Response to Heavy, Non-Floating Oil Spilled in a Great Lakes River Environment: A Multiple-Lines-Of-Evidence Approach for Submerged Oil Assessment and Recovery

On July 25, 2010, during a rainstorm and flood a rupture of a 30” Enbridge Energy (Enbridge) oil pipeline (Line 6B) released a heavy crude oil composed of bitumen diluted with natural gas condensate (dilbit) into a wetland and adjacent river system in a Great Lakes watershed near Marshall, Michigan. The Enbridge Line 6B pipeline is part of the 1,900-mile Lakehead system that transports heavy crude oil as well as other oil products and natural gas products from production fields in Western Canada and Western U.S. to refineries in the Upper Midwest and Ontario, Canada. Line 5 and Line 6 intersect watersheds of three of the Great Lakes – Superior, Michigan, and Huron. With increased production in these regions the volume of oil transported through pipelines and via railcar adjacent to water bodies in the Great Lakes region has increased dramatically.

After Line 6B ruptured, oil spilled for over 17 hours during a product change from Western Canadian Select to Cold Lake Blend. Enbridge has reported that approximately 843,000 gallons of oil were discharged from the pipeline into the nearby Talmadge Creek and Kalamazoo River (Fig. 1). This discharge represents one of the largest inland oil spills into a freshwater system in North American history.

Flood flows with an annual exceedance probability of 4 percent (Hoard et al., 2010) rapidly distributed the oil along 2 miles of Talmadge Creek and adjacent wetlands and downstream into the Kalamazoo River, eventually reaching 38 miles of the river's channels, as well as associated backwaters, floodplains, islands, and wetlands (Fig. 1). During transport the lighter diluent volatilized and a portion of the oil submerged beneath the water surface, presumably by adhering to and mixing with sediment and organic matter. The submerged oil was subject to further transport along with the remaining surface oil until eventually settling into quiescent areas of the Kalamazoo River. As floodwaters receded, drapes of oil of various thicknesses also covered floodplain, wetland, and island surfaces. Some of the oil passed through two impoundments and dams at the Village of Ceresco and City of Battle Creek, and a significant fraction eventually ended up in a third impounded area at the delta of Morrow Lake, approximately 36.5 miles downstream of the release location and approximately 70 miles upstream of the mouth of the Kalamazoo River at Lake Michigan. Based on rough estimates of time-of-travel of water releases between the power plant at Marshall and the USGS stream gage on the Kalamazoo River at Battle Creek, the initial leading edge of the oil probably took about 30 hours to be transported to Morrow Lake, travelling at about 1.25 miles/hour.

Personnel and contractors of the United States Environmental Protection Agency (U.S. EPA) responded to the Line 6B spill on July 26, 2010, the date that it was reported (Fig. 2). The U.S. EPA issued a Removal Administrative Order (Order) to Enbridge on July 27, 2010. The Order was issued pursuant to the authority vested in the President of the United States by Section 311(c) of the Federal Water Pollution Control Act §1321(c), as amended, and commonly known as the Clean Water Act (CWA). The amended CWA also includes the Oil Pollution Act of 1990, 33 U.S.C. §2701 et seq. The response efforts were managed utilizing the Incident Command System (ICS) under a Unified Command Structure with Incident Commanders from the U.S. EPA, Michigan Department of Environmental Quality (MDEQ), and Enbridge among others coordinating on response, assessment, and monitoring activities.

The physical setting of the Kalamazoo River—meandering, low gradient, diverse channel and floodplain features, extensive floodplain forests and wetlands, off-channel water bodies, and impoundments—is typical of many medium- to large-sized rivers in the Great Lakes region and contributed to the lengthiness and complexity of response operations. The average gradient of the Kalamazoo River in the spill-affected reach is 3.14 feet/mile or about 0.06 percent.

The objectives of this paper are to describe and discuss how conventional recovery techniques and a multiple-lines-of-evidence approach were used in the assessment and recovery of dilbit in a freshwater riverine ecosystem within the Great Lakes Region. The multiple-lines-of evidence approach included six science-based tools used specifically for assessment and recovery of submerged oil.

Pursuant to U.S. EPA direction, over 1.1 million gallons of Line 6B oil were recovered relatively early in the response by aggressive use of conventional oil spill containment and recovery strategies and tactics. This volume exceeds Enbridge's initial estimate of the total oil released from the pipeline. Nearly 2,500 responders were organized under an ICS with operations being driven by comprehensive Shoreline Cleanup and Assessment Technique (SCAT) activities developed and directed by U.S. EPA (Dollhopf and Durno, 2011).

