Answers are grounded in this paper and cite their sources.
The National Hurricane Center (NHC) does not verify official or model forecasts if those forecasts call for a tropical cyclone to dissipate or if the real tropical cyclone dissipates. A new technique in which these forecasts are included in a contingency table with all other forecasts is presented. Skill scores and probabilities are calculated. Forecast verifications with the currently used technique have shown a slight improvement in intensity forecasts. The new technique, taking into account all forecasts, suggests that the probability of a forecast having a large (>30 kt) error is decreasing, and the likelihood of the error being less than about 10 kt is increasing in time, at all forecast lead times except 12 h when the forecasts are already quite good.
No takes yet. Share an insight, caveat, or question.
Sim D. Aberson (2008) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: