Presentation is loading. Please wait.

Presentation is loading. Please wait.

Why do models fail? Problems, problems …

Similar presentations


Presentation on theme: "Why do models fail? Problems, problems …"— Presentation transcript:

1 Why do models fail? Problems, problems …
John Rees BGS Head of Policy and Science Co-ordination Andrew Hughes BGS Groundwater Modeller and NMPI Co-organiser

2 Ideal vs reality

3 Generic issues. as seen from a government institute
Generic issues as seen from a government institute - the British Geological Survey

4 BGS and numerical modelling
Groundwater - Model Developer and user Coastal change - Poacher and gamekeeper Contamination - often the ‘honest broker’

5 Typical reasons for failure
Lack of consistency of approach Turnover and education of staff in client organisations Language between user and modellers, and between modellers Appropriate and sufficient data Trust between parties Unrealistic expectations Honesty about the limitations of models Cost of taking-up new developments

6 \ Policy makers \ Modellers \ Resource Managers

7 Key organisations Universities Consultants Government Institutes
Regulators Utilities Government Ministries Pan-governmental

8 Regulator issues Limited number of staff that have modelling skills
Accessibility of models is often poor – they are often not easy to run The maintenance and updating of models has a high cost (the leaky roof syndrome) Inadequate data to support development or understanding Expectations are often not met Stakeholders are not adequately consulted

9 Resource manager issues
Common uncertainty about the specification and scale of model required Lack of clarity about the needs of regulators Find that models have too much uncertainty for detailed use Inadequate timely stakeholder involvement Find that models are not as flexible as managers would like

10 Model developer issues
Models are driven by policy and not the other way round e.g. Habitats Directive Few examples of models being trusted by all parties Stakeholders are recognised as important, but involvement is very variable Personalities are important in defining how much stakeholders interact Managing expectations is very important Those consultants who deliver to spec and on time may not be the best to drive modelling forwards

11 Broader implications Lack of efficiency associated with development encourages adoption or tweaking of older ‘industry standard’ models instead of adoption of newer models. Drive to conservatism encourages usage of ‘tried-and-tested’ consultants who focus on delivery, rather than more innovative scientists who will introduce new concepts and stretching the modelling. Acceptance of limitations (e.g. empirical constants, black-boxes) instead of driving better modelling.

12 Towards solutions Need common understanding before attempting to formulate solutions - Climate change models are not taken up by some national governments that do not accept reality! Language – differs markedly between disciplines Definition of problem type, organisations involved and geographic extent Problem is potentially huge, so need boundaries Guidelines are needed

13 Summary

14 Questions ?


Download ppt "Why do models fail? Problems, problems …"

Similar presentations


Ads by Google