Marine Hydrocarbon Spill Assessments: From Baseline Information through to Decision Support Tools

Marine Hydrocarbon Spill Assessments: From Risk of Spill through to Probabilities Estimates describes the methods used for estimating hydrocarbon spill risks and the potential consequences. Throughout the book, mathematical methodologies and algorithms are included to aid the reader in the solving of applied tasks presented. Marine Hydrocarbon Spill Assessments: From Risk of Spill through to Probabilities Estimates provides a fundamental understanding of the oil properties and processes which determine the persistence and impacts of oils in the marine environment. It informs the reader of the current research in hydrocarbon spill assessments, starting from an assessment of a risk of a spill, and moving on to modelling approaches to impact assessments, laboratory toxicity assessments, field impact assessments and response options, and prevention and contingency planning.

• Identifies efficient solutions to protect coastal regions from the marine pollution of hydrocarbon spills
• Includes case studies examining and analyzing spills, providing lessons to prevent these in the future
• Covers the science of oil spills from risk analysis to cleanup and the effects on the environment

• Language: English
• Publisher: Elsevier Science
• Release date: Aug 19, 2021
• ISBN: 9780128193730

Book preview

Chapter 1

Baseline data for spill assessments: ambient conditions, socioeconomic data, sensitivity maps

Lucy Romeo¹, Patrick Wingo², Michael Sabbatino³ and Jennifer Bauer²,    ¹National Energy Technology Laboratory, Morgantown, WV, United State,    ²National Energy Technology Laboratory, Albany, OR, United States,    ³NETL, Albany, OR, United States

Abstract

Effective oil spill preparedness and response relies heavily on the availability of baseline data. Baselines comprise measurements and information collected prior to natural or anthropogenic disasters and can be applied to predict the transport and fate of pollutants, plan for socioeconomic stressors, and overall mitigate impacts. Spatial and temporal in nature, these datasets represent the current state of a specific area. Baselines representing offshore areas comprise ambient conditions, socioeconomic statuses, and environmental sensitivities. This chapter will highlight the value of baselines and identify means to collect and build representative databases for marine and coastal ecosystems to aid in spill assessments.

Keywords

Baseline; data; spill preparedness; spill response; ambient; socioeconomic; sensitivity mapping; machine learning

1.1 Why and for what baseline data are needed?

Efforts to effectively prevent, prepare for, and respond to natural or anthropogenic disasters often rely heavily on the availability and accuracy of baseline data. Baseline data are measurements and information collected prior to an event or specific activity, such as a disaster, restoration project, management change, or industry-driven development. Baseline data are typically spatial or temporal in nature, representing a state of existence in the real world. In marine and coastal ecosystems, baseline data can encompass a variety of ambient, environmental, social, economic, and human health variables. This chapter will highlight the value of baselines and identify means to collect and build representative databases for marine and coastal ecosystems to aid in oil spill preperation, assessment, and response.

Major marine oil spill events, such as the Gulf War oil spill, the Deepwater Horizon blowout, and the Sanchi collision have stressed the value of baseline data to evaluate the magnitude of environmental, economic, and human health effects (Carswell, 2018; Goldstein, Osofsky, & Lichtveld, 2011; Joyner & Kirkhope, 1992; Nelson, Bauer, & Rose, 2014; Zhang et al., 2020). During a spill, baseline data are commonly used to guide cleanup and response efforts to protect vulnerable and sensitive communities and environments through comparison with information on the spill itself, such as the amount of oil, the trajectory and movement of oil, duration oil expected to be present, and its physical properties (Nelson, Grubesic, Sim, & Rose, 2018). Following a disaster, restoration efforts rely on baseline data to effectively monitor and assess the degree of success for both ongoing and completed projects. Beyond their value for spill response and restoration, baseline data are critical for planning, prevention, and preparedness efforts.

