Trevor Page

Position Paper: The role of expert opinion in environmental modelling

The inevitable though frequently informal use of expert opinion in modelling, the increasing number of models that incorporate formally expert opinion from a diverse range of experience and stakeholders, arguments for participatory modelling and analytic-deliberative-adaptive approaches to managing complex environmental problems, and an expanding but uneven literature prompt this critical review and analysis. Aims are to propose common definitions, identify and categorise existing concepts and practice, and provide a frame of reference and guidance for future environmental modelling. The extensive literature review and classification conducted demonstrate that a broad and inclusive definition of experts and expert opinion is both required and part of current practice. Thus an expert can be anyone with relevant and extensive or in-depth experience in relation to a topic of interest. The literature review also exposes informal model assumptions and modeller subjectivity, examines in detail the formal uses of expert opinion and expert systems, and critically analyses the main concepts of, and issues arising in, expert elicitation and the modelling of associated uncertainty. It is noted that model scrutiny and use of expert opinion in modelling will benefit from formal, systematic and transparent procedures that include as wide a range of stakeholders as possible. Enhanced awareness and utilisation of expert opinion is required for modelling that meets the informational needs of deliberative fora. These conclusions in no way diminish the importance of conventional science and scientific opinion but recognise the need for a paradigmatic shift from traditional ideals of unbiased and impartial experts towards unbiased processes of expert contestation and a plurality of expertise and eventually models. Priority must be given to the quality of the enquiry for those responsible for environmental management and policy formulation, and this review emphasises the role for science to maintain and enhance the rigour and formality of the information that informs decision making.

Investigating the uncertainty in predicting responses to atmospheric deposition using the model of acidification of groundwater in catchments (MAGIC) within a generalised likelihood uncertainty estimation (GLUE) framework

This study investigates the uncertainty associated with the modelled response of a catchment to historic and predicted future acidic deposition for the period 1851–2041. The MAGICmodel is applied within a GLUE framework to the 3.88 km2 Afon Gwy catchment at Plynlimon, Wales. Nine million Monte Carlo simulations resulted in 5700 being accepted as behaviouralas defined by a fuzzy measure comparing observed to simulated variables. Model output and parameter sensitivity analysis indicate that, for this example where weathering rates are low,model dynamics are limited compared to control exerted by modelinitial conditions and by the specified acidic deposition boundary conditions. The results show that despite the small number of behavioural simulations, they are widely spread acrossthe ranges for most of the parameters varied. The GLUE methodology allows simulated prediction ranges for important variables to be presented as quantitative likelihood weighteduncertainty estimates rather than a single prediction for eachvariable over time.

Re-shaping models of E. coli population dynamics in livestock faeces: Increased bacterial risk to humans?

Dung-pats excreted directly on pasture from grazing animals can contribute a significant burden of faecal microbes to agricultural land. The aim of this study was to use a combined field and modelling approach to determine the importance of Escherichia coli growth in dung-pats when predicting faecal bacteria accumulation on grazed grassland. To do this an empirical model was developed to predict the dynamics of an E. coli reservoir within 1ha plots each grazed by four beef steers for six months. Published first-order die-off coefficients were used within the model to describe the expected decline of E. coli in dung-pats. Modelled estimates using first-order kinetics led to an underestimation of the observed E. coli land reservoir, when using site-specific die-off coefficients. A simultaneous experiment determined the die-off profiles of E. coli within fresh faeces of beef cattle under field relevant conditions and suggested that faecal bacteria may experience growth and re-growth in the period post defecation when exposed to a complex interaction of environmental drivers such as variable temperature, UV radiation and moisture levels. This growth phase in dung-pats is not accounted for in models based on first-order die-off coefficients. When the model was amended to incorporate the growth of E. coli, equivalent to that observed in the field study, the prediction of the E. coli reservoir was improved with respect to the observed data and produced a previously unquantified step-change improvement in model predictions of the accumulation of these faecal bacteria on grasslands. Results from this study suggest that the use of first-order kinetic equations for determining land-based reservoirs of faecal bacteria should be approached with caution and greater emphasis placed on accounting for actual survival patterns observed under field relevant conditions.

Predicting microbial water quality with models : over-arching questions for managing risk in agricultural catchments.

The application of models to predict concentrations of faecal indicator organisms (FIOs) in environmental systems plays an important role for guiding decision-making associated with the management of microbial water quality. In recent years there has been an increasing demand by policy-makers for models to help inform FIO dynamics in order to prioritise efforts for environmental and human-health protection. However, given the limited evidence-base on which FIO models are built relative to other agricultural pollutants (e.g. nutrients) it is imperative that the end-user expectations of FIO models are appropriately managed. In response, this commentary highlights four over-arching questions associated with: (i) model purpose; (ii) modelling approach; (iii) data availability; and (iv) model application, that must be considered as part of good practice prior to the deployment of any modelling approach to predict FIO behaviour in catchment systems. A series of short and longer-term research priorities are proposed in response to these questions in order to promote better model deployment in the field of catchment microbial dynamics.

Development and testing of a risk indexing framework to determine field-scale critical source areas of faecal bacteria on grassland

This paper draws on lessons from a UK case study in the management of diffuse microbial pollution from grassland farm systems in the Taw catchment, southwest England. We report on the development and preliminary testing of a field-scale faecal indicator organism risk indexing tool (FIORIT). This tool aims to prioritise those fields most vulnerable in terms of their risk of contributing FIOs to water. FIORIT risk indices were related to recorded microbial water quality parameters (faecal coliforms [FC] and intestinal enterococci [IE]) to provide a concurrent on-farm evaluation of the tool. There was a significant upward trend in Log[FC] and Log[IE] values with FIORIT risk score classification (r^2 = 0.87 and 0.70, respectively and P < 0.01 for both FIOs). The FIORIT was then applied to 162 representative grassland fields through different seasons for ten farms in the case study catchment to determine the distribution of on-farm spatial and temporal risk. The high risk fields made up only a small proportion (1%, 2%, 2% and 3% for winter, spring, summer and autumn, respectively) of the total number of fields assessed (and less than 10% of the total area), but the likelihood of the hydrological connection of high FIO source areas to receiving watercourses makes them a priority for mitigation efforts. The FIORIT provides a preliminary and evolving mechanism through which we can combine risk assessment with risk communication to end-users and provides a framework for prioritising future empirical research. Continued testing of FIORIT across different geographical areas under both low and high flow conditions is now needed to initiate its long-term development into a robust indexing tool.

Determining E. coli burden on pasture in a headwater catchment: Combined field and modelling approach

Empirical monitoring studies of catchment-scale Escherichia coli burden to land from agriculture are scarce. This is not surprising given the complexity associated with the temporal and spatial heterogeneity in the excretion of livestock faecal deposits and variability in microbial content of faeces. However, such information is needed to appreciate better how land management and landscape features impact on water quality draining agricultural landscapes. The aim of this study was to develop and test a field-based protocol for determining the burden of E. coli in a small headwater catchment in the UK. Predictions of E. coli burden using an empirical model based on previous best estimates of excretion and shedding rates were also evaluated against observed data. The results indicated that an empirical model utilising key parameters was able to satisfactorily predict E. coli burden on pasture most of the time, with 89% of observed values falling within the minimum and maximum range of predicted values. In particular, the overall temporal pattern of E. coli burden on pasture is captured by the model. The observed and predicted values recorded a disagreement of >1 order of magnitude on only one of the nine sampling dates throughout an annual period. While a first approximation of E. coli burden to land, this field-based protocol represents one of the first comprehensive approaches for providing a real estimate of a dynamic microbial reservoir at the headwater catchment scale and highlights the utility of a simple dynamic empirical model for a more economical prediction of catchment-scale E. coli burden.

Assessment of uncertainties in a long range atmospheric transport model: Methodology, application and implications in a UK context

Acid deposition models are inherently simplified representations of real world behaviour and their performance is best evaluated by comparison with observations. National and international acid rain policy assessments handle observed and modelled deposition fields in different ways. Here, both the observed and modelled deposition fields are seen as uncertain and the Generalised Likelihood Uncertainty Estimation (GLUE) framework is used to choose acceptable sets of model input parameters that minimise the differences between them. These acceptable sets of model parameters are then used to estimate deposition budgets to the UK and to provide a probabilistic treatment of excess deposition over environmental quality standards (critical loads).

A guide to expert opinion in environmental modelling and management

1364-8152/

Effects of seasonal meteorological variables on E. coli persistence in livestock faeces and implications for environmental and human health

– see front matter 2012 Elsevier Ltd. doi:10.1016/j.envsoft.2012.01.006 Compilation of this thematic issue has been prompted by observing an emerging field of professional practice, well supported by theoretical development and review. There are an increasing number of models for environmental assessment and management that formally incorporate and rely on expert opinion sourced from a diverse range of experience and stakeholders. Development and application of thesemodels is also increasingly setwithin the context of participatory, analytic-deliberative and adaptive approaches to managing complex environmental problems. As such applications grow in number, and in acceptance and influence, it is timely to draw together in this specialist journal selected leading examples of theoretical and applied best practice. Multiand trans-disciplinary working is inherent in the examples included here. We present papers offering a range of relevant perspectives and methodologies: from environmental sciences, sociology and political science, philosophy of science, and statistics and mathematics. Some of these perspectives may be unfamiliar to the regular reader of Environmental Modelling & Software, but we believe this diversity of contributions is important for a full understanding of the practice of environmental modelling and its strengths and limitations. In particular, this issue seeks to explore how to make best use of all existing knowledge in responding to the environmental challenges we face as society. Although by no way exclusive, we contend that the methodologies and disciplines showcased in this collection of papers will enhance the analysis and use of expert opinion in models. A starting point is the position paper by Krueger et al. (2012) which offers a frame of reference for the different roles of expert opinion in environmentalmodels. The paper exposes informalmodel assumptions and subjective choices made by modellers as forms of expert opinion, and recognises the increasing inclusion of expertise from beyond the conventional domains of science and technology in contemporary modelling studies. It is demonstrated that a broad and inclusive definition of experts and expert opinion is both required and part of current practice. The review and definition of

Adaptive forecasting of phytoplankton communities

Agriculture contributes significant volumes of livestock faeces to land. Understanding how faecal microbes respond to shifts in meteorological patterns of contrasting seasons is important in order to gauge how environmental (and human health) risks may alter under a changing climate. The aim of this study was to: (i) quantify the temporal pattern of E. coli growth within dairy faeces post defecation; and (ii) derive E. coli seasonal population change profiles associated with contrasting environmental drivers. Evaluation of the die-off dynamics of E. coli revealed that a treatment mimicking drought and warming conditions significantly enhanced persistence relative to E. coli in faeces that were exposed to field conditions, and that this pattern was consistent across consecutive years. The internal temperature of faeces was important in driving the rate of change in the E. coli population in the immediate period post defecation, with most E. coli activity (as either die-off or growth) occurring at low dry matter content. This study highlighted that the use of seasonal E. coli persistence profiles should be approached with caution when modelling environmental and human health risks given the increased likelihood of atypical seasonal meteorological variables impacting on E. coli growth and die-off.