Environmental risk assessment of marine activities

Based on Le Roy, D., Volckaert, A, Vermoote, S., De Wachter, B., Maes, F., Coene, J. and Calewaert, JB., 2006. Risk analysis of marine activities in the Belgian Part of the North Sea (RAMA). Research in the framework of the BELSPO Global change, ecosystems and biodiversity – SPSDII, April 2006, 107 pp + Annexes. Available at [1]

In broad terms, risk assessment is the procedure in which the risks posed by inherent hazards involved in processes or situations are estimated either quantitatively or qualitatively. Environmental Risk Assessments (ERA) are carried out to examine the effects of an entity or agent on humans (Health Risk Assessment) and ecosystems (Ecological Risk Assessment). They cover a broad spectrum of risks, receptors and end-points: an ERA can focus on biological, chemical, radiation and/or physical risks towards impacts on receptors such as human beings (individuals or population), fauna and flora (single species or whole ecosystems), and materials (e.g. impacts on building by acid rain, loss or damage of property). For each of these risk receptors different end-points are defined: for example mortality and morbidity in human health assessment, property loss in fire, revenue loss for people depending on the harmed ecosystem in the economic impacts assessment, extinction or total catch in ecological risk assessment.

Dependant on which risks, receptors and end-points one wants to investigate, the steps and methods to be used in ERA will differ. An adapted approach of ERA is required since a general description of the key tasks and methods in risk analysis is not possible.

Here we focus on Ecological Risk Assessment (EcoRA) of Marine activities. We first describe the general aspects and steps of the ERA process that most risk assessments are confronted with. The application of risk assessment process is then illustrated in a case study focusing on marine activities (Merchant Shipping more precisely) in the Belgian Part of the North Sea (BPNS).

Environmental risk assessment of marine activities

The usefulness of a risk assessment of marine activities depends on the method(s) used and the purpose of the results. The analysis of risks can be approached in several ways, from very general and rough to very detailed. A risk analysis consists of several consecutive steps and for each step in the risk assessment process a number of methods are available. Each of these methods come with advantages and disadvantages. A number of factors influence the most appropriate method to be used in a given circumstance, including the type of activities, the study area, potential gaps in data and/or models and the uncertainty issue of risk assessment. Here we focus on methods relevant to risk sources related to marine activities and to marine and coastal ecosystems as potential receptor.

QUANTITATIVE VERSUS QUALITATIVE RISK ASSESSMENT APPROACHES

Depending on the characteristics of the problem under review and the availability and form of data required, the annalist needs to decide upon the use of a qualitative, semi-quantitative or quantitative approach. All approaches have their own set of possible methods that can be used for each of the consecutive steps of the risk analysis.

Quantitative risk assessment

Quantitative Risk Analysis (QRA) is the determination of the probability and consequences of potential losses in numerical terms. The assignment of probability values to the various events in the risk model provides for a quantitative assessment of risk.

An important aspect of risk assessment is the estimation of the associated uncertainty. Therefore, the process may be completed through the use of statistical models such as probability analysis, Poisson distributions or Bayesian theory. These statistical models require the use of past data and assumptions about future trends. Much of the data may be accumulated from different sources.

Qualitative risk assessment

Although the bulk of the effort in developing methods of risk analysis has been addressed to quantitative methods, critical aspects of risk frequently require qualitative evaluation. Qualitative risk analysis may use “expert” opinion to estimate probability (or frequency) and consequence (or impacts) often through linguistic expressions. Based on expert judgement different qualitative consequence categories can be defined in terms of for example high, medium, low, etc. The same can be done for qualitative probability categories in terms of expressions as likely, may occur, not likely, very unlikely. This subjective approach may be sufficient to assess the risk of a system, depending on the decisions to be made and available resources. Formal processes for expert-opinion elicitation have been developed to provide consistency in qualitative information gathering (e.g. Delphi technique). Concerning qualitative uncertainty estimates, one has to rely on subjective estimates of uncertainty.

Conclusion

Within the quantitative and qualitative approaches a wide range of methods exist, each with its own characteristics, advantages and disadvantages and fields of application.

