This paper will document the most recent developments in the author's continuing research into predicting the impact of extremes of storminess, rainfall, and temperature during the Florida Dry Season (1 November through 30 April) from the major teleconnections; ENSO, PNA, NAO, and AO