Spectroscopic remote sensing of the distribution and persistence of oil from the Deepwater Horizon spill in Barataria Bay marshes

REMOTE SENSING OF ENVIRONMENT

卷:129

Article

Kokaly, Raymond F.¹  Couvillion, Brady R.²  Holloway, JoAnn M.¹  Roberts, Dar A.³  Ustin, Susan L.⁴  Peterson, Seth H.³  Khanna, Shruti⁴  Piazza, Sarai C.² 

[1] US Geol Survey, Denver Fed Ctr, Denver, CO 80225 USA

[2] US Geol Survey, Livestock Show Off, Baton Rouge, LA 70803 USA

[3] Univ Calif Santa Barbara, Dept Geog, Santa Barbara, CA 93106 USA

[4] Univ Calif Davis, Dept Land Air & Water Resources, Davis, CA 95616 USA

关键词: Hyperspectral;    Ecosystem disturbance and response;    Imaging spectroscopy;    Hydrocarbon;    Oil spill;    Coastal wetlands;   

DOI  :  10.1016/j.rse.2012.10.028

来源: Elsevier

【 摘 要 】

We applied a spectroscopic analysis to Airborne Visible/InfraRed Imaging Spectrometer (AVIRIS) data collected from low and medium altitudes during and after the Deepwater Horizon oil spill to delineate the distribution of oil-damaged canopies in the marshes of Barataria Bay, Louisiana. Spectral feature analysis compared the AVIRIS data to reference spectra of oiled marsh by using absorption features centered near 1.7 and 23 mu m, which arise from C-H bonds in oil. AVIRIS-derived maps of oiled shorelines from the individual dates of July 31, September 14, and October 4, 2010, were 89.3%, 89.8%, and 90.6% accurate, respectively. A composite map at 3.5 m grid spacing, accumulated from the three dates, was 93.4% accurate in detecting oiled shorelines. The composite map had 100% accuracy for detecting damaged plant canopy in oiled areas that extended more than 1.2 m into the marsh. Spatial resampling of the AVIRIS data to 30 m reduced the accuracy to 73.6% overall. However, detection accuracy remained high for oiled canopies that extended more than 4 m into the marsh (23 of 28 field reference points with oil were detected). Spectral resampling of the 3.5 m AVIRIS data to Landsat Enhanced Thematic Mapper (ETM) spectral response greatly reduced the detection of oil spectral signatures. With spatial resampling of simulated Landsat ETM data to 30 m, oil signatures were not detected. Overall, similar to 40 km of coastline, marsh comprised mainly of Spartina alterniflora and Juncus roemerianus, were found to be oiled in narrow zones at the shorelines. Zones of oiled canopies reached on average 11 m into the marsh, with a maximum reach of 21 m. The field and airborne data showed that, in many areas, weathered oil persisted in the marsh from the first field survey. July 10, to the latest airborne survey. October 4, 2010. The results demonstrate the applicability of high spatial resolution imaging spectrometer data to identifying contaminants in the environment for use in evaluating ecosystem disturbance and response. Published by Elsevier Inc.

【 授权许可】

Free