How Likely is that Chance of Thunderstorms? A Study of the National Weather Service use of Words of Estimative Probability

One of the most challenging aspects of weather forecasting is effectively communicating forecast information to the public. No forecast is ever completely certain, and no meteorological phenomenon is guaranteed to occur. As such, the uncertainty in forecast information should be communicated in a way that makes sense to end users. Previous studies of the communication of probabilistic information suggest that, while the general public are more apt to communicate uncertainty with Words of Estimative Probability (WEPs), they prefer to receive that information numerically. Other work has suggested that a combination of numbers and WEPs is the best method for communicating probability, but little has been done to assess the communication and interpretation of probabilistic, meteorological information. In this work, we code 8900 tweets from the National Weather Service (NWS) Weather Forecast Offices (WFOs) and analyze them to find how they communicate probabilistic forecast information to the public via Twitter. This analysis reveals that WFO messaging is dominated by the use of WEPs, with few numerical descriptions of probability. These WEPs are generally unqualified, which could further impede the public’s ability to determine whether probability or uncertainty information is being communicated. Very rarely are probabilities expressed using both numbers and words, indicating a clear difference between recommended best practices from past work with current NWS methods of expressing probability. Based on this analysis, two publically fielded surveys are also analyzed in order to understand how participants tend to interpret such qualified and unqualified WEPs. Findings suggest that people tend to interpret qualified WEPs more concisely than the unqualified ones. However, both categories experience a wide range of interpretations, implying that WEPs communicate relatively vague notions of probability that mean very different things to different people.