Deepwater Horizon REVISITED
n the early months of 2010, a technological marvel floated in the Gulf of Mexico, 50 miles off the Louisiana coast. The Deepwater Horizon was drilling the first oil well in the newly discovered Macondo Oilfield. The giant semisubmersible rig was 400 feet long, 250 feet wide and stood 14 stories tall. Designed to withstand heavy weather and operate in extreme deep-water environments, the rig set a record at the end of 2009 by drilling a well nearly seven miles into the Earth in water 4,000 feet deep.
Then on April 20, 2010, something went terribly wrong. At about 10 p.m., a series of explosions ripped through the rig, killing 11 workers and injuring 17 others. Fire raged unabated for nearly two days, as emergency teams raced to the site and poured seawater on the blaze. Despite their efforts, the rig listed heavily to one side, and on April 22—ironically Earth Day—the Deepwater Horizon slipped beneath the waves and plunged to the muddy seafloor 5,000 feet below, setting the stage for an oil leak of nearly unimaginable magnitude.
Satellite Imagery and Oil Slick Detection
The remote sensing community has seen repeatedly that satellite imagery and other datasets are useful for detecting and monitoring pollution at sea caused by offshore oil and gas development. But satellite images have limitations. Systems like the Moderate-Resolution Imaging Spectroradiometer (MODIS) instrument, carried on NASA’s Aqua and Terra satellites, measures visible to infrared wavelengths of light. Such systems rely on the sun for illumination and are impeded by clouds, fog, dust and haze. They’re also affected by the angle of the sun relative to the position of the satellite as it passes overhead.
Certain conditions of illumination geometry and sea state create a pattern of “sunglint” over the target area that can reveal the presence of oil slicks and sheen (very thin slicks), but those favorable conditions aren’t always met. For these reasons, satellite images aren’t uniformly suitable for mapping the full extent of large oil spill events at sea.
The use of radar imaging satellites helps fill some of the gaps in coverage left by visible-infrared sensors. Imaging radars create their own illumination of the target, beaming radar energy down to the ground as they pass overhead and measuring the radar energy that bounces back up to the sensor.
This long-wavelength microwave energy is able to penetrate clouds, dust, haze, fog and all but the heaviest tropical rainfall or hail, day or night, reliably imaging Earth’s surface. Because they’re sensitive to the “roughness” of the ocean surface, radar images can provide excellent detection of even thin oil slicks and sheen, which tend to smooth the ocean’s surface.
But other factors can create smooth patches on the ocean, including calm wind conditions, heavy rain, coastal upwelling, and the presence of oily surfactants emanating from phytoplankton blooms, coral spawn and other natural sources. High sea-state conditions due to strong surface winds can mechanically disaggregate oil slicks and overwhelm the tendency of the oil to smooth the ocean’s surface. Therefore, confident delineation of human-caused oil slicks depends on the analyst’s experience, understanding of the context of the oil spill and characteristics of the target area, knowledge of wind and weather conditions when the images are acquired, and familiarity with the radar sensor’s particular features.
Throughout the duration of the British Petroleum/Deepwater Horizon Oil and Gas Disaster, from April-August 2010, SkyTruth, a nonprofit organization that promotes environmental awareness with remote sensing and digital mapping technology, acquired satellite images collected on a daily basis from two main sources. Visible-infrared images were downloaded from NASA’s MODIS Rapid Response Web site. Radar images and visible-infrared images from a variety of non-U.S.-orbiting sensors were collected from the Center for Southeastern Tropical Advanced Remote Sensing (CSTARS) facility operated by the University of Miami.
What the Imagery Revealed
The following sections describe significant observations and conclusions, in chronological order, made by SkyTruth using imagery combined with other sources of information and analysis. The most important observations were that initial government and industry estimates of the daily spill rate were far too low and that a government report suggesting most of the spilled oil had dissipated by early August was overly optimistic and relied on unsupported assumptions. Throughout the spill and in its aftermath, SkyTruth’s analysis of satellite imagery provided timely and credible information that in key instances differed from official statements by industry and government sources.
Spill Overwhelms Response Capability
At about 7:30 a.m. EDT on April 21, 2010, news reports indicated there had been an explosion and fire aboard the Deepwater Horizon rig. The following day, the rig sank and was replaced by a giant oil slick.
Official estimates of the spill rate varied dramatically. On April 23, SkyTruth alerted colleagues at Florida State University and the CSTARS satellite imaging facility about the potential for a major oil spill, although the Coast Guard believed that no oil was leaking from either the sunken rig or the Macondo wellhead at the seafloor. The following day, the Coast Guard discovered that the well was indeed leaking at the seafloor at a rate estimated to be 42,000 gallons (1,000 barrels) per day, comfortably within the level of the response capability that BP and the Coast Guard had claimed a few days into the incident.
SkyTruth began acquiring visible-infrared satellite images from the MODIS sensors on NASA’s Aqua and Terra satellites. An image from April 25 showed obvious oil slicks and sheen covering more than 800 square miles. Most disturbing was another satellite image from that day taken by the Advanced Land Imager sensor on the European Space Agency’s Earth Observing (EO)-1 satellite (Figure 1). This was the first public image that showed a portion of the growing oil slick in great detail. Several skimmer vessels work around the edges of the slick to collect oil from the surface, but the scale of the response effort appears to be overwhelmed by the magnitude of the slick. The next good image SkyTruth obtained, taken on April 27, showed the slick had nearly tripled in size in just two days.
On April 29, despite the earlier Coast Guard assurances, oil began to come ashore on the Mississippi Delta. By May 1, satellite imagery showed the slick had reached 2,600 square miles, and SkyTruth got its first radar satellite image from CSTARS to supplement the visible-infrared imagery.
Spill Much Worse Than Reported
On April 27, SkyTruth and Dr. Ian MacDonald of Florida State University calculated that the spill rate was 210,000 gallons (5,000 barrels) per day at an absolute minimum to generate a slick covering 2,233 square miles in just seven days—much higher than the Coast Guard’s estimate of 42,000 gallons per day. Based on statements made that same day by a BP executive about the thickness of the oil slick, SkyTruth concluded the rate was more likely on the order of 840,000 gallons (20,000 barrels) per day, and more than 6 million gallons of oil already had been released into the Gulf during the first week of the spill.
The day after SkyTruth published its estimate, the National Oceanic and Atmospheric Administration (NOAA) weighed in on the rate of the spill, claiming it was 210,000 gallons (5,000 barrels) per day. BP objected, but ultimately accepted that estimate, and it remained the official spill-rate figure for the next four weeks.
SkyTruth published a refined estimate from Dr. MacDonald on May 13 that was based on a Coast Guard map of the oil slick, compiled from observations made during low-altitude reconnaissance overflights. At a rate of 1.1 million gallons (26,500 barrels) per day, SkyTruth predicted that the previous worst oil spill in U.S. history, the Exxon Valdez disaster, was surpassed that day.
Possibly due in part to pressure from SkyTruth, other scientists and the media calling on BP and the federal government to defend the 5,000 barrel per day spill-rate estimate, on May 19, the federal government convened a panel of scientists designated the Flow Rate Technical Group (FRTG), tasked with producing scientifically robust estimates of the spill rate. On June 10, the estimate was raised to 840,000-1.7 million gallons (20,000-40,000 barrels) per day. On June 15, the estimate was upped again to 1.5-2.5 million gallons (35,000-60,000 barrels) per day.
Finally, on Aug. 2, FRTG announced that the initial spill rate from the leaking Macondo Well was 2.6 million gallons (62,000 barrels) per day and that the amount of oil spewed into the Gulf during the duration of the spill totaled 172.2 million gallons (4.1 million barrels), ranking the disaster as the worst unintentional oil spill in history.
Slick Entrained in Loop Current, Approaches Florida Straits
A strong surface current in the eastern Gulf of Mexico, called the Loop Current, acts like an ocean conveyor belt, transporting water through the Gulf and out into the Atlantic Ocean. Many observers voiced concern that the oil spilling into the Gulf would move through the Florida Straits and into the Gulf Stream, potentially reaching the eastern seaboard from Florida to North Carolina and beyond.
On May 17, SkyTruth published a MODIS image (Figure 2) that showed the oil slick indeed was entrained in the Loop Current. Analysts at The Weather Channel concurred, as did NOAA’s administrator the following day. Images on subsequent days confirmed the analysis. On May 27, SkyTruth tentatively identified a thin sheen of oil entering the Florida Straits, but water-sampling data collected from the area couldn’t confirm that oil from the spill reached either the straits or the Gulf Stream.
Oil Slick Peaks, Spans Most of Northeastern Gulf
Throughout May and June, satellite images showed the oil slick generally increasing in size, although SkyTruth noted significant day-to-day fluctuations that were attributed to cleanup and containment activity, weather and sea-state conditions, as well as variability in imaging conditions that affected the ability of satellite images to reveal the full extent of oil slicks and sheen.
On June 25 and 26, a MODIS image (Figure 3) showed slicks and sheen ominously spread across more than 24,000 square miles of the northeastern Gulf of Mexico—an area the size of West Virginia—clearly affecting hundreds of miles of beaches and marshes from Louisiana to Florida. Later reports of tar balls coming ashore on Texas beaches added a fifth state to those experiencing direct impacts from the spill.
This was the peak size of the spill that SkyTruth could observe at the Gulf’s surface. In coming days, tropical storms Alex and Bonnie brought strong winds, large waves and heavy rain that helped dissipate the slicks.
Macondo Fully Capped
On July 15, all flow of oil and gas from the Macondo Well was stopped. Tropical Depression Bonnie tracked directly over the spill site on July 24, dispersing some of the oil slick. Satellite images on July 26 revealed widely scattered patches of oil. Oily sheen covered almost 11,000 square miles again on July 28, but subsequent visible-infrared and radar images indicated a progressive reduction of oil floating on the Gulf ’s surface during the next two weeks. This wasn’t unexpected; with no new oil leaking from the Macondo Well, oil slicks at the surface would be steadily diminished by evaporation, photolysis, biodegradation, natural dispersal and the efforts of cleanup crews.
On Aug. 5, the well was plugged from the top with cement. The final operation to permanently kill the well—injecting cement into the bottom via a relief well—occurred a few weeks later.
Cumulative Observed Surface Extent Larger than Oklahoma
Satellite observations of the surface oil slicks and sheen from April 25 through July 16 showed that, at one time or another, 68,000 square miles of Gulf waters were covered with oil—an area about the size of Oklahoma. That doesn’t include the distribution of hydrocarbons that couldn’t be seen suspended in the water column, dissolved in the water or driven beneath the surface by the application of chemical dispersants. Ocean currents at depth potentially transported that material in different directions than oil at the surface, making it likely that the total area of the Gulf that was directly impacted by oil and natural gas from this spill is larger than the cumulative oil-slick footprint determined from satellite image analysis (Figure 4).
Government Assessment of Oil Fate Overly Optimistic
On Aug. 4, NOAA and the U.S. Geological Survey (USGS) reported that a total of 205.8 million gallons (4.9 million barrels) of oil gushed from the Macondo Well over 85 days before it was completely shut off on July 15. Subtracting 33.6 million gallons (800,000 barrels) of oil that was kept out of the water by direct capture at the wellhead yields a total spill of 172.2 million gallons (4.1 million barrels) of oil directly into Gulf waters.
The NOAA/USGS report included a pie chart that suggested three-fourths of the spilled oil was dispersed or destroyed, leading White House Office of Energy and Climate Change Policy Director Carol Browner to conclude “more than three-quarters of the oil is gone. The vast majority of the oil is gone.”
How much of the oil, injected at high pressure into frigid water 5,000 feet deep and treated at the wellhead with chemical dispersants, never reached the surface? The NOAA/USGS report doesn’t directly address this question. The SkyTruth-MacDonald estimates suggest 44 percent of the oil that leaked from the well remained underwater (or was driven back underwater by dispersants), out of sight of satellite images and Coast Guard observers. Given a total spill of 172.2 million gallons, extrapolation from that first week leads to the conclusion that 76 million gallons may have lingered beneath the surface. NOAA is assuming rapid biodegradation of the dispersed and dissolved oil, which may be reasonable in relative terms, i.e., biodegradation in the warm shallows of the Gulf of Mexico progresses more quickly than in the frigid Arctic Ocean, for example. But with no data provided on the actual rates of biodegradation at various depths, there’s no way to determine confidently how much oil had naturally biodegraded by early August. At best, one can say that 25 percent of the total amount of oil released from the well has been accounted for by direct recovery from the wellhead and by burning and skimming at the surface. Evaporation, biodegradation and other natural processes are attacking the remainder, but at unknown rates and with unknown efficacy.
Conclusions and Recommendations
As SkyTruth and others demonstrated throughout the BP/Deepwater Horizon oil and gas disaster, satellite imagery is a valuable source of unique information in response to catastrophic spills and offers the ability to detect and assess the frequency and significance of smaller, chronic pollution events. Therefore, the federal government should institute a program of routine, comprehensive satellite image monitoring to detect and assess pollution problems wherever offshore drilling occurs or is being contemplated in both state and federal waters.
The cost for such a program, covering oil and gas infrastructure in the Gulf of Mexico, would likely fall in the range of $3 million to $4 million per year. The program could be funded by offshore operators, their insurers, government agencies and revenues collected from lease sales and petroleum production. The design, implementation and results of a monitoring program should be publicly transparent and should engage active participation by the academic and nongovernmental organization communities.
Publisher’s Note: This article is an adapted excerpt from “Impacts, Perception, and Policy Implications of the BP/Deepwater Horizon Oil and Gas Disaster,” Elliott Norse and John Amos, which appeared in the November 2010 issue of the Environmental Law Reporter’s (ELR’s) News & Analysis. Copyright © 2010 Environmental Law Institute, Washington, D.C. Reprinted with permission from ELR.