The selection of a quantitative or qualitative method depends upon the availability of data for evaluating the hazard and the level of analysis needed to make a confident decision. Quantitative risk assessment (QRA) is unambiguously defined as a “frequency x impact” and provides a more uniform understanding among different individuals than qualitative risk assessment. QRA is the most correct and practical approach and combines the advantages of various techniques. However, not all of the relevant risk sources and receptor specific aspects can be covered in quantitative terms and quality data essential for accurate results are not always available. In this sense, a less detailed analysis based on the use of qualitative analysis methods can be appropriate. A semi-quantitative approach, using quantitative methods if possible and qualitative methods if needed, is a pragmatic and often the most suitable approach.

ENVIRONMENTAL RISK ASSESSMENT

Dependant on which risks, receptors and end-points one wants to investigate, the different steps and methods to be used in ERA will differ. This illustrates that an adapted approach of ERA is required and that a general description of the key tasks and methods in risk analysis is not possible. There are however a number of unifying principles underlying all risk assessments. Despite the diversity of approaches, we can state that in general seven steps can be identified addressing the key questions in an Ecological Risk Assessment (ERA): Problem Formulation, Hazard Identification, Release Assessment, Exposure Assessment, Consequence or Effect Assessment, Risk Characterisation and Estimation and Risk Evaluation (Covello and Merkhofer 1993; Fairman et al., 1999).

Table 1. Seven steps of an ERA and associated key-questions (based on Fairman et al., 1999)

1. Problem Formulation	What needs to be assessed?
2. Hazard Identification	What can go wrong?
3. Release Assessment	How often or how likely?
4. Exposure Assessment	How does the released material reach the receptor, at which intensity, for how long and/or how frequent? How likely will the receptors be exposed to the released pollution?
5. Consequence or Effect Assessment	What is the effect on the receptors?
6. Risk Characterisation and Estimation	What are the risks (quantitative or qualitative measure)?
7. Risk Evaluation	How important is the risk to those affected, those who create it and those who control it?

The conclusions made in the Risk Characterisation and/or Risk Evaluation are used as input for Risk Management: Which actions should be taken and how should the remaining risks be handled? The different consecutive steps are presented in the following figure:

Presentation of the general key tasks in environmental risk assessment (Based on Fairman et al. (1999)

Whatever method chosen, two major topics that need to be taken into account throughout each consecutive step of the risk assessment are (1) uncertainty rating and (2) quality assessment of the input. Identification of potential gaps is also an important topic of risk assessment in order to assure the quality and relevance of the available information.

Problem Formulation

The problem formulation step is crucial in ERA. Initially the problem has to be defined and certain issues must be clear before the assessment starts (Fairman et al., 1999; Mac Donald et al., 2001):

• What are the risk sources we want to assess? Are these point sources (e.g. wind energy park) or mobile sources (e.g. maritime transport, fishing fleet) and what are the characteristics of these risk sources?
• Are we concerned with the production, use or disposal of the hazard? What are the environmental hazards to be taken into account: mineral oil, chemicals, garbage, sewage, ballast water, tributyltin, emissions, noise etc;
• Which are the pathways in which the created hazard can reach the receptor and which are the receptors and end-points?
• Will we focus on pre-defined sensitive ecosystems (e.g. area of special conservation under the Habitats Directive, EC Birds Directive or areas with a high value in recreational amenity or commercially exploitable biological resources) or do we cover the risks for a broader area?

At this stage, a generic model should be defined to describe the functions, features, characteristics and attributes of the system under investigation. Other questions that need to be handled in this first step are those related to legal and policy frameworks relevant to the risk assessment. Will we rely on regulatory standards and policy frameworks as a guide to determine "acceptable" risk and the significance of including specific end-points? Is there a legal framework that determines how we should approach the risk assessment? (Fairman et al., 1999)

Hazard Identification

The purpose of this step is to identify all of the conceivable and relevant hazards that could possibly cause harm to the receptor of interest. The identification may involve the establishment of those agents that may cause harm and working backwards to identify how this harm could occur. Alternatively, hazard identification may arise from examining all possible outcomes of routine operation and identifying the consequences from normal operation. (Fairman et al., 1999). The hazards identification step is closely linked to the next step, release assessment in the sense that these steps are both risk source related while the exposure and consequence steps are risk receptor related. Often, no distinction is made between hazard identification and release assessment, and is simply denominated as "hazard identification".

