Fingerprinting of Weathered Oil Residues from the Deepwater Horizon Oil Spill: The Importance of Multiple Lines of Investigation

On April 20, 2010, the Deepwater Horizon (DWH) offshore drilling rig experienced an uncontrollable blowout and catastrophic failure resulting in a subsurface oil release lasting 86 days. In response to this accident, BP and governmental agencies serving as natural resource damage trustees engaged in one of the largest environmental data gathering efforts in history. These collection efforts resulted in the need to distinguish oil samples associated with this release from other natural and anthropogenic petroleum sources. The interpretation of the hydrocarbon chemistry of sediment samples taken as part of the release of oil from the Mississippi Canyon (MC)252 well in the Gulf of Mexico (GOM) was confounded by the presence of naturally occurring hydrocarbons as well as other anthropogenic sources. Oil and gas production has been occurring in the shallow waters of the GOM for more than sixty years (Overton, 2004). Natural inputs of hydrocarbons include petroleum related hydrocarbons from petroleum seeps and biogenic hydrocarbons (e.g., plant wax alkanes, algal hydrocarbons, perylene, etc.), while anthropogenic sources include releases from other oil and gas platforms (often of similar chemical composition), combustion-related sources from shipping and land based sources, and runoff from terrestrial areas. Previous investigations of the hydrocarbon concentrations in GOM sediments before 2010 have shown total polycyclic aromatic hydrocarbon (TPAH) concentrations ranging over an order of magnitude, from less than 100 ppb to more than 1,000 ppb (Rowe and Kennicutt, 2009). The authors noted that the stations with the highest TPAH concentrations did not have a uniform PAH composition, indicative of the multiple sources in the study area. Additionally, the TPAH concentrations measured at a single station during multiple cruises varied up to 109%, suggesting that the measured numbers were highly spatially or temporally variable. The median concentration measured in the study was 100 ppb, although the authors acknowledge that it is about four-times lower than values reported in previous studies.

In this paper, a protocol has been developed to determine the presence or absence of MC252 oil in sediments of the offshore regions of the GOM. Chemical data from more than 4,000 offshore sediment samples, collected from dozens of separate studies during the DWH Response and Natural Resource Damage Assessment (NRDA) have been interpreted using this procedure. Analytical chemistry measurements, included PAHs and alkanes but specifically focused on the relatively degradation-resistant petroleum geochemical biomarkers combined with other lines of chemical and non-chemical information.

Much effort has focused on interpretation of petroleum chemistry data from oil samples, tar balls, and sediment so as to determine if MC252 oil is present or whether a sample is from a different source or a mix of petroleum sources (Stout and Payne, 2016). While the interpretation of the source or sources of oil and tar ball samples requires great care in analyzing data on PAHs, alkanes, biomarkers, etc., the fingerprinting task becomes far more complex when applied to the interpretation of data from sediment samples. These complexities arise from the changes in chemical compositions of MC252 oil that occur as a result of weathering processes before and after sedimentation and the multiplicity of potential sources of petroleum-related chemical compounds in sediments. Thus, a more complex evaluation has been used in the development of this methodology.

Several contributing factors complicate the fingerprinting of petroleum residues in these sediments. Sediments from all coastal and offshore areas, especially those in the “spill zone” in the GOM, are marked by important influences from the Mississippi River and natural sources inclusive of, but not limited to, natural oil seeps and as such contain a background mix of: 1) multiple petroleum or petrogenic sources; 2) multiple combustion-related (pyrogenic) sources; and 3) biogenic sources. Misattribution of background and petrogenic sediment PAHs to the spilled oil has, in other events (e.g., Exxon Valdez [Bohem et al., 1997]), created uncertainty and confusion until the proper approaches to interpreting the PAH data were discussed, applied, and published (Bence et al., 1996; Page et al., 1996).

Several detailed issues were considered in the development of the methodology presented here: 1) the natural and anthropogenic background of petroleum related chemicals in the GOM, including the PAHs, is considerable and associated with multiple sources; 2) chemical fingerprints of natural seepages or releases from other area oil production platforms may be very similar to chemical fingerprints from MC252 oil; and 3) extensive weathering via physical, chemical (e.g., evaporation and dissolution) processes and especially biodegradation (i.e., microbial degradation through utilization of oil as a microbial food source) of MC252 have taken place during water column and surface transport. The extensive use of dispersants during the release likely contributed to the rapid rates of weathering and dissolution, but is not expected to alter the chemical composition of the oil.