Lessons learned in various areas of marine ecological management and disaster response have demonstrated the value that baseline data provide in understanding which areas are most vulnerable or sensitive. These data and the derived insights are often used to develop sensitivity maps for various marine and coastal ecosystems, helping decision makers identify what resources might be impacted, and informing disaster preparedness and response efforts as needed. Baseline data also offer support for modeling efforts, affording researchers and scientists the ability to simulate spill events, assess what-if scenarios, and use the findings to evaluate the effectiveness of different technologies and various preparedness and response strategies (Duran et al., 2018; Stelzenmüller, Lee, South, Foden, & Rogers, 2013). In addition, persistent observations and measurements found in baseline data often provide the larger volumes of data necessary to support the training, testing, and validation of new technologies and models that involve integrating artificial intelligence and machine learning techniques and algorithms.

Outside of spill prevention, preparedness, and response, the collection and recording of baseline data allows for trends to be tracked and problems to be identified. In addition to offering insight into current conditions, a robust collection of baseline data might identify where data are otherwise missing, incomplete, or poor. Baseline data can also provide key insights for decision makers designing new policies or management strategies, as well as facilitate research to evaluate the effectiveness of said management and policy changes (McLeod & Leslie, 2009; White, Halpern, & Kappel, 2012). As such, the determination of which baseline data to include in an analysis or assessment should largely be situational, based off criteria set to determine the degree of impact, success, or failure for a given event or activity.

1.2 Types and sources of baseline data

This section covers three categories that combined makeup most spatial baseline datasets: ambient data, socioeconomic data, and sensitivity mapping (Fig. 1.1). Ambient data represent marine environmental phenomena, including currents and wind. These data, whether from monitoring stations, field observations, or derived from external models, enable an understanding of the surrounding environment wherein oil spills might occur. These are critical for understanding the fate and transport of oil spill scenarios. Socioeconomic data represent quantitative and qualitative measurements for civil, industrial, and commercial factions, which might be impacted by oil spills. Sensitivity maps contain data representing coastal resources, including critical habitats and the distributions of threatened or endangered species. For each category we provide a high-level discussion of how data can be used, the formats they are commonly found in, and their availability.

Figure 1.1 Venn diagram representing how the combination of ambient, socioeconomic, and sensitivity mapping data build a baseline for a given marine or coastal area.

1.2.1 Ambient

Ambient data represent environmental factors like currents, wind, and waves and may be obtained from a network of observational or monitoring stations, buoys, and drifters or from remote sensing data sources, such as satellites (Maximenko et al., 2009). Ambient data are critical for modeling and understanding the transport and fate of spill or blowout events. Monitoring data come in an array of spatial and temporal resolutions and might be too sparse for deriving useful information due to physical, environmental, or monetary restraints.

To produce a detailed hindcast, present time, or forecast picture of the oceanic and atmospheric processes over an area of specific concern, ambient data are often generated using vetted numeric ocean models. These models utilize a wide array of inputs, assumptions, and equations to approximate the conditions found within an ocean’s water column and at the water surface (NOM Group, 2003). Some common attributes simulated by ocean models include water velocity, temperature, salinity, pressure, and surface wind interaction (NOM Group, 2003). Examples of ocean models include the Navy Coastal Ocean Model (NCOM; https://www.ncdc.noaa.gov/data-access/model-data/model-datasets/navoceano-ncom-reg) and the Hybrid Coordinate Ocean Model (HYCOM; https://hycom.org/). Sometimes a combination of models can be used; for example, one model could simulate the conditions in the subsurface water column, while another model could provide values used for simulating wind. Spatially explicit ocean models are typically gridded, meaning that all ambient characteristics are consistent across a given region (NOM Group, 2003). Cartesian and block-grids can be spatially structured or unstructured along the x–y or longitude–latitude plane (NOM Group, 2003). Curvilinear structured grids divide the model region into quadrilateral units, which can be linear, or curved (Thompson et al., 1999). The grid cells in a curvilinear grid are distorted along one or more parallel curves (Thompson et al., 1999), which is useful for fitting data in extreme polar regions. Unstructured grids typically take the form of a trimesh; that is, each region is shaped as a triangle, although there are other geometric meshes occasionally used for building unstructured grids, such as orthogonal, hexagonal, or mixed shape meshes (Thompson et al., 1999). Regions in an unstructured grid can vary wildly in size, with smaller cells used in regions of high geographic complexity, such as coastlines, inland seas, and dense archipelagos (Thompson et al., 1999). Examples of these gridding methods are shown in Fig. 1.2.

