Principle 1

Good decision-making is best supported by scientific information deemed adequate for purpose. Where today's best available information is not adequate, this must be made obvious.


For a given decision, some information will be relevant and adequate for the purpose of informing that decision. A single forecast may be adequate for some purposes and not others. For instance, if today is Wednesday and I wish to decide whether to hold a party this weekend on Saturday or Sunday, the weather forecast will be a useful input. But if I am setting a date for a party in three months’ time, and wish to decide between the Saturday and Sunday of a given weekend, then even today’s best available weather forecast is not adequate for this purpose. Decision-makers would benefit if it were made clearer where today’s best available information is thought to be, or not be, adequate for different kinds of decision support.  


How can we assess adequacy for a given purpose?  In the example of a weather forecast, this is reasonably easy: we can look at a large sample of previous weather forecasts for a similar situation (location, lead time, type of weather) and, with reference to observations of the actual weather, generate a statistical description of the forecast quality. This can then to be used to decide, for each purpose, whether this weather forecast is adequate as a source of information, or not.

This relies on the existence of a relevant archive of previous forecasts and outcomes. Where we are forecasting climate, firstly we may have fewer or no previous forecasts to assess, and secondly, we are not sure whether skill in the past will be directly related to skill in the future, since the underlying climate is itself known to be changing.

References and further reading

Parker, W. S. (2009, June). Confirmation and adequacy‐for‐purpose in climate modelling. In Aristotelian Society Supplementary Volume (Vol. 83, No. 1, pp. 233-249). Oxford, UK.

Smith, L. A. (2002). What might we learn from climate forecasts? Proceedings of the National Academy of Sciences, 99 (suppl 1), 2487-2492.

Leave a comment


Comments on Principle 1

Reason Machete - P01-0719
The example given in the WHY is a weather example. I think it would be better to contrast two climate scenarios. For instance, information required to decide which crop to farm next season in a given region compared with deciding whether to build a dam today based on decadal projections. Of course the weather example illustrates the point. 

Dewi Le Bars - P01-0725
The weather forecast example is a nice illustration. Unfortunately, in practice in climate science most of the time there is no clear cut between information that is adequate or not adequate for purpose. Making that decision is difficult and requires a dialogue between “expert users” and climate scientists. An example for sea level: Hinkel, J., Church, J. A., Gregory, J. M., Lambert, E., Le Cozannet, G., Lowe, J., … Wal, R. (2019). Meeting User Needs for Sea Level Rise Information: A Decision Analysis Perspective. Earth’s Future, 7(3), 320–337.