The method described below allows for the categorization of samples into one of five categories based on the chemical and supporting evidence. The categories are 1) MC252 oil, 2) probable MC252 oil, 3) indeterminate, 4) no evidence of MC252, and 5) below screening threshold. Classification in Category 1 or 2 indicates a sample which has likely been impacted by oil from the DWH oil spill. Category 3 samples contained petroleum hydrocarbons but source attribution could not be determined. Category 4 samples definitively indicated no evidence of petroleum from the DWH oil spill. Background level of hydrocarbons in deep-water sediments of the GOM, typically ranging up to more than 500 ppb (0.5 ppm) of TPAH, contain petroleum from natural seeps and other sources. These non-MC252 sources complicate any fingerprinting interpretations. Furthermore, samples below this TPAH level unlikely pose any incremental risk over background (Rowe and Kennicutt, 2009). Because of these factors and based on an initial review of the data, the presence of MC252 oil in those samples with TPAH concentrations of 500 ppb (0.5 ppm) or less could not be definitively determined nor could the specific source of mix of sources in samples. Therefore, those samples (TPAH < 500 ppb) were excluded from the detailed fingerprinting analysis and assigned to Category 5 (below screening threshold) with no source interpretation provided.

To determine the presence/absence of MC252 oil in a sediment sample, multiple lines of chemical and other evidence are needed. Due to background sources, the weathering of MC252 oil, and other factors, no one chemical parameter or analytical approach suffices to definitively determine the source of petroleum hydrocarbons in a sediment sample. To control for analytical variability and obtain standard references for comparative purposes, an MC252 source oil control oil should be run with each analytical batch. When batch control oil analyses were not available, laboratory-specific average concentrations for MC252 were used, but results were interpreted with caution. The chemical parameters required for source interpretation are listed in Table 1. To aid in interpretation of the chemical data, other non-chemical lines of evidence were also reviewed in conjunction with the chemical data. Examples of these types of information are listed in Table 2, but the list is not exhaustive. Full analyte lists and analytical methods can be found in DHW analytical quality assurance project plans (NOAA, 2014 and previous versions). Data review and interpretation

A detailed review of chemical and associated data from more than 4,000 offshore sediment samples was used to assign one of the 5 interpretation categories to each sample. Logical pathways for assignment to each of these categories are illustrated in the schematic in Figure 1 and detailed specific criteria for each category are presented in Table 3.

For all samples with TPAH concentrations above the fingerprinting threshold, the three primary chemical lines of evidence (petroleum biomarkers, alkanes [including gas chromatographs (GC) using flame ionization detector (FID)], and PAH; Table 1) were examined and compared to an MC252 oil standard run with the same batch. Consideration was given that as oil remained in the environment, weathering occurred which altered the profile(s) compared to the initial oil released. For example, MC252 oil chromatograms showing different stages of weathering are provided in Figure 2. These chromatograms show a suite of both alkane and an unresolved complex mixture (UCM) within the expected boiling range of an MC252 oil source, as determined from patterns of degradation noted in surface oils and tar balls after the spill. As MC252 oil weathers, although components such as the n-alkanes degrade and ratios of individual components may change, the shape and boiling range of the UCM are characteristic of an MC252 source.

Source determinations made on the basis of the chemical evidence were confirmed with additional lines of evidence including the consideration of known seep locations and non-MC252 oil and gas infrastructure. Seep presence was additionally considered through the use of depth profiles of PAH concentration (Figure 3, when available). Sediment sample depth within a sediment core was also considered. Samples which showed characteristic surface or near-surface maximums (i.e., upper 1–3 cm of the sediment column) of the deposition of oil from the water column were considered more likely to contain MC252, while samples with no concentration gradient over core depth were considered likely to be seep-related or unimpacted depending on concentration.

One of the most striking features of the biomarker data from the DWH oil spill was that biodegradation of biomarkers was rapid relative to other oil spills (e.g., Exxon Valdez) and followed a predictable pattern. As a result, the biomarker subset used for the regression analysis needed to be refined to focus on those biomarkers most resistant to weathering. Examination of the MC252 data set has revealed a distinct and predictable weathering pattern of the C27 sterane biomarkers. The initially affected biomarkers include the C27 diacholestanes and cholestanes (13b(H),17a(H)-20S-diacholestane, 13b(H),17a(H)-20R-diacholestane, 14b(H),17b(H)-20R-cholestane, and 14b(H),17b(H)-20S-cholestane). This degradation pattern is accompanied by an alkane profile that exhibits a high-molecular-weight (i.e., waxy) alkane pattern, as all of the lighter compounds have been lost. Coefficient of determination (R²) was used to assess similarity of the biomarker profile in the sediment samples as compared to the control oils (Saba and Boehm, 2011). The sterane weathering can decrease the R² between the sample and control biomarkers, so similarity between samples and control oils was evaluated both with and without these compounds. In the most heavily weathered samples, this list was refined to exclude several additional biomarkers that were predictably impacted by more extensive degradation.

An example of a match of the biomarkers (R² >0.95), confirming the dominance of the MC252 biomarkers in an unweathered sample, is shown in Figure 4; a non-match (R² < 0.8) is shown in Figure 4B. Two examples of weathered matches, where the steranes show depletion, are shown in Figure 4C and D. In 4C, the interferent C30-moretane (a possible indicator of terrestrial organic material [French et al., 2012]) is also present. Pending confirmation from other lines of evidence, these biomarker comparisons indicate the probable presence of MC252 oil which has been significantly biodegraded.