Figure 1.2 Examples of the various gridding methods applied. (A) Cartesian grids, (B) curvilinear grids, and (C) unstructured trimesh.

With ocean models that also capture depth, there are several approaches to take as the depth regions may follow a regular interval or may become shorter near the surface and the ocean floor. Shorter intervals typically represent depth regions of higher complexity (NOM Group, 2003). Additionally, the contours between regions may follow specific depths or follow the approximate contours of the underlying bathymetry (NOM Group, 2003). There are a number of ways to represent ambient attributes across space; it is important for a spill modeler to understand these representations as appropriate for the ocean models used to derive baseline simulated ambient data.

Data originating from an ocean model can be used as surrogates for static or dynamic conditions in other physical models, such as those that simulate oil transport throughout the water column (Duran, Beron-Vera, & Olascoaga, 2018). Many of the traits mentioned above are used to provide data to drive the ambient conditions, which are critical for simulations (Sim et al., 2015). Oil spill simulation models, such as the US National Energy Technology Laboratory’s (NETL) Blowout and Spill Occurrence Model™ (https://edx.netl.doe.gov/blosom/) or the National Oceanic and Atmospheric Administration’s (NOAA) General NOAA Operational Modeling Environment (https://response.restoration.noaa.gov/oil-and-chemical-spills/oil-spills/response-tools/gnome.html), frequently take advantage of the simulated ambient characteristics produced by ocean models. This can increase the complexity of processes being simulated without the associated computational costs of running a full-scale ocean model simulation in addition to simulated oil spill fate and transport processes.

As previously discussed, ambient data are the driving force behind the transport and destination of each representative spill parcel within many oil spill simulation models. Without baseline data, there will be little variation in simulation runs, and there would be no defensible connection to observable real-world phenomena. While large-scale ocean models may produce data in a custom format (Wallcraft, Carroll, Kelly, & Rushing, 2003), data are often shared using file formats which are universally available to the research population. One of the most widely used formats for distributing model-derived ambient data is Network Common Data Form, also referred to as netCDF (Unidata, https://www.unidata.ucar.edu/software/netcdf/). The netCDF file format was designed with scientific data in mind, with data stored as series of platform-independent arrays annotated with metadata describing their representations (Unidata). Due to the large nature of datasets generated by ambient models, several specialized data transfer protocols exist for pulling down slices of available data; one of the most common protocols is the Open Network Data Access Protocol (OPeNDAP, https://www.opendap.org/).

While not all ocean modeling teams make their data public, many do, particularly those hosted by public institution, such as NOAA’s National Centers for Environmental Information (NOAA; https://www.ncei.noaa.gov/). Model output data can usually be downloaded directly, packaged in files (often netCDF), or accessed using a special connection protocol, such as the previously mentioned OpenDAP. Often times, the model data are available through a variety of means via a Thematic Real-Time Environmental Distributed Data Services (THREDDS) data server (Unidata; https://www.unidata.ucar.edu/software/thredds/current/tds/). Data produced by both the NCOM and HYCOM models are provided on their own THREDDS data server; at the time of this writing, NCOM-derived data can be retrieved from https://www.ncdc.noaa.gov/data-access/model-data/model-datasets/navoceano-ncom-glb by clicking on one of the TDS links in the table found on this page. Similarly, HYCOM results can be found on https://hycom.org/dataserver.

