Lessons Learned On Oil Spill Environment Impact Assessment: 10 Years of Petrobras Experience Review

In the beginning of the 2000s, the experience on four oil spills fostered a major change in health, safety and environment (HSE) practices in Petrobras. Three of the spills encompassed pipeline ruptures and one consisted of an offshore platform sinking. The spills took place in: Guanabara Bay, close to REDUC refinery, January 2000; Iguaçu river, close to REPAR refinery, July 2000; Serra do Mar mountain ridge, February 2001 and platform P-36, offshore Campos Basin, March 2001, respectively denominated henceforward Guanabara Bay, OSPAR, OLAPA and P-36 spills. Previous information about these accidents were reported in the literature (Michel, 2000; Gabardo et al. 2001; Meniconi et al 2001; Meniconi at al. 2002,, Barcellos et al, 2003; Falkiewicz 2003, Millet et al. 2003; Faria et al. 2005).

Guanabara Bay spill: On January 18^th, 2000, a spill of roughly 1300 m³ of marine fuel MF-380 took place in Guanabara Bay, in Rio de Janeiro State, due to a pipeline rupture of a Petrobras refinery. The spill occurred in the fringe of a mangrove, reaching some nearby coastal ecosystems such as rocky shores, sandy beaches and mangrove forests (Michel, 2000). The spilled oil characteristics were: density (0,9817), 12 °API, viscosity (20°C) of 5313 cP, boiling point range from 174°C to 750°C (26% w/w residue), BTEX content of 1.56 mg.g⁻¹ and PAH of 50 mg.g⁻¹. Initial efforts included mechanical oil recovery from the water surface, mangrove areas protection with booms, beach cleanup, rocky shore cold water flushing and oiled birds rehabilitation (Barcellos et al, 2003). Evaluation of oil on the shoreline and environmental impact assessment with water and sediment sampling were conducted afterwards. Fish were collected to assess contamination, and the results allowed fishery activities opening on the 10^th of March/2000, (Meniconi et al. 2001). Long term environmental monitoring has been performed in Guanabara Bay since the accident.

OSPAR spill: On July 16^th, 2000, approximately 4000 m³ of crude oil spilled due to a pipeline rupture in the scraper area of a Petrobras refinery located in the state of Parana (REPAR), southern Brazil. The spill affected a small stream, a wetland zone and Barigui and Iguaçu rivers. The spilled oil characteristics were: density (0.8164), 41 °API, viscosity (20°C) of 3716 cP, IBP < 36°C to 750°C (11%w/w residue), BTEX content of 41.8 mg.g⁻¹ and PAH of 16 mg.g⁻¹. The oil contention and recovery on both rivers was quite efficient but part of the oil remained trapped on the wetland, demanding a restoration program. Long term environmental monitoring has been performed in the OSPAR affected area since the accident.

OLAPA spill: On February 16^th 2001, a big landslide in Serra do Mar mountain ridge, in the state of Parana, caused a disruption of OLAPA pipeline, releasing approximately 52 m³ of light fuel oil in creeks, rivers and Nhundiaquara estuary. The spilled oil characteristics were: density (0.9427), 18 °API, viscosity (20°C) of 5,000 cP, boiling point range from 120°C to 380°C, BTEX content of 2.74 mg.g⁻¹ and PAH of 276.9 mg.g⁻¹.

P-36 spill: In March 2001, following the sinking of Stationary Production Unit P-36 in water depths of 1,300 m, diesel fuel and oil stored on the platform spilled on the ocean. An environmental sampling campaign 8 to 10 days after the accident was conducted and results indicated no changes in seawater quality.

Those events mobilized a significant number of professionals at Petrobras in order to minimize effects and assess environmental changes after oil spills in a more effective way. Based on the experience of the above four oil spill impact assessments, PETROBRAS focused on developing and improving the following tools for emergency preparedness and response: modeling of oil spill scenarios for contingency planning, sensitivity maps, remote sensing, forensic chemical analysis, environmental assessment and monitoring, mangrove and wetland restoration and underground contaminant transport simulation model.

Furthermore, one of the lessons learned by PETROBRAS in the past decade is that legal aspects must be one of the concerns necessarily involved in the design, planning and execution of emergency response.

An oil spill modeling project for 25 unities (TRANSPETRO oil terminals and Petrobras refineries) along the Brazilian coast was developed for contingency planning and environmental permitting, using a detailed meteorological and oceanographic characterization of each study area. Based on the requirement of scenarios for both typical and critical weather conditions, the oil fates and trajectory (deterministic and probabilistic simulations) modeling system OILMAP was applied to simulate oil spill incidents in all 25 sites. Due to Brazil's continental dimensions, the simulated environmental conditions ranged from the subtropical Patos Lagoon in the south to the equatorial Solimões River in the Amazon Basin, including estuaries, coastal zones, and offshore sites (Figure 1).