Release Assessment

The Release Assessment step involves the identification of the potential of the risk source to introduce hazardous agents into the environment. This may be descriptive or involve the quantification of the release. Release assessment attempts to give a measure of the likelihood of a release. It will include a description of the types, amounts, timings and probabilities of the release of hazards into the environment and a description of how these attributes might change as a result of various actions or events (Fairman et al., 1999). Release assessment is also risk source related and therefore often executed together with the hazard identification step.

In quantitative risk analysis (QRA), a quantitative estimation of the probability of release can be approached in two ways:

• The historical approach which uses direct statistical data on the system under investigation. This may be collected monitoring data or data from similar marine activities. This includes data on undesired events as well as data on recovery and control measures which mitigates the potential impacts.
• The approach which uses analytical and simulation techniques, breaking the system down into contributing factors and causes. Collected monitoring data or data from similar marine activities are also used to verify the modelling results.

Expert judgement can be used to estimate the likelihood or probability of a release of hazards in a non-quantitative way. Based on the results of the hazard identification, the likelihood is divided in different categories in terms of terms of expressions as likely, may occur, not likely, very unlikely (Wilcox et al., 2000).

Exposure Assessment

Exposure assessment attempts to quantify the potential exposure levels of the hazard at the receptor site. It includes a description of the intensity, frequency and duration of exposure through the various exposure media (routes of exposure) and the nature of the population exposed. Risk assessment on ecosystems has to deal with a multitude of organisms, all with varying sensitivities to chemicals and various groups have distinct exposure scenarios (e.g. free swimming species have another exposure pathway than benthonic species). The exposure assessment step requires the use of monitoring data, exposure modelling techniques and also mapping models to locate ecological sensitivity incorporating GIS techniques (Fairman et al., 1999; Ecotoc, 2001).

Most of the time, exposure of ecosystems to produced hazards is determined in terms of PEC or the Predicted Environmental Concentration. The PEC is calculated on both local and regional spatial scales from monitoring data where available (also called MEC or Monitored Environmental Concentration), or by using realistic worst-case scenarios. If this information is not available, estimates are made from exposure models. The PEC is calculated for each environmental compartment using the information available on release quantities and subsequent degradation processes in the "standard" environment. Site-specific information is used when available and appropriate. The relevant compartments of the marine environment are (Ecotoc, 2001):

• Water-exposure of aquatic organisms across respiratory and other permeable surfaces;
• Sediment-exposure of sediment dwelling (benthic) organisms by ingestion of, or direct contact with, sediment particles;
• Biota-exposure of higher trophic levels via the food chain (secondary poisoning), by predation on organisms that have been exposed via the water, sediment or predation on other organisms.
• Air-exposure for marine birds and mammals by inhalation of the chemical in the air they breath (likely less significant than the other three)

Consequence or Effect Assessment

A Consequence Assessment will examine the consequences of the release or production of the hazards, to the specified population and the quantification of the relationship between specified exposures to the hazard and the consequences of those exposures. The consequences examined in ecological systems are varied and few defined end-points exist at present. Environmental risk assessment on ecosystems is concerned with different populations and communities and the effects of substances on their mortality and fecundity (Fairman et al., 1999).

In ecological impact assessment, the consequences or effects can be estimated in terms of PNEC or the Predicted No Effect Concentration (based on EC Directive 93/67/EEC). Separate PNEC values need to be derived for the relevant compartments of interest: water compartment, benthic compartment (sediments) and biota (representing organisms which are eaten by avian and mammalian predators). PNEC values can be derived using ecotoxicity tests. In these tests, the estimation of the PNEC is primarily made on the basis of results from monospecies laboratory tests or, in some cases, from model ecosystem tests. The available ecotoxicity data are used to derive a No Observed Effect Concentration (NOEC) or a Lowest Observed Effect Concentration (LOEC). The test species used are selected to represent the sensitivities of different taxonomic groups in each environmental compartment. For aquatic effects assessments, ecotoxicity data are required on representatives of fish species, daphnia and algae (Fairman et al., 1999).

Assessment (safety) factors are applied to the toxicity value to enable extrapolation from laboratory experiments to the field, acute to chronic effects and for inter and intra species variations. The size of the assessment factor varies according to the number and type of data available and the likely duration of exposure. (Fairman et al., 1999; Ecotoc, 2001). Ecotoxicological Assessment Criteria (EACs) are defined as effects benchmarks against which the results of environmental monitoring can be assessed in an attempt to identify possible areas of concern. The determination of EACs is based on the same principles as for the assessment factors. EACs are only derived when data which meet predefined quality criteria are available from at least three species.