1.2.2 Socioeconomic

Socioeconomic data can represent an array of civil, industrial, and commercial factions and are valuable to quantify the potential costs of oil spill events. Socioeconomic data can also represent vulnerability measurements, monetary values, including cost of the oil lost or cost of resources or activities impacted, or indices representing the demographics of an area. These data can originate from surveys, industry reports, previously published datasets, papers, news, and models.

Understanding the cultural and economic importance of marine sector industries and activities, such as fishing, tourism and recreation, and the energy industry, is key for building a representative baseline. This information can be used to assess an area’s spatial vulnerability, predict potential economic costs, and build better response plans. The use of socioeconomic data in a baseline varies regionally and is dependent on present industries and activities, data availability and accuracy, and spatial scale. For example, when evaluating future oil spill response preparedness at a spatially explicit port complex in Brazil, a study mapped area vulnerability using digital maps and satellite images, and gathered survey information on neighborhood income, education, and dependence on fishing (de Andrade, Szlafsztein, Souza-Filho, dos Reis Araújo, & Gomes, 2010). At a global scale, the International Oil Pollution Compensation Fund analyzed oil spill cleanup cost and total cost to build a baseline for the International Maritime Organization (IMO) to better understand how to evaluate environmental risk (Kontovas, Psaraftis, & Ventikos, 2010). This large-scale study applied regression analysis on past spill statistics, including the total amount of oil spilled and number of spills, to quantify historical cost and estimate costs moving forward.

The formats and availability of socioeconomic data vary. Formats include data already spatially structured, with geometric or coordinate information tied to areas or locations on the earth, such as comma-separated values (CSV) files and shapefiles. These structured formats can be more easily reformatted, transformed, and analyzed. Less accessible data formats include Portable Document Format (PDF) and hardcopy papers, both of which require additional processing to convert the data into a format suitable for analysis. These formats are especially common for archiving historical socioeconomic data records.

Depending on the region of interest and spatial scale, socioeconomic data can be accessed through online databases, government websites, industry reports, or literature on past spills or spill cleanup reports. These data are typically already processed into products such as publications and static maps, which take additional sleuthing to find individual, spatial, and temporal values. Socioeconomic data can also be obtained as outputs from models. For example, to better understand response costs and socioeconomic damages of offshore oil spills in the US, the Environmental Protection Agency (EPA) developed the Basic Oil Spill Cost Estimation Model (BOSCEM) (Etkin, 2004). BOSCEM incorporates data on the amount spilled, type of oil, location-specific socioeconomic values, environmental vulnerability, and response effectiveness (Etkin, 2004). Building an understanding of models like BOSCEM, users can leverage the framework for their own areas of interest and purposes.

For areas where credible publicly available data are hard to come by, researchers would either need to collect and assemble data, which is dependent on funding and local interest, or rely on what resources are available and derive a socioeconomic proxy from what information they have. In a 2015 study assessing the potential socioeconomic and environmental impacts of oil spills in the US Gulf of Mexico, researchers were unable to acquire credible, spatial, publicly available socioeconomic data relating to tourism, which is a common and valuable indicator of coastal economies (Bauer et al., 2015). Coastal tourism in the Gulf of Mexico contributed 11% to the overall Gross Domestic Product of the US economy in 2010, with the industry employing 50.6% of people working in ocean sectors (National Oceanic Atmospheric Administration, 2013). To overcome this hurdle, researchers leveraged available hotel data and seasonal information by conducted interviews with hotel staff to calculate profit associated with room occupancy as a proxy for tourism (Bauer et al., 2015). Results of this data collection process are shown in Fig. 1.3, which displays the seasonal normalized profit per hotel in the US along the Gulf of Mexico’s coastline.

