A probabilistic approach to Tropical Cyclone Conditions of Readiness (TCCOR)

Abstract

Tropical Cyclone Conditions of Readiness (TCCOR) are set at DoD installations in the Western Pacific to convey the risk associated with the onset of destructive winds from approaching tropical cyclones. In this thesis, the methods by which TCCOR are set were analyzed to determine if objective and/or probabilistic guidance could improve the process. The Tropical Prediction Utility (TPU) was developed by forecasters at Yokosuka, JA and the Joint Typhoon Warning Center as a means of standardizing TCCOR forecasts using elements from JTWC official warnings. The TPU was used to recreate TCCOR timelines for 42 different cases affecting military bases at Guam, Kadena, JA, Sasebo, JA, and Yokosuka, JA during the 2002 -- 2007 typhoon seasons. These timelines were then compared to historical TCCOR timelines and wind observations to identify any trends and biases in set time and duration for each TCCOR. A wind speed probability model was also used to compare the timelines to the wind observations and to categorize them based on consistent trends in probability at each predicted and historical TCCOR. The results suggest that potential biases exist in the Tropical Prediction Utility that tend to predict TCCOR earlier than they were set in practice. Although clear trends were identified between wind speed probabilities and elevated TCCOR, statistical uncertainties exist when using the probabilities to discern between "its" and 'false alarms." While this thesis identified basic traits in TCCOR settings, a larger sample of cases is needed for further study to determine factors that discriminate between hits and false alarms.