The interpretation category of each sample location was used to map the overall extent of MC252 oil in the offshore sediments (Figure 5). For locations where more than one depth was fingerprinted, the lowest (most likely to be MC252) category was plotted. The distribution of Category 1 and 2 samples showed that the majority of these samples were limited to an area near the wellhead, within approximately 20 nmi with most of these samples, located toward the southwest from the wellhead, consistent with the trajectory of subsurface “plume” (Camilli et al., 2010; Boehm et al., 2016).

In addition to providing information regarding the extent of the transport of MC252 oil, fingerprinting provides some information regarding the transport mechanisms which delivered the oil to the sediment. Three distinct chemical signatures suggestive of the sedimentation pathway were identified. In the first, drilling muds and sediments from the failed attempt to “top kill” the well (the process of injecting drilling mud into the well to stabilize the blowout) combined with MC252 oil had settled out of the water column in areas adjacent to the wellhead. In the second, oil reaching the sea surface was subjected to in situ burning, which in turn produced residues with altered chemistry some of which were deposited on the seafloor in discrete deposits. The final and primary mechanism of transport was the diffuse deposition of the oil from the anomalous layer of elevated dissolved and droplet oil that moved to the southwest (also known as “the plume”; see Camilli et al., 2010) and became incorporated with settling marine snow, which resulted in patchy oil remnants in the sediments to the southwest of the wellhead. These particular residues often showed signs of biomarker degradation and heavy waxy alkane compounds in the chemical compositions. The typical, often highly weathered profile seen in the majority of sediments was attributed to the marine snow pathway, while the drilling mud impacted sediments were limited to the area proximate to the wellhead.

Drilling mud used in the “top kill” procedure was a synthetic-based mud (SBM) that contained compounds called “olefins.” The presence of olefins is evident as a unique cluster of peaks in the GC chromatogram (see Figure 6). Though olefin presence alone did not indicate an MC252 source, as their presence may be related to other local drilling activity, olefin presence added a line of evidence to the confirmation of MC252 related oil in Category 1 and 2 samples proximal to the wellhead and suggested the mechanism by which this oil was deposited; olefin absence did not provide lack of confirmation because MC252 may have reached the sediment by mechanisms other than its association with SBM transport. Select sediment samples within 2 miles of the wellhead were found to contain olefins.

Controlled in situ burns were performed during the course of the DWH oil spill to remove oil from the water surface. A total of 411 controlled burns were performed with the volume of oil burned estimated between 9.3 and 13.1 million gallons (Ross, 2010). In situ burn crews reported visual observations of sinking post-burn residues. Chemical fingerprints of Category 1 and 2 sediment samples were further evaluated to determine the potential presence of burn residue in these samples. Identification of a sample as potential burn residue did not change the interpretation category but provided information about the likely spatial impact of a given sample. Photographic evidence of burn residue samples suggests these discrete tar-ball-like samples would have limited spatial impact regarding chemical concentrations.

Several chemical indicators were examined to determine whether a sample contained burn residue. Identification of burn residues is complicated in sediment samples by background contributions of pyrogenic PAHs and thus requires more detailed evaluation of all available lines of evidence. The presence of the following indicated a burn residue: Presence of select heavy pyrogenic PAHs which are absent in fresh MC252, including benzo(k)fluoranthene, benzo(a)pyrene, and indeno(1,2,3-cd)pyrene, enhanced concentrations of benzo(g,h,i)perylene relative to dibenzo(a,h)anthracene, and a chromatographic UCM shaped similar to a scalene triangle (Figure 7). Some of these characteristics were also noted in burn residues by other researchers (Stout and Payne, 2016; Shigenaka et al., 2015). Diagnostic ratios including BKF/BBF, BAP/BEP, and INDP/Hopane were used to confirm the categorization. Very few sediment samples were found to contain evidence of burn residues, which were more commonly found as oil/tar ball samples targeted during ROV transects.

Overall, the method described in this paper allowed for the identification of MC252-containing samples in offshore sediments in the GOM despite significant weathering which affected not only the more easily degraded components of the oil but some of the more recalcitrant steranes. This multiple-lines-of-evidence approach allowed for the inclusion of these highly weathered samples, while also excluding very similar oils which occur in the GOM. While the general approach to identifying MC252 oil presented in this paper is applicable to other data sets (e.g., shoreline samples), the use of this method requires a detailed understanding of the processes which may be acting on the oil as it is transported. Specific tolerances and criteria require refinements appropriate to the target environment and degree of weathering. For example, shoreline samples from the Gulf of Mexico would be expected to contain oil which has been significantly weathered, and the presence of fresh oil on the shoreline would not be consistent with an MC252 source. As the understanding of the fate and transport of the oil increases, the ability to identify residues and develop a complete picture of the extent of oiling will improve, and the extent of MC252 oiling can be further refined. The general framework presented here can be further applied to non-MC252 spills, but requires modifications based on an understanding of the oil and the environment in which the spill occurred.

This research was supported by BP Exploration and Production Co. We thank the many researchers from multiple consulting firms, laboratories, agencies, and institutions who participated in these sample collections and analyses.