Expert judgement may also be used to assess the magnitude of the consequences in qualitative terms. Dependent on the pollution source and ecosystem characteristics, the potential consequences on the ecosystem are divided in different categories (e.g. “minor” to “catastrophic”). (Wilcox et al., 2000)

Risk Characterisation and Estimation

Risk characterisation consists of integrating the results from the release assessment, exposure assessment and the consequence assessment to produce measures of environmental risks. This may include an estimate of the numbers of measures indicating environmental damage, and the uncertainty involved in these estimates (Fairman et al., 1999).

In the risk characterisation as described above, PEC incorporates the results of the release and the exposure assessment step while PNEC incorporates the results of the consequence assessment step. Current risk assessment practice compares the PEC with the PNEC for the relevant ecosystem using data from representative species. Implicit in this approach is the assumption that there is a tolerable threshold of any chemical substance in the environment (via the PNEC). An element of precaution is built into the approach via the use of conservative/worse-case assumptions within exposure and effects assessments (Ecotoc, 2001).

The EU practice on risk characterisation involves the calculation of a quotient – the PEC/PNEC ratio. This PEC/PNEC ratio should be calculated for all relevant endpoints. If the PEC/PNEC is less than 1, the substance of concern is considered to present no risk to the environment and there is no need for further testing or risk reduction measures. If the ratio cannot be reduced to below 1 by refinement of the ratio (by gathering of further information and further testing), risk reduction measures are necessary (Fairman et al., 1999).

The PEC/PNEC ratio risk characterisation method does not allow us to assess the effective risk expressed in e.g. terms of number of affected individuals or reduced population density in a specific region resulting from a particular activity. An overall estimation of risk can be defined as the multiplication of the consequence for each damage-causing event with the frequency of that event. The frequency of an event is a result of the hazard identification and release step (e.g. frequency of collisions, powered grounding, etc. within a particular area). The consequence of a damage-causing event is usually defined as casualty probabilities. This is presented in the PECs (e.g. amount of fuel oil spilled due to collisions at the receptor site), taking into account the relevant PNECs representing the thresholds below which no damage exists for the investigated species (e.g. no effect concentrations of fuel oil in the different relevant marine ecosystem compartments for seagulls). The population of the species under investigation (e.g. seagulls) present in the areas covered by each probability band is multiplied by the appropriate casualty probability producing the total number of the population predicted to be affected by each event. When combined with the frequency for each event, a risk estimate can be produced for this specific species. This process can be repeated for a number of key species in order to have an overall idea about the risks for the whole ecosystem.

Although a quantitative risk assessment approach is preferred, there may be cases where this can not be carried out (e.g. no PEC or PNEC can be properly calculated). Qualitative risk assessment can be used as an alternative. In this case, the risk characterisation shall entail a qualitative evaluation of the likelihood that an effect will occur under the expected conditions of exposure. The results of the qualitative risk characterisation can be used as a base to prioritise risk reduction measures.

Risk Evaluation

Risk Evaluation is the examination of what the characterised risks actually means in practice. What is the significance or value of the identified hazards and estimated risks? Risk evaluation deals with the trade-off between the perceived risks and benefits. This includes acknowledgement of the public perception of the risk and the influence that this will have on the acceptability of risk and risk decisions. On its turn, the public perception of risk depends on the economic, social, legal and political context in which the affected and/or concerned population lives (see Figure 3 1) (Fairman et al., 1999).

The risk evaluation may take account of these perceived risks and benefits and incorporate them in the final risk assessment. The results from this risk evaluation may serve as an input to the risk management process. Based on the acceptable level of risk eventual choices of action are determined needed to achieve the desired level of risk. If a system has a risk value above the risk acceptance level, actions should be taken to address concerned risks and to improve the system though risk reduction measures.