The system allows oil spill contingency specialists to develop response plans for typical spills in the selected locations. Figure 2 shows a result of the probabilistic modeling for Ilha d'Agua Terminal in Guanabara Bay. The colors indicate oil probability in water.

Operational hydrodynamic models were also implemented for three terminals along the Brazilian coast: Ilha Grande and São Sebastião, in the southeast region and Guamaré, in the northeast, providing a 72 hours forecast for the currents conditions.

To prevent and minimize environmental damages due to oil spills, great efforts were taken for the improvement in sensitivity maps to be used for prevention and response in emergencies. Twenty-one coastline areas and two areas in the Amazon Region (Brazil) have been mapped for their sensitivities to oil spills, in collaboration with several universities. Afterwards, onshore areas around refineries and pipelines were also mapped (Araujo, 2007).

A specific software was developed (MAPS) to produce these sensitivity maps and make it possible to update, print and promptly retrieve the huge amount of data gathered during the surveys. The methodology was a customized version from the NOAA methodology. The great advantage is that the software can be integrated with other emergency response computational tools. Nowadays a new approach is being developed for inland areas, mainly concerning rivers, which is based on geomorphology and also includes vulnerability indexes to oil spills.

For the Guanabara Bay spill, Landsat-5/TM and Radarsat-1 images were used. Image processing techniques improved the detection of different oil-covered areas on the bay surface (Figure 3). The information was used to confirm the spilled volume and as a starting geometry for oil drift models. As the oil patch spread out around the bay, the oil detection was more difficult due to false targets, such as wind shadow zones and biogenic oil. The use of ancillary information from aerial inspections allowed the identification of the real oil covered areas.

Orbital remote sensing data was the most cost-effective tool to elaborate oil spill maps with cartographic precision. The use of SAR (Synthetic Aperture Radar) data proved to be more effective due to its capacity of emergency tasking, revisit and near real time delivery. The low frequency of acquisition and the prevailing cloud cover limited the use of visible and near-infrared data. Oil spill maps were used to follow the trajectory and fate models results. The GIS integrated analysis of the image dataset, field data and ancillary information acquired through aerial inspections presented an important decision tool during the emergency, improving the Guanabara Bay contingency plans and evaluation of the spill impacts.

Nowadays PETROBRAS is using spaceborne multi-sensor remote sensing for sea surface monitoring along the Brazilian coast. An integrated methodology was established based on the experiences and knowledge acquired in the past decade spills events. Synthetic Aperture Radar (RADARSAT and ENVISAT), ocean color (MODIS and MERIS), thermal infrared (NOAA/AVHRR), scatterometer (QuikSCAT) and data were integrated in order to detect and characterize different sorts of marine pollution and meteo-oceanographic phenomena.

The coexistence of multiple sources of hydrocarbons in the environment (natural and anthropogenic) resulted in complex scenarios, showing the need to use the best available techniques in order to identify and differentiate for source studies of areas affected by these compounds. Moreover, drifting oil patches or unidentified oil spills bring out the greatest need for forensic chemical analysis, due to their liability and responsibility.

Several analyses have been performed for oil characterization during the 2000 and 2001 spills in order to correlate source and spilled oil samples. Efforts to speed up time of response for analytical results, maintaining QA/QC practices in high status, were important mainly in differentiation of false targets found in sampling campaigns. The techniques used by Petrobras for differentiating sources of hydrocarbons in the environment not only in the oil spill impact assessments but in monitoring projects include: a) Diagnostic indexes of aliphatic hydrocarbons including n-alkanes, isoprenoids, branched resolved hydrocarbons and UCM; b) Diagnostic ratios of PAH (Wang et al., 1999; Yunker et al., 2002); c) Principal Component Analysis for PAH (Yunker et al., 2000) and d) Geochemical distribution of saturated and aromatic hydrocarbons, including the biomarkers hopanes and steranes, among others.

In addition, Petrobras has improved its knowledge in forensic methodology for identifying waterborne oils in aquatic environments by participating of the Bonn-OSINet Round Robin Oil Spill Identification. Investment is also being done on Compound-Specific Stable Isotope Analysis (CSIA) for differentiation of sources of light fuels (O'Malley et al, 1994; Benson et al, 2006; Meier-Augenstein, 1999; Brenna et al, 1997).

