OR/16/022 Recommendations for future work

Beriro, D, Cave, M, Wragg, J and Hughes, A. 2016. Agent Based Modelling: Initial assessment for use on soil bioaccessibility. British Geological Survey Internal Report, OR/16/022.

Summary and next steps

By exploring the examples to a) understand how they work and b) attempt to answer the question ‘Is Agent Based Modelling (ABM) useful for extending the work of bioaccessibility and environmental interaction?’ then the following conclusion are reached:

1. Ease of use: Insight-maker > NetLogo > Repast-simphony
2. Flexibility: Repast-simphony > NetLogo > Insight-maker

However what has become apparent is that there are two clear favourites for further use: repast- simphony and NetLogo. Both offer the flexibility to undertake complex problems with ABM.

It is clear that by the introduction of agents as representatives of real-world entities it can make the assessment of human health assessment potentially much easier. There are significant benefits in the conceptual understanding of the modelling and how this can be exploited to explain the process to end-users. Examples of Local Authorities (LAs) would benefit from modelling of contaminated land problems with ABM.

Next steps

The work packages planned for FY 16/17 are as follows:

1. Literature review to determine gaps in applying ABM to human health effects of exposure to contaminated land
2. Identify and characterise groups doing ABM including which software they use e.g.

UCL

Warwick

Lancaster

Risk Solutions Ltd

Other groups — who are they

3. Identify the use of external groups as mentors as well as collaborators
4. Do simple Northampton example (houses, roads and parks along with a range of agents — children of different ages and adults)
5. Build on human exposure matrix initiated as part of the work reported here
6. Produce database/spreadsheet of available environmental and exposure data
7. Improve Java knowledge within the group before next sprint
8. Undertake sprints early in next FY (June/July) to capitalise on current momentum
9. Identify potential recipients of new BGS skills and ABM problem solving e.g. Université de Lorraine (Nancy, Fr)

Details of the example to be developed

The ‘real-world’ ABM model should be based on a geographical context relying on attributed and gridded raster data representative of the local environment of Northampton (or similar area). External data are expected to include:

• Basemap e.g. OS Open Map

o  Buildings

o  Public open space

o  Roads

o  Gardens

• Total and bioaccessible As concentrations
• Soil type based on parent material maps or superficial deposits (e.g. clay, sand, loam)
• Organic carbon content of soils

Some assistance from a BGS GIS specialist is expected to be required to assist with the creation of the data layers. The attributed geographical data will be used to reflect environmental constraints on the movement and flow of people.

The agents used in the model are expect to include children of different age classes and adults of the same age class. These age classes should be derived from the Environment Agency Contaminated Land Exposure Assessment (CLEA) model used to estimate chronic exposure to PHS in soil in humans. Doing so will help to ensure that the outputs are in a language that risk- based land management and land-use planning professionals can understand.

The behaviour of the agents should also be derived from CLEA. For the ingestion pathway this would include the amount of soil ingested per day based on a given number of hours spend within a particular land-use scenario e.g. public open space. Additional data sources may also be called on that are not generic e.g. actual number of visits to a park per year.

The aim of the model is to test whether ABM can simulate the exposure to As in humans from soil as well as the industry standard deterministic model (CLEA). If the answer is ‘yes’ then ‘what if’ scenarios would be explored e.g. what if a child only plays in one park or another, or in a park for 50% of the time and in a garden for 50% of the time. The results would then be used to explore land-use planning scenarios e.g. developing park #1 and extending park #2 or where the most suitable location is for park #3.

The potential results are believed to be innovative and relevant to not only to regulators e.g. planners and contaminated land officers, but also consultant business advising their clients on how a particular area of land might be used. The ABM would also be a beneficial way of exploring how existing BGS dataset, products and applied science might be made to be more impactful. Outputs of this work are expected include publication in a high impact journal and an application for funding under a relevant scheme e.g. Innovate UK, NERC Innovation Projects or commercially funded research. Links with existing project such as NERC KE Fellowships and other national initiative such as the Future Cities Catapult will be explored.