2 days in a room with facilitated post-it notes and people from Defra, the EA, Universities and elsewhere to discuss the (still potential) modelling program to be associated with the Defra DTC (Demonstration Test Catchments) programme. One day on the needs and relevance of modelling the impacts of mitigation measures to the DTC, and one day on what might actually need to be done. The discussions ranged more widely than the DTC project alone, and (somewhat frustratingly) did not get as far as the details of what might be included in a DTC modelling framework (to be worked out by a smaller working group). However, one of the final small group discussions was assigned to look at some Guiding Principles for modelling in this context. In bullet point form, these were (as scribbled down quickly today as the group reported):
- Models should represent what matters
- Models should contribute to the solution, they are not a solution themselves
- Models should be framed by the question to be asked (including level of certainty of required)
- Models should be adaptive learning tools for interaction with stakeholders as a way of developing understanding and trust
- Modelling culture should be an important component of decision making
- Model contributions should aid no regrets decisions
- The ethical dimension of modelling should require the involvement of stakeholders
- Uncertainty is not a dirty word
- Standard inputs and outputs should facilitate model integration and coupling of models
- Everyone can be wrong from modeller to decision maker to stakeholder.
These Guiding Principles gloss over some very difficult issues (such as what level of complexity is implied by “representing what matters”, or how to deal with epistemic errors) but do reflect the emphasis in the discussions on stakeholder involvement in the modelling process and the impact of model uncertainties in the decision making and policy setting process; both of which hold regardless of what particular model structures are considered as components of the DTC modelling framework.
I thought that the outcome was of interest to the CCN, the concept of Guidelines for Good Practice that underlies CCN was in part developed to address some of the issues raised by these principles (and CCN was already quite well represented at the meeting). Anyone want to comment on these or add further important principles that should be included……principles of model evaluation are not really included for example unless considered to be implicit in the adaptive learning process.
I am sure there will be more to report on the DTC modelling component, and its links to the NERC Virtual Observatory project in the coming months.