A new approach for spilled oil characterization and weathering evaluation using chemometric analysis of sections of chromatograms (Christensen et al, 2007) was applied in the characterization of the complex PAHs pollution patterns in sediments from Guanabara Bay (Christensen et al, 2010), resulting in the discrimination of the main distinct sources of 3 to 6-rings PAHs and the identification of new potential diagnostic ratios of PAHs (Figure 4):. By this methodology it was also possible to conclude: 1) Rio de Janeiro harbour (station BG 05) is the most contaminated site in the bay with mainly pyrogenic sources mixed with a fraction of high-molecular-weight petrogenic PAHs; the same pattern was observed in a plume stretched from the harbour to the North East direction (BG08, BG10, BG11, BG16, BG25 and BG27); 2) São João de Meriti River (BG37), due to an urban and highly industrialized area, is the second largest source of PAHs, introducing mainly a fraction of low-molecular-weight petrogenic PAHs, and 3) the sites close to the ruptured pipeline (BG31) in the Guanabara Bay spill, which also receives a chronic input of pollution from the Sarapuí and Iguaçu rivers (BG32), show a distinctive pattern indicating presence of heavy PAHs petroleum fraction.

Finally, beyond focusing only on analytical parameters correlated to oil industry, externalities deriving from other activities such as organochlorine pesticides and polychlorinated biphenyls (PCBs) in sediments from Guanabara Bay have been investigated, as they may play important role in toxicity when present in the environment. Unpublished results in sediments from Guanabara Bay indicated the presence of both classes of compounds in levels of concentration higher than the Threshold Effect Level (TEL - Buchman, 1999) in some of the investigated stations and PCB higher than the Probable Effect Level (PEL) in the mouth of São João do Meriti River.

So, in ten years of experience, Petrobras has accumulated experiences to distinguish the influence of other chronic pollutants sources in the areas of influence of their facilities and keeps looking for new methods to face analytical challenges.

Among the Brazilian oil spill lessons learned, some of the most important were the need for sound environmental information. As Petrobras has a widespread presence all over Brazil, offshore and onshore, with refineries, terminals and E&P platforms, our greatest challenge has been to cope with the lack of baseline surveys of ecosystems in all our areas of influence. The baseline ecological state can be used to evaluate environmental changes and be a goal for ecological restoration, after an oil spill.

Due to the need of biodiversity knowledge and a lack of large scale ecological understanding of most Brazilian ecosystems, especially in deep sea and other offshore areas, a huge effort is being undertaken to accomplish regional comprehensive assessments carried out by several research institutes under the coordination of Petrobras.

This task comprehends previous data consolidation and synthesis as well as huge characterization efforts with original data collection which have been done or are still on their way from North to South of Brazil and from coastal areas to the deep ocean. Ecosystems range from the Amazon and the Atlantic rainforests to river basins, up to coastal resources in bays, mangroves forests and plunging down to the deep sea.

The challenge of expanding our offshore activities along the continental margin and into deeper domains has led to some extensive environmental assessments which take place in five large marine ecosystems in different sedimentary basins: a) Rio Grande do Norte and Ceara; b) Sergipe and Alagoas; c) Espirito Santo; d) Campos and e) Santos. The latter three, encompass pre-salt oilfields, the new frontier for E&P in Brazil. Onshore, the challenge is to deal with very sensitive areas in specific river basins, such as Iguaçu (PR) in southern Brazil and Urucu petroleum province in the Amazon Forest (North), both under intensive research programs. Very broad multidisciplinary studies were also carried out in Guanabara Bay (RJ) and Todos os Santos Bay (BA), coastal areas with high degree of complexity, where Petrobras has several operational units. In Guanabara Bay spill, some interesting results came out from the comparison of data from oiled rocky shores that were cleaned with cold water flushing (higher mortality) with data from an untreated oiled rocky shore and previous data. The decision to flush was more due to stetic and touristic reasons and required by the environmental agency.

Besides all characterization efforts, environmental monitoring has been an important activity for the company around its marine and inland sites for decades. In areas affected by oil spills, long term monitoring is being performed to follow ecological recovery and evaluate remediation strategies.