Figure 1.3 Maps illustrating the seasonal fluctuations of tourism along the Gulf of Mexico's US coastline. The maps shown above applied normalized hotel profit differences season-by-season as a proxy for coastal tourism, the data of which were created as a socioeconomic representation of the baseline for the US Gulf of Mexico (Bauer et al., 2015). Data were derived from locational information, state-wide seasonality obtained from literature and hotel interviews, room rates, and number of rooms per hotel.

1.2.3 Sensitivity maps

Sensitivity maps contain data representing areas that are environmentally vulnerable to oil spills, including shorelines, critical habitats, and natural resources. Information derived from these maps is critical for preparation and response mitigation to reduce the potential environmental consequences of oil spills. Sensitivity maps vary regionally, and in some cases, need to be created using environmental and natural resource data.

The most widely used approach to sensitivity mapping throughout the world is Environmental Sensitivity Index (ESI) (Jensen, Halls, & Michel, 1998). ESI are spatial information that are comprised of three factors: shoreline types, oil-sensitive biological resources, and commercial, recreational, and human-use resources (Fig. 1.4). ESI maps were first applied in 1979, when ESI maps of the Texas coast were prepared to evaluate impacts from the Ixtoc I well blowout in the Gulf of Mexico (Jensen et al., 1998). Since then, sensitivity maps, including ESI, have been built, updated, and applied in many regions. The ESI approach has been implemented throughout the world including the US, Brazil, Canada, United Arab Emirates, Greenland, India, Israel, Jordan, El Salvador, Germany, South Africa, Mauritius, Nigeria, and New Zealand.

Figure 1.4 Example of Environmental Sensitivity Index data for the state of Louisiana (the US), including (A) a subset of resources, (B) a subset of habitats, and (C) shoreline types. Source: Data were procured from Navy Coastal Ocean Model’s Office of Response and Restorations online Environmental Sensitivity Index data page: https://response.restoration.noaa.gov (National Oceanic, 2014).

Along Brazil’s coast sits one of the world’s most continuous mangrove habitats, which are considered by NOAA to be a sensitive cover-type flora to oil spills (NOAA, 1997). ESI maps were produced along this region for oil spill contingency planning, applying remotely sensed elevation (i.e., topography and bathymetry), geomorphological, meterological and oceanographic, biological, and socioeconomic data (Souza Filho, Goncalves, de Miranda, Beisl, & de Faria Almeida, 2004).

Nigeria is the largest oil producer in Africa, with the maximum crude oil production capacity of 2.5 million barrels a day according to the Nigerian National Petroleum Corporation (2018). The Nigerian government estimates that approximately 7,000 oil spills have occurred between 1970 and 2000 (Baird, 2010). In an effort to reduce the impacts of oil spills along Nigeria’s coast, the Federal government, nongovernment agencies, and oil firms developed oil spill management policies, which included ESI mapping developed by the Environmental Systems Research Institute (Nwilo & Badejo, 2006).

A risk assessment with sensitivity maps was conducted along India’s Chennai coast to better prioritize resources more likely to be impacted by oil spills (Kankara, Arockiaraj, & Prabhu, 2016). This assessment combined coastal resource information to build ESI maps and outputs from oil spill models to better understand the potential movement of oil spills in proximity to coastal resources.

Where sensitivity maps are available, they can take multiple forms. Static and web maps are two products of sensitivity mapping. Static maps, which can be printed or available digitally as PDF files or images, comprise standalone visual information that can be used to understand proximities but does not link features, attributes, and boundaries for deeper analyses. Because of their static nature, it is important for users to gather metadata on these maps to make sure they are not out of date. Time permitting, users can take static maps and digitize them into spatial data, which can then be updated and analyzed, leveraging additional data sources. As the name implies, web maps are hosted online and enable users to visualize the data in multiple ways. Web maps are more easily updated as the data can be automatically fed to the platform through background processes or have linked sources. Furthermore, data displayed in web maps can be made available for download in spatial