Including expressions of uncertainty within local forecast products during tropical cyclone wind situations

In preparation for tropical cyclone events, critical decision-makers not only need a meteorologist's best deterministic wind speed forecast, but they also require an accompanying expression of uncertainty in order to make responsible decisions. This requirement reveals the inherent shortcoming of deterministic-only wind speed forecasts such as those found in the current Zone Forecast Product (ZFP) and Coastal Waters Forecast (CWF). To address the problem, the Miami and Melbourne Weather Forecast Offices (WFOs) have developed an experimental method in which tropical cyclone wind speed probabilities are used to enhance the language of the ZFP and CWF by introducing expressions of uncertainty (e.g., forecast error) within the products. These enhancements have been incorporated within the legacy (zone-based) versions, and also within the dynamic point-and-click versions found on the WFO Web sites.

Since 2005, the National Hurricane Center (NHC) has produced gridded tropical cyclone wind speed probabilities for 34-, 50-, and 64-knot winds through 120 hours during operational forecast cycles for active systems in the Atlantic and Pacific Basins. The probabilities are centered about the NHC's official track, intensity, and wind radii forecast, based on average error statistics over recent years for those variables. Together with tropical cyclone watch/warning information and deterministic wind speed information, incremental probabilities of threshold wind speeds are used to trigger, through automated text formatters, predefined expressions of uncertainty. The formatters weigh various gridded inputs to identify the appropriate expression of uncertainty. For each period of the 5-day forecast, the products convey whether hurricane or tropical storm conditions are possible, expected, or imminent/ongoing. Preliminary results from the use of these products have been positive and, if transitioned to official policy, would increase the usefulness of the ZFP and CWF. This would foster a greater consistency between the NHC and coastal WFOs, while reducing the workload associated with manual editing. Similar improvements in the utility of tabular products, such as forecast matrices, could be easily accommodated. This paper will present a summary of recent experiences collected through real-time and post-event testing, as well as planned improvements for the 2008 season. Some issues that must be addressed before operational transition can occur will be discussed.