To evaluate the recovery of the affected mangrove in the Guanabara Bay spill, spatial analysis of the vegetation cover evolution was performed, by processing satellite and aerial images, adopting GIS routines (Blasco et al. 1998, Dahdouh-Guebas et al., 2004). Also, comparative analysis of flora and fauna variables, using significance tests to assess reforestation effectiveness, was performed in both impacted areas: reforested, and naturally recolonized. No significant differences were pointed out between reforested and natural colonized areas on the analyzed floristic parameters (total and individual species basal area, forest and individual species average height, and the ecosystem complexity by the Holdridge Complexity Index. The results also indicate that natural recolonization after the spill was abundant and effective in most affected areas. No significant differences were found on the infauna between reforested and natural colonization impacted areas. The crabs fauna survey accomplished on mangrove reforested areas about 10 years after the spill, presented similar results in both areas. The results lead to conclusions that reforested areas are in very similar conditions to those of natural recolonization.

During the OSPAR spill, a freshwater wetland region covering more than 14 hectares was partially contaminated by the oil. In a concordance with the Parana State Environmental Protection Agency (IAP), the decision was to remediate and restore this wetland excluding the use of intrusive technologies, avoiding more damages to the ecosystem. As this region has an important ecological function, a persistent contamination could result not only in a source of oil to the rivers, but also a contamination risk for the resident biodiversity. In this way, some intervention was necessary in order to improve natural flushing and biodegradation of the remaining oil. Technologies were proposed and implemented to stimulate biodegradation by aeration, flow up the oil trapped in soil by a hydraulic system, and a final treatment by water-oil separators. A small creek in the wetland area was deviated from its original channel, in order to facilitate aeration for biodegradation stimulation. The PAH, BTEX and TPH concentration were monitored both in soil and groundwater, in several different stations covering the whole wetland.

After 10 years, the major part of the region presented TPH concentrations well below 5,000 mg/kg, which is the value considered as intervention limit, and in groundwater the PAH and BTEX concentrations were bellow the Brazilian federal standard for soil and groundwater (CONAMA 420/2009).

Several studies concerning biological communities, such as aquatic benthic invertebrates, fishes, terrestrial plants, mammals and birds are being conducted by local universities. As no information prior to the accident is available, the results are being compared to non affected areas, in a way to estimate the recovery of those communities.

The SCBR (acronym in portuguese for “Risk Based Corrective Solutions”) model is a computational tool developed by Petrobras Research Center and Federal University of Santa Catarina, for simulation of transport and fate of contaminants; for risk assessment of the exposure pathways in soil, water and air; and simulation of remediation technologies. An advantage in comparison to other groundwater transport models is its application to different products, including biofuels.

The model development included field based tests in an experimental site with different Brazilian fuels, released in soil and groundwater. One case study was conducted in São Sebastião Oil Terminal and the methodology was based on the Contaminated Areas Management Manual of São Paulo State Environmental Protection Agency (CETESB). The results obtained were the groundwater flow model for the area, the knowledge of the contaminant behavior for this scenario (Figure 4), the risk area mapping and information on how an emergency intervention could minimize the risk. The model application during an oil spill therefore results in a gain of time and efficiency in order to reduce soil and ground water impacts.

Under Brazilian law, any technical or technological approach (emergency response, environmental monitoring, damage assessment and restoration) should undergo legal validity, prior to its use under Brazilian jurisdiction, otherwise one may suffer negative consequences in three different and virtually independent aspects: civil, administrative and criminal.

In Brazil, environmental liability law must be considered before the adoption of the emergency response actions, for example the use of dispersants.

Another aspect which differs substantially from other legal systems (where consequences from oil spills may be resolved solely by paying compensation or damages) is that restorarion or remediation of the affected environment must be prioritized, whereas the option of paying damages will only exist if restoration proves impossible - at least in part. Prior consent from the environmental agency is always required.

All the case studies presented on this paper have been carried out complying to these legal requirements.

After the spills in the beginning of the decade, Petrobras applied different tools to assess oil spill impacts and remediate ecosystems, reaching new levels of preparedness and response capability to incidents.

Oil fate modeling system, sensitivity maps, remote sensing, forensic chemical analysis, environmental assessment and monitoring, mangrove and wetland restoration and SCBR modeling for contaminated areas were the most important tools developed and customized in the company.

Environmental projects carried out by Petrobras up to date and the several ecosystem baselines provided by them are now playing a crucial role in the company's decision making process, as well as producing relevant data and scientific knowledge, stimulating national taxonomic, chemical and ecotoxicological expertise and sponsoring biological and chemical laboratories infrastructure and biological collections in Brazil.

The authors would like to thank all Petrobras team without whom this challenge could not have been accomplished. We are grateful to Denis Millette, from Hidrogeoplus for the wetland restoration studies and all other academic institutions and companies who worked for Petrobras and contributed for the results. Finally, we specially acknowledge the Petrobras/HES Department for their assistance and cooperation.