Abstract
The Tropical Meteorology Project at Colorado State University, led by Dr. William Gray, has been issuing Atlantic basin seasonal hurricane forecasts in early June and early August since 1984. This paper examines the skill of these forecasts over the past twenty‐five years (1984–2008) compared with climatology and a previous 3‐year, 5‐year and 10‐year mean. Seasonal forecasts are shown to have smaller mean‐squared errors than any of these metrics for most parameters, although improvements have been modest. The forecast models used by the Tropical Meteorology Project have changed over time, and the current early June and early August model's skill at hindcasting the past twenty‐five years are investigated using cross‐validation. Preliminary research indicates that using these new models may result in more significant skill improvements in future years.
Published Version (Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have