2006 as red. This global map is available at http://www.ngdc.noaa.gov/dmsp/interest/gas_flares.html.

Four major gas flaring regions of the world in a Mollweide 1-km equal area projection are the Gulf of Guinea region of Africa, Western Siberia, Persian Gulf (third image), and North Africa. The images are color composites made with the annual data from 1992 (blue), 2000 (green), and 2006 (red) (Fig. 3).
Click here to enlarge image

Areas with many gas flares include the Gulf of Guinea, North Africa, Persian Gulf, and Western Siberia (Fig. 3).

Estimated volumes

The study developed a method for estimating gas flaring volumes for individual countries based on a “sum-of-lights index” and a set of reported gas flaring volumes for countries (from the GGFR) and individual flares.

The “sum-of-lights” index values are the tallies of the digital number values extracted for the gas flares of a particular country or an individual flare. The values are intercalibrated so that the data from each year can be pooled and quantitatively compared. Note that the study was unable adequately to intercalibrate the data from the early part of the OLS record (1992-94).

Click here to enlarge image

The steps in developing the calibration included removal of outliers, regression modeling, and establishing the prediction interval of the model (Fig. 4).

Following is the model:

Volume in billion cu m =
0.00002646 × Sum of lights index
R 2 = 0.978
P-value <2.2e-16

The study used this model to estimate the billion cu m of flared gas for the individual countries with a 90% prediction interval. The prediction interval of the regression model is about 1.61 billion cu m, which defines the upper and lower bound for the volume estimates as a form of an error bar for each estimate.

Click here to enlarge image

Gas flaring volume estimates for individual countries were based on the sum of lights index values. Fig. 5 shows the estimates for individual countries or areas for 2004. The estimates for Russia are broken out for the autonomous territory of Khanty-Mansiysk and the rest of Russia.

The combined estimates indicate that gas flared in Russia was about twice as much as in Nigeria.

The combined volume estimates for all countries provides an estimate of global gas flaring (Fig. 6).

Click here to enlarge image

The figure indicates that overall flaring from 1995 to 2006 has remained largely stable at 150-170 billion cu m. There were dips in gas flaring in 1999 and 2002. Gas flaring increased by more than 10 billion cu m 2002-03 and then declined for 2 years after that before increasing again in 2006.

Study results

The study provides gas flaring estimates for 60 countries or areas, tripling the number listed by GGFR.

While Nigeria has been widely reported as the country with the largest volume of gas flaring, satellite data indicate that Russia flares twice as much gas as Nigeria.

Global gas flaring has remained largely stable for the past 12 years, remaining in the range of 150 to 170 billion cu m.

The global gas flaring estimate for the year 2004 is 160 billion cu m, slightly higher than the GGFR estimate of 150 billion cu m. The DMSP estimate of 160 billion cu m of flaring in 2004 is 25% of the US natural gas consumption that year and represents an added carbon dioxide emission burden to the atmosphere of 400 million tonnes.

It should be noted that the study could not address several areas of uncertainty, including possible errors in the reported gas flaring volumes used in the calibration and environmental effects on flare size and brightness in the OLS data.

We fully expect gas flaring volumes estimates will improve in the future through the inclusion of multiple satellite data sources. It is also clear that improvements in satellite estimates of gas flaring will require reliable sources of in situ measurements of gas flaring volumes for calibration.

The independent estimates of gas flaring volumes from satellite observations will play a key role in guiding efforts to reduce gas flaring. In many cases national governments responsible for establishing the regulatory framework for resource extraction have not known the magnitude of the flaring.

Companies engaged in building infrastructure for using or marketing associated gas may also find the estimates useful. Also with the data, international petroleum companies can assess the efficacy of efforts made to reduce gas flaring in remote locations under the direction of their subsidiaries and contractors.

Satellite remote sensing has moved from a curiosity to an operational and vital capability in the effort to reduce and ultimately eliminate most gas flaring.


The study was funded by the World Bank global gas flaring reduction (GGFR) initiative. Dee Pack’s participation was funded by the Aerospace Corp. internal research and development program.

The authors

Chris Elvidge ([email protected]) leads the NOAA-NGDC earth observation group and has worked on satellite observed nighttime lights since 1994. Elvidge has a PhD in applied earth sciences from Stanford University.

Edward H. Erwin ([email protected]) is a physical scientist with the National Geophysical Data Center (NGDC). He has 12 years experience in ingesting and archiving DMSP satellite data at NGDC. Erwin has a BS in meteorology from Texas A&M University and an MS in space sciences from Utah State University.

Kimberly E. Baugh works as a scientific programmer for the Cooperative Institute for Research in Environmental Sciences at the University of Colorado, Boulder. She is the primary developer of the software used to produce the nighttime lights cloud-free composites. Baugh has an MS in applied math.

Benjamin T. Tuttle is an associate scientist at the Cooperative Institute for Research in Environmental Sciences at the University of Colorado, Boulder, and is currently a PhD student in the department of geography at the University of Denver. Tuttle holds a BA in geography and a BA in environmental studies from the University of Colorado and an MS in GIScience from the University of Denver.

Ara T. Howard writes software code at the University of Colorado.

Dee Pack ([email protected]) directs the remote sensing department in the physical sciences laboratories of the Aerospace Corp. and leads research on global monitoring with Department of Defense and civil agency satellite sensors. Pack has a BS from the University of Virginia and a PhD from Princeton University, both in physical chemistry.

Cristina Milesi ([email protected]) is a senior research scientist with the Foundation of California State University Monterey Bay, which is based at the NASA Ames Research Center. Milesi holds a PhD in remote sensing and ecosystem modeling from the University of Montana.