Communicating Hazard Risk as Possibility, not Probability

Decision-making is difficult under high uncertainty. Further, uncertainty is difficult to communicate. Typically, meteorologists express likelihood of event occurrence via probabilities. However, there is a difference between a 50% risk stemming from total ignorance (e.g., risk of unprecedented snowfall) and 50% drawn confidently from a reliable ensemble (Monte Carlo) prediction system simulating the given extreme event in half the ensemble’s members. It is important, therefore, to understand the nature of uncertainty when predicting cold-pool formation in the Uinta Basin, Utah, where winter ozone concentrations may reach unhealthy levels after heavy snow.

Thus, we turn to possibility theory, based on fuzzy logic, whereby each variable’s lexical category (e.g., “high snowfall”) has a possibility: an event’s likelihood that is supported in the data. Possibility is a dual measure, and its counterpart is necessity. This is the certainty of an event occurring due to alternatives having a low possibility (the given event must happen). Because the possibility distribution may not sum to 1, as they must in probability theory, the paradigm gives a deeper insight into risk analysis drawn from support in data (“known unknowns”) versus total ignorance of the likelihood of a black swan event (a “failure of imagination” that unprecedented snowfall will occur). Interpreted as an upper and lower bound on likelihood, a possibility paradigm provides more information to risk-averse and risk-tolerant users, respectively.

While the theory of possibility is unfamiliar to many meteorologists and end-users alike, its initial deployment in 2024 will be interpreted by forecasters supporting the Bingham Research Center’s Ozone Alert program that communicates risk to stakeholders sensitive to high-ozone episodes. We show various visualisations that aim to provide simpler (but richer) information to the layperson. We further propose use of a large language model (LLM) fine-tuned to interpret this concept of possibility and discuss with a decision-maker at an appropriate level of complexity regarding the potential extreme event.