Wednesday, 31 January 2024: 2:00 PM
327 (The Baltimore Convention Center)
Kim Klockow McClain, NCEP/UCAR, Norman, OK
In recent years, many studies have noted the positive benefits of forecast uncertainty information to support informed decision-making. A recent effort provided a high level summary of this research, alongside recommendations from this literature about the stylistic presentation formats that seem to do best. Importantly, the works reviewed in that study all had a few features in common: namely, they specified the decision scenario and the reference class of the probabilistic information. In this way, they constrained the decision problems and distilled the complex reality of the forecast information, often eliminating the spatio-temporal complexity of the decision problem that exists in the real world. This could carry significant implications for the application of the recommendations from this body of work.
Picking up from that work, this presentation proposes a taxonomy of forecast uncertainty information and its presentation attributes. Because of the spatio-temporal context of the information, the presentation argues that stylistic considerations, like the format of a probability, should be considered at the end of a process rather than being seen as an end unto itself. Starting with the perspective of the decision-maker, several other issues take precedence and should be decided before generating a probability: the reference class, the reference frame, and the update frequency. Once these are specified, then designs can be optimized. The presentation will conclude with an outline of the items a forecaster should think about as they craft probabilistic messages, especially across the lifecycle of an event.

- Indicates paper has been withdrawn from meeting

- Indicates an Award Winner