The three major approaches to evaluate risks are:

• Professional judgement: technical experts most knowledgeable in their fields examine the risks and make conclusions based on ‘best judgement’. Expert judgement may be used to estimate probability (step 3 and 4, see 1.3.2 and 1.3.3) and consequence (step 5, see 1.3.5). Based on a ranking of the probability and consequences of the concerned risk, experts may defineacceptance levels.
• Formal analysis: Cost-benefit, cost-risk-benefit and decision analysis are the most common of formal analysis techniques for alternative risk management options. In cost benefit analysis and cost-risk-benefit analysis, benefits (e.g. avoided pollution, risk) and costs (cost of pollution reduction or risk reduction measures) associated with a particular risk management option are evaluated against each other. Decision analysis is an axiomatic theory for making choices in uncertain conditions.
• Bootstrapping: Bootstrapping approaches identify and continue policies that have evolved over time. It is argued that society achieves a reasonable balance between risks and benefits only through experience. The safety levels achieved with old risks provide the best guide as to how to manage new risks.

Professional judgement is a qualitative approach, while formal analysis and bootstrapping are both defined as quantitative approaches. For each of these approaches different methods exist.

UNCERTAINTY

Uncertainty is inherent to all risk assessments. It is important to assess the magnitude of the uncertainty to determine the "relevance" of the quantified risk. Risks associated with a specific risk source and receptor and under pre-specified surrounding conditions will be expressed in terms of a range (with a lower and upper bound) rather than a single figure. The best estimate of risk is situated between the upper and lower bound. Comparing the magnitude of this range with the best estimate gives an idea about its relevance or value. Knowing the uncertainty is also important to ensure that the input of the results into the risk evaluation step is realistic (i.e. using cost benefit analysis methods) and thus to ensure that appropriate risk management decisions are made (Wilcox et al., 2000; MCA, 2003).

Sources of uncertainty

There are several potential sources of uncertainty. These include:

• Uncertainty inherent to the used methods in each of the ERA steps (e.g. choice of model, assumptions made in used models);
• Uncertainty related to the collected data and parameters (e.g. gaps in historic/recent data, use of data from other situations and extrapolations to fill out gaps);
• Idiosyncrasies of the analyst: interpretation of ambiguous or incomplete information , human error;
• Uncertainty about the future (e.g. improved techniques and management to prevent and control pollution: improved ship structure, training of crew, adaptation of shipment routes according to pollution sensitivity areas, improved emergency plans, etc.).

The applicability of historical data to the current situation

Over a period of time there are likely to be changes to the risks associated with a system. This might be due to older equipment being replaced by modern items, degradation of existing equipment and structures, changes in management systems, changes in operating conditions, etc. These will tend to move the actual risk levels away from the average historical levels, so that the present-day risk is different from the risk used as a basis for calculation. The net result is often a lowering of the risk over a period of time. However such changes are usually very slow to occur and often have a minimal impact on accident statistics. In the shipping industry in particular there is unlikely to be a sudden step-change in overall risk levels as vessels are likely to trade for over 20 years and practices evolve rather than being replaced by entirely novel methods. It is thus expected that this will have a small impact on the uncertainty inherent in the analysis (MCA, 2003).

Uncertainty in the completeness of the data

Hazard identification

It is extremely unlikely that every accident will be reported. This will lead to an historical risk level that is lower than the risk in reality. This is expected to be the major cause of uncertainty in the estimation of the base case risk levels. The shipping industry is very diverse, and there is no central body to which all accidents must be reported. However, there are a number of organisations which do collect shipping accident data and it is very likely that major accidents, particularly those involving loss of life, or major pollution will be known by those organisations. It is thus expected that, whilst there will be some uncertainty in the results, the high risk areas will have been adequately identified (MCA, 2003).

Exposure and Consequence assessment

The consequence and exposure steps are one of the most important areas in which completeness of data are problematic. An example is the need of extrapolation from laboratory experiments to the field, acute to chronic effects and for inter and intra species variations because of lacking data, especially in risks assessment in marine environments. These extrapolations entail additional uncertainty which is dealt with by the introduction of assessment or safety factors.

Methods to assess uncertainty

Quantifying all sources of uncertainty is difficult (especially idiosyncrasies of the analyst). Methods for estimating the uncertainty are for example statistical analysis (for uncertainty related to data and parameters and models), expert judgement (for uncertainty related to models) and sensitivity analysis (for uncertainty related to future trends). Uncertainty should be assessed for each of the ERA steps. When passing on results to other steps in the methodology, it is important that the uncertainty bounds are passed also, along with information on the key areas of uncertainty and what effect they might have on the risk levels.

References

See also References for environmental risk assessment.