Because much of the oil remained floating in the days immediately following the release, conventional methods and equipment were effective in containment and recovery of this floating component. Over 12,000 linear feet of conventional surface containment boom and over 8,000 feet of absorbent boom were deployed to contain and aid recovery of the floating oil within the first week following the release (Fig. 3A). During this time, however it became apparent that these would not work for the sinking component, leading to application of alternative tactics like agitation (Fig. 3C).

Containment barriers were constructed within the source area. Vacuum trucks and excavation equipment were utilized to remove pooled oil and oil-saturated soils. A series of earthen dams were constructed within the forested wetland between the pipeline break location and Talmadge Creek to contain and stop the flow of oil into Talmadge Creek. In addition, Enbridge constructed a flume system consisting of a series of inverted weirs within Talmadge Creek to trap and contain oil (Fig. 3B). Eventually, the entire 2.25 miles of the creek overbank areas were excavated to recover oil saturated soils.

To collect and remove the boom-contained oil in Talmadge Creek and the Kalamazoo River, Enbridge utilized conventional equipment including vacuum trucks, drum skimmers, and absorbent materials. Collected oil and oil/water mixtures were transported to a central on-site storage area and placed into large portable tanks for consolidation, separation, and temporary storage.

Over the next few weeks, containment and collection locations were added along 38 miles of the affected Kalamazoo River. In total, approximately 35 surface containment locations were established along the Kalamazoo River, and over 175,000 linear feet of containment and absorbent boom were deployed during the first month of the response.

Because of the flood at the time of the spill, oil was deposited on the banks above the normal river stage. These “overbank” areas were assessed for oil impact and cleanup utilizing the SCAT process by separating the shoreline and overbanks along both sides of the Kalamazoo River into one-quarter mile segments. Recovery actions for overbank areas included: direct pooled oil removal; pooled oil removal through low-pressure/high volume flushing with collection via snare boom or absorbent boom; stained vegetation removal leaving root systems intact; manual removal of oiled debris; cutting, bagging, and removal of low hanging oil-impacted limbs; and scrape/removal of oil impacted soils. SCAT teams also re-inspected areas to determine the effectiveness of the initial recovery. Oil was also recovered by excavation of oil-saturated soils, resulting in the recovery of an additional estimated 300,000 gallons of released oil.

Even as conventional containment/recovery strategies were being implemented, U.S. EPA personnel began to evaluate the extent to which some of the oil had become submerged within the Kalamazoo River system. A Submerged Oil Task Force (SOTF) was created in late August 2010 to perform field assessments, characterization, and mapping of submerged oil (Fig. 2). Enbridge was directed to begin installing submerged oil containment structures at numerous locations within the Kalamazoo River. These structures included gabion baskets filled with absorbent snare booms installed on the bottom of the river and silt curtains extending from the river bed to the water's surface. In addition, a sediment basin was constructed within Talmadge Creek to enhance sediment and submerged oil deposition upstream of its confluence with the Kalamazoo River.

Initial SOTF assessment efforts in 2010 consisted of mapping the geomorphic settings throughout the river, followed by collection, visual assessment, and analysis of sediment core samples from representative depositional areas, and qualitative assessment of sheening and globs from submerged oil through manual agitation (poling) of the sediments. These activities resulted in the identification of 18 priority locations in the Kalamazoo River for further evaluation and implementation of submerged oil recovery actions.

In the fall 2010, ecological assessments at the 18 priority location preceded submerged oil recovery actions. Three levels of recovery were recommended based on results from the ecological assessments: (1) aggressive agitation techniques to liberate the submerged oil from the sediment (raking, flushing, aggressive aeration, and skimming for locations with limited ecological value) (Fig. 3C), (2) less aggressive agitation to limit damage to flora and fauna (cautious raking and flushing in areas deemed to have high ecological value), or (3) dredging within the Ceresco dam impoundment.

By 2011, submerged oil had proved much more challenging to recover than floating and overbank oils and increasingly became the focus of cleanup efforts (Fig. 2) (U.S. Environmental Protection Agency, 2011). It was clear that there was no single technology that would be operationally effective. During the summer of 2011, submerged oil recovery efforts focused on various manual and mechanical sediment agitation techniques to liberate the oil from the sediment, followed by conventional recovery of the oil from the water surface. Although these techniques were successful in recovering Line 6B oil, the results of the Late Summer 2011 Reassessment poling led the U.S. EPA to question the continued use of agitation as a recovery tool. The Federal On-Scene Coordinators (FOSCs) tasked U.S. EPA's Scientific Support Coordinators (SSCs) and a Scientific Support Coordination Group (SSCG) led by the SSCs to devise a multiple-lines-of-evidence strategy to inform submerged oil characterization, recovery strategies and endpoint determinations. Some of the components of this strategy were developed by operations and situation personnel (geomorphological mapping, poling assessments, and sheen tracking) and some were driven by scientific support staff (hydrodynamic and sediment transport modeling, forensic oil chemistry, and Net Environmental Benefit Analysis (NEBA)).

An understanding of the geomorphic behavior and characteristics in the Kalamazoo River provided the backbone for submerged oil assessment and recovery activities throughout the response. The physical behavior of the submerged oil was not well understood, but initial indications were that it accumulated in depositional areas of the river and had an affinity for aggregation with fine-grained soft sediment (silt, clay and organic matter accumulations with high water content) (Fig. 4). Slow-moving areas with submerged oil included reaches where the slope of the river flattened, such as at the three impoundments, or where the river widened enough to allow for depositional areas along channel margins. Submerged oil also was associated with secondary channels, oxbows, the downstream side of islands, and tributary mouths.

In spring 2011, delineation of geomorphic surface units (GSUs) for the oil-affected reach of the Kalamazoo River was done in a Geographic Information System (GIS) based on the synthesis of multiple data sources (Enbridge Energy, L.P., 2012a) (Fig. 5). Channel longitudinal profile and slope data were collected in the summer and fall 2010 (Tetra Tech, Inc., 2011). Fluvial landforms, anthropogenic features, bank lines, and channel widths were interpreted from April 2011 leaf-off aerial photographs. Particle size of river sediment was visually assessed during the Spring 2011 Reassessment poling and core collection and logging and was grouped into eight categories – gravel and larger, sand and gravel, sand, sand and silt, sand over silt, silt over sand, soft sediment, and organic. Water depths from the Spring 2011 Reassessment poling were used as a final refinement of the GSUs. The resultant 28 geomorphology-based categories were used to delineate areas of the river channel that were prone to either erosion or deposition. This technique of mapping river geomorphic settings and looking for submerged oil in depositional areas became an important tool to locate remaining problem areas after the floating oil was removed.

Sediment poling became a primary submerged oil assessment tool. The sediment was agitated using a graduated aluminum pole with an 8-inch diameter metal disc on the submerged end. If submerged oil was present in the sediment, the agitation action liberated oil from the sediment, allowing it to float to the water surface. The percent coverage of oil sheen and number of globs at the water's surface within one square yard were observed and categorized as ‘none’, ‘light’, ‘moderate’, or ‘heavy’ according to the Field Observation Submerged Oil Flowchart (Fig. 6).

During the initial qualitative assessment activities conducted in 2010, over 4,000 poling points were assessed throughout the affected river system. Global positioning system (GPS) coordinates and field observations were recorded upon poling and managed in a GIS. When moderate or heavy submerged oil observations were made, points were delineated with additional poling points made away from the initial point in all directions (step-outs) until the moderate/heavy area was delineated with observations of light and/or no sheening.

In spring 2011, teams conducted a comprehensive reassessment of the entire 38 miles of affected Kalamazoo River. The Spring 2011 Reassessment poling resulted in the collection of over 6,000 poling points and identification of over 200 acres in the heavy and moderate submerged oil poling categories, including 90 acres within the Morrow Lake delta and fan. Subsequent to the submerged oil agitation recovery actions conducted in 2011, a Late Summer 2011 Reassessment poling was conducted to evaluate the effectiveness of the agitation recovery efforts and to monitor the submerged oil accumulation footprints throughout the system. Reassessment poling results indicated that acreage of moderate and heavy oil categories had decreased but acreage with light categories had increased.

During the poling efforts conducted throughout 2011, crews observed an important correlation between water and sediment temperature and the relative liberation of submerged oil from the sediments. A Temperature Effects Study was conducted to determine the minimum water and sediment temperature above which poling activities could be effectively conducted. Based on this study results, the FOSC made a decision that 60oF was the minimum river water temperature for poling assessments to be conducted in the Kalamazoo River, which was a compromise between higher temperatures leading to increased sheening from agitated sediment and lower temperatures allowing for a longer field season (start and end date) for assessments to be conducted.

Poling reassessments were again conducted in spring 2012, late summer 2012, and spring 2013 (Fig. 2). These reassessments were all conducted above the minimum water and sediment temperature threshold of 60o F. Poling data from these reassessment activities have been valuable in tracking and mapping the submerged oil footprint over time, and have assisted the FOSC in directing Enbridge to conduct submerged oil recovery activities via dredging within the three impounded areas of the affected Kalamazoo River system in 2013–14.

The persistence of submerged oil resulted in continuing observation of surface oil globules and sheen for over three years after the spill, often in response to localized sediment agitation associated with boat traffic. In addition, oil globules and sheen appeared “spontaneously” (i.e., in the absence of any obvious or direct action). U.S. EPA required Enbridge to contain and collect surface sheen when it appeared. Soon after the sheen oil recovery activity began, U.S. EPA initiated the tabulation of sheen observations and collection responses, including dates, times, and locations of observations and recovery activities. Petroleum-based sheens were differentiated from biogenic sheens mainly in the field by use of a standard operating procedure for a tiered, four-part testing methodology (Enbridge Energy, L.P., 2012b).

When sheen location data were mapped, it became readily apparent that there was visual correlation among problematic sheening locations, depositional geomorphic settings, and heavy/moderate poling results (Fig. 7). These three independent lines of evidence converged on a picture of submerged oil being associated with deposits of fine-grained or soft organic-rich sediment. In addition, forensic oil chemistry results confirmed that Line 6B oil was the source of the vast majority of surface expressions of oil globs and sheen in the river (greater than 95% of sheen samples collected were confirmed as derived from Line 6B oil, see Forensic Oil Chemistry section below).

In 2011, hydrodynamic modeling became an integral part of operations to help answer questions about the fate and transport of remaining submerged oil in the Kalamazoo River and whether the oil could migrate out of the Morrow Lake delta and past Morrow Dam. The modeling served an important purpose of being able to extend the range of flow conditions that had been observed in the time since the spill.

A set of hydrodynamic and sediment transport models using the 2-dimensional Environmental Fluid Dynamics Code (EFDC) was developed by Tetra Tech for Enbridge in 2011–12 to simulate river water levels, flows, velocities, shear stresses, sediment loads, and erosion and deposition rates along the 38 affected miles of the Kalamazoo River (Enbridge Energy, L.P., 2012a). An important assumption of the models was that the physical properties of clay and silt-sized fine-grained sediment could be used as a surrogate for submerged oil and oiled sediment because the remaining submerged oil was found associated with fine-grained soft sediment deposits in slow-moving depositional areas of the river.

The Enbridge EFDC models were later updated and expanded by U.S. EPA in 2013–14 to help answer continued questions about the migration potential of the submerged oil and to evaluate new recovery and containment strategies. In addition, ongoing field and laboratory flume studies helped to identify the physical properties of oiled sediment and to describe the formation and breakup of oil-mineral aggregates in a freshwater river system.

After the Line 6B oil submerged and became distributed throughout the river system, it became apparent that forensic oil chemistry methods would be needed to determine whether it was possible to distinguish Line 6B oil from other residual background hydrocarbons (RBHs) that might be present in the Kalamazoo River sediments. Nearby urban areas likely contributed RBHs from nonpoint sources as well as occasional dumping and small spills. The objective of the work was to determine whether it would be possible to identify Line 6B oil in environmental samples with a high degree of certainty because of high background levels of other sources of oil contaminants. Based on this work, it was shown that Line 6B oil was enriched in certain triaromatic stearene biomarkers (Fig. 8), that could be used to differentiate the spilled oil from background oil sources (G. Douglas, in prep). These results were used to demonstrate definitively whether sheen/globule samples contained Line 6B oil. Environmental samples that contained Kalamazoo River sediment were more challenging to evaluate. Nevertheless, methods were developed to identify Line 6B oil in sediment samples that contained varying levels of RBH (G. Douglas, NewFields, written commun., 2014).

A NEBA was developed by the SSCG in 2012, using the Efroymson et al. (2003) application for marine environments and Rayburn et al. (2004) application for oil spill planning in the Great Lakes as guides (Bejarano et al., 2012). This provided another tool to help in decision-making and planning for remaining areas of submerged oil in the Kalamazoo River after human health and safety factors were accounted for. The NEBA offered a means for the FOSC and operations staff to weigh the ecological risks associated with leaving the residual submerged oil in place and allowing for potential natural attenuation, or removing the oil with selected recovery actions.

The NEBA conceptual design resulted in relative risk matrices for eight recovery actions (monitored natural attenuation, enhanced deposition in designated sediment traps, agitation toolbox techniques, dredging/vacuum truck, dewater/excavate, sweep/push, scraping, and sheen collection) that encompassed eight habitat types (impounded waters and deltas, flowing channels, depositional backwaters, bars, emergent wetlands, islands, oxbows and meander cutoffs, and forested scrub-shrub wetlands), and six ecological resource categories (plants, mammals, birds, amphibians/reptiles, fish, and invertebrates). Risks of exposure via five pathways (aqueous exposure, sediment exposure, physical trauma, physical oiling/smothering, and indirect) were taken into account in terms of magnitude of impact and length of recovery.

The final color-coded matrix of relative risk rankings ranged from “low impact” with an estimated level of resource impact of less than 10% relative to baseline or reference and less than 1 year for recovery to “very high impact” with an estimated resource impact of greater than 60% and greater than 7 years for recovery. The rankings were based on the current knowledge of the degree of oiling starting in the fall of 2011, after two seasons of intensive recovery actions.

The NEBA was integrated with approximately 200 individual submerged oil tactical areas in 2012. The integration was repeated as new poling results and tactical areas were updated by operations through spring 2013. Monitored natural attenuation and sheen collection were recommended for most of tactical areas, many of which were small in extent, but there were some important exceptions:

• A number of depositional areas along the river were observed to trap submerged oil after initial recovery operations had removed the oil, and were therefore designated as “sediment traps” that would be expected to need repeated assessment and recovery.

• For these designated sediment traps, the recommendation was to follow the sediment trap monitoring and maintenance plan. It was assumed that sediment traps would require repeated active submerged oil recovery, possibly every 6 months or after a major flood.

• If the number of moderate and heavy poling results stayed the same or increased in a tactical area then the recommendation was to increase monitoring frequency and continue to evaluate for possible future recovery.

• Agitation toolbox techniques were not recommended for recovery. Dredging was the only recovery tactic recommended.

By December 2010, about 5 months after the spill, most floating oil and oiled vegetation and debris had been recovered and a plan for overbank oil-saturated soil recovery (excavation) was in place that would continue into 2011. In addition, application of fluvial geomorphological science aided by a comprehensive GIS data base was beginning to inform submerged oil recovery strategies. The development of poling techniques, and GIS mapping of that information were designed to validate the geomorphological science and to provide greater resolution of submerged oil depositional patterns so that recovery priorities could be established and ecological impacts could be better understood. These were the first steps in developing a SCAT-like process for submerged oil.

The comprehensive mapping of submerged oil provided a baseline to determine submerged oil distribution throughout the system which could then be compared with periodic poling events in the future as well as special event-related poling events (e.g., post flooding or post recovery) to evaluate oil movement or recovery effectiveness. A third SCAT-like effort was necessary to “integrate” the NEBA relative risk matrix with tactical areas of the river containing moderate and heavy submerged oil. This step was strategically critical to optimize recovery and maintain sensitivity to the potential negative environmental effects of recovery. The integration resulted in a systematic recovery approach for these remaining tactical areas for the FOSC's consideration.

Additional evidentiary convergence on recovery targets was provided by tracking sheen observations and recovery, which demonstrated that the incidence and intensity of sheening corresponded to depositional areas of the river with moderate and heavy poling results. Finally, forensic chemistry analyses of oiled sediment samples also confirmed that poling and geomorphological prediction could indicate where submerged Line 6B oil was concentrated, thereby solidifying confidence in the SCAT process.

Continuing efforts to remove submerged oil from the system included dredging of oiled sediments from the impoundments: Ceresco, Mill Ponds, Morrow Lake delta, and from smaller designated sediment traps throughout the system. The tools described in this paper that informed decision making on current strategy and endpoints may be employed again for any areas that prove persistently problematic going forward.

The Enbridge Line 6B diluted bitumen oil spill into the Kalamazoo River created a challenging situation that was overcome by using a combination of conventional techniques for floating oil and a multiple-lines-of-evidence approach for submerged oil. Recovery of submerged oil has been particularly difficult and protracted, with recovery continuing into the 4th year after the spill. Given that the magnitude and composition of this release into a freshwater ecosystem has no historical precedent, it is valuable to consider the lessons learned in its cleanup and remediation for planning for other heavy oil spills in the Great Lakes Region.

At the onset of cleanup, conventional recovery techniques for the Kalamazoo River targeted floating oil and oil deposited in overbank areas on vegetation and soils. However, as attention shifted to the substantial occurrence of submerged oil in the river sediment, techniques were developed to raise the oil to the surface to allow its recovery. Ultimately, it was necessary to recover oiled sediment and sludge by conventional methods such as hydraulic dredging.

The multiple-lines-of-evidence approach developed for assessment and recovery of submerged oil, when integrated with a robust GIS helped to optimize submerged oil assessment and recovery techniques, inform decision-making related to transitions and endpoints, and ultimately clean as much oil as possible while limiting long-term ecological damages from the oil recovery activities.