OR/14/042 In the future most models will at some point in their life-cycle need to be linked to other models. What needs to be achieved in order for this to happen
Standards are critically important to enable integrated modelling to happen. Linked to this, is a need to motivate and incentivise researchers to adopt standards during the modelling process. The reasons given for this were:
- For all models, researchers need to be able capture the scientific accuracy of the final model results
- There is a danger that by re-using model components; users can make inappropriate links and be using model components beyond the design limitations
- Quantification and assessment of uncertainty, necessitates the use of a standard set of metadata to describe the model, its inputs and outputs and assumptions used
- Need to be able to bench-mark model performance
- Need to encourage multiple agencies to use the models whose development is funded by NERC workflows and frameworks
- Once we have a defined set of standards linking models becomes much easier
It was clearly recognised that we need a ‘Taxonomy’ of standards which would allow researchers to select and adopt those standards that best suited the modelling work being undertaken. There is a need for a minimum set of standards which must be light touch and simple to implement and standard for all components within a linked modelled system.
To be included within an IEM standard:
- A clear ontology defining types, properties and inter-relationships of the modelling processes and data manipulation undertaken, i.e. what has been inputted and outputted by the models and what operations (modelled processes) have been performed.
- The necessary information required to understand a model, such as a description of the models and the assumptions used, and the parameter types that the model can link to. This will enable future users to more easily understand the scientific accuracy that the final model can produce.
Another element was, understanding what would be gained from linking models together. Exemplars should therefore examine the benefits and time required for linking and we should clearly be able to demonstrate the advantage in dynamic linking. Technology platforms that will be generated need to be appropriate to the scale of the problem we are looking at.
There will be a need to generate a critical mass of linkable components (both models and the data used to serve them) in order to gain momentum. By delivering the infrastructure development through high profile exemplar projects, it should be possible to generate the momentum required.
We should incentivise by creating a culture where the individual contributions of researchers are rewarded i.e. citations given to modellers when their models are re-used and at a corporate level; work with RCUK/NERC to encourage the use of a limited number of standards and frameworks.
IPR and Corporate Liability will need to be addressed, to ensure models can be made available to be linked. For the IEM community to grow, an open source and open access ethos will need to be adopted wherever possible. The IPR value will be in being the expert in your model component and data, not actually in the possession of it. If models are linked together to provide solutions to decisions-makers, who is it that assumes liability if something goes wrong?
There are significant skill shortages in staff with expertise to manage linking technologies, fitting models to standards, assisting with metadata capture and developing automated metadata capture systems.