Summary: This project is designed to provide long-range probabilistic forecasts
of tropical cyclones that develop in association with convectively coupled
atmospheric waves and intraseasonal oscillations, as motivated by Frank
and Roundy (2006, Monthly Weather Review).

Roundy and Schreck (2008, submitted to Quart. Journ. Roy. Met. Soc.) derives the predictor set.
Another manuscript that is presently in preparation describes and analyzes the skill
of the forecast scheme, and it should be available Fall 2008. The progress of the modes
utilized for the forecasts can be viewed here.
Dr. Roundy presented a summary of the technique at the
28th Conference on Hurricanes and Tropical Meteorology in Orlando.


This is not a public warning system, and is not intended to
compete with warning authorities or organizations. Please
contact your appropriate warning authority for public forecasts and
warnings.

Predictions represent the likelyhood, given the large-scale state of atmospheric
waves, intraseasonal oscillations, climate variations, and the seasonal cycle
that a tropical cyclone of some intensity would be located within 7.5 degrees of
latitude and longitude of a given point. This system is not designed to forecast
the likelyhood of genesis alone, since it may also be useful for long-range
prediction of the most likely tracks of potential storms.

Occasionally, tropical cyclones form which are not associated with tropical waves
and intraseasonal oscillations. These would appear as 'missed' forecasts, or low
probability events.

Since the forecast depends on the historical storm record,
regions where storms are most frequent will exhibit the greatest
skill. The algorithm is not sensitive to many types of error prevalent
in the storm record, including, but not limited to specific storm locations
or intensities. However, the algorithm would be sensitive to storms missing
from the track data.

This forecast is not made for every grid point in the map domain:
Probabilities are not calculated for regions where storms do not
occur sufficiently often in the climatology to close the models (e.g.,
the eastern Atlantic, parts of the Central Pacific, parts of the
Arabian Sea, and large expanses of the extratropics (especially
over land).