Evaluating Subjective Uncertainty Information in National Hurricane Center Tropical Cyclone Discussions

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 5 January 2015
Andrea B. Schumacher, CIRA/Colorado State Univ., Fort Collins, CO; and O. Vila and V. M. Vincente

The National Hurricane Center (NHC) issues several forecast products that include uncertainty information. Products such as the Uncertainty Forecast Cone and Wind Speed Probabilities are objective products that communicate uncertainty in the context of past forecast errors. Although their quantitative construction may make these products relatively straightforward to interpret and validate, both of these products are based on climatology and lack information specific to a particular forecast scenario. More situation-specific uncertainty information is contained in NHC Tropical Cyclone Discussions (TCD). These text products contain subjective assessments of the relative uncertainty of a given forecast and the sources of that uncertainty. Although it can be argued that TCDs contain the most relevant information on the uncertainty of a particular forecast, their subjective form makes them challenging to validate.

A pilot study was conducted to systematically evaluate the track and intensity forecast uncertainty information contained in NHC TCDs. A quantitative content analysis of past years' discussions was conducted and findings regarding the frequency and types of uncertainty information and the types of evidence cited (e.g., model agreement or spread, synoptic conditions) will be presented. A statistical comparison of uncertainty statements in TCDs and actual forecast errors will also be presented. It is hoped that this preliminary analysis will help guide future improvements to this product and provide a skill benchmark that can be used in developing and evaluating new uncertainty guidance tools. In addition, we believe this pilot study merely scratches the surface of this topic and is motivation for future work to investigate topics such as the estimated value, intended and actual use, interpretation by end users, and mode of dissemination of this unique type of tropical cyclone forecast uncertainty information.