Brevia

Magnitude of the 2010 Gulf of Mexico Oil Leak

See allHide authors and affiliations

Science  29 Oct 2010:
Vol. 330, Issue 6004, pp. 634
DOI: 10.1126/science.1195840

Abstract

To fully understand the environmental and ecological impacts of the Deepwater Horizon disaster, an accurate estimate of the total oil released is required. We used optical plume velocimetry to estimate the velocity of fluids issuing from the damaged well both before and after the collapsed riser pipe was removed. We then calculated the volumetric flow rate under a range of assumptions. With a liquid oil fraction of 0.4, we estimated that the average flow rate from 22 April 2010 to 3 June 2010 was 5.6 × 104 ± 21% barrels/day (1.0 × 10−1 meter3/second), excluding secondary leaks. After the riser was removed, the flow was 6.8 × 104 ± 19% barrels/day (1.2 × 10−1 meters3/second). Taking into account the oil collected at the seafloor, this suggests that 4.4 × 106 ± 20% barrels of oil (7.0 × 105 meters3) was released into the ocean.

The Deepwater Horizon well was sealed on 15 July 2010 after flowing oil at the seafloor for approximately 84 days. A full accounting of the oil released will be required in order to fully understand the environmental and ecological impacts of this disaster. One method for determining this volume is to measure the flow at the discharge sites and integrate these measurements over time. We used optical plume velocimetry (OPV) to estimate the mean velocity of fluids issuing from the well with videos from before and after the removal of the collapsed riser pipe from the blowout preventer. We analyzed two short (20 to 30 s) high-resolution video sequences (1) representing each regime. We focused our analyses on flow near the nozzle, where momentum forces are dominant and velocities scale linearly with the average nozzle rate (2). In this part of the flow, differences between the image velocity and the average flow rate at the nozzle can be accounted for with a constant “shear layer” correction factor (2), and the median image velocity can be used to determine flow rate.

In the study of fluid dynamics, spatial cross-correlation methods (e.g., particle image velocimetry) are often used to calculate the image velocity field. However, such methods can yield velocities that are significantly lower than expected with this kind of flow (2). OPV was developed for measuring flow rates in seafloor hydrothermal systems (3) and uses temporal instead of spatial cross-correlation. Interpolated temporal cross-correlation functions of image intensity are calculated across the entire region of interest for pixel pairs separated by some distance horizontally and vertically. The distance is chosen on the basis of the direction of flow, the frame rate, and the resolution of the imagery. To increase accuracy, the separation is then refined so that the distance is maximized while ensuring that the signals still correlate (2). The lag value corresponding to the maximum of each cross-correlation function defines the time (in frames) required for flow features to traverse the distance defined by the pixel separation. In this way, a velocity (in pixels/frame) can be calculated at every pixel within the region of interest (Fig. 1A).

Fig. 1

(A) Colored contours of image velocity magnitude in pixels/frame. Areas of the image not hosting flow or containing values outside the measurable range are masked. The dashed boxes indicate the two areas used in our analyses. (B) Contours of volumetric flow rate in barrels/day (m3/s) over a range of image velocities and effective flow diameters. Values associated with the lighter and darker flows are indicated with gray dots. The value calculated for the combined flow is shown with a black dot. Vertical error bars indicate the 12% combined uncertainty associated with the image resolution, the median image velocity, and the shear layer correction factor, and horizontal error bars indicate the 3% uncertainty associated with the area estimate (6). For a more detailed version, see fig. S3.

After the riser was removed, the flow was separated into two flows, one lighter and one darker in color (Fig. 1A and fig. S1), which we analyzed separately. Figure S2, A and B, shows the distribution of the image velocity magnitudes for the lighter and darker colored flows, which have median values of 13.6 and 9.43 pixels per frame, respectively. Assuming the smaller lighter-colored flow comprised 10% of the cross-sectional area of the riser pipe (4), we calculated an effective image velocity magnitude of 9.85 pixels/frame for the entire flow.

With a spatial resolution of 3.85 pixels/cm (fig. S1), a video frame rate of 30 frames/s, and a shear layer correction factor of 2.10 (5), we contoured the volumetric flow rate from this leak over a range of image velocities and effective flow diameters (Fig. 1B and fig. S3). Assuming the fraction of liquid oil in this fluid was 0.4 (6), we estimated the average flow rate after riser removal to be 9.3 × 103 and 5.8 × 104 barrels/day (1.7 × 10−2 and 1.1 × 10−1 m3/s) for the lighter and darker flows, respectively, or 6.8 × 104 barrels/day (1.2 × 10−1 m3/s) total. Our analysis from before riser removal (figs. S4 to S7) yielded a flow rate of 5.6 × 104 barrels/day (1.0 × 10−1 m3/s). Because leaks at the kink above the blowout preventer were not included, this total is likely an underestimate. Thus, we cannot say with certainty that flow rates increased after riser removal.

Aside from temporal variability, we estimated the total combined uncertainty to be about 19 and 21% for the flows before and after riser removal (6). It was not possible to quantify temporal variability with the available data; however, changes were likely. Temporal variability may have been induced by variations in the gas content of the fluid, changes in the well integrity or abrasion of components within the system, changes in the formation pressure, or changes in the size and number of openings in the riser wall at the kink. Assuming a constant flow rate and subtracting the 804,877 barrels of oil (127,965 m3) collected at the seafloor (7), we estimated that the total oil released from the Deepwater Horizon leak was 4.4 × 106 ± 20% barrels (7.0 × 105 m3). This estimate may be refined if additional video allows the temporal variability to be assessed or the flow from the secondary leaks to be added. Despite the uncertainties, it is clear that this oil release exceeds the Exxon Valdez spill by about an order of magnitude, with flow rates at least one order of magnitude higher than initially reported.

Supporting Online Material

References and Notes

  1. High-resolution video data was provided by the office of Senator Bill Nelson and by the Senate Committee on Environment and Public Works.
  2. We analyzed original data from the laboratory experiments to obtain this correction factor.
  3. A full discussion of all uncertainties is available as supporting material on Science Online.
  4. This research was supported by the Lamont–Doherty Earth Observatory and by NSF under grants 0623285 and 0917955.
View Abstract

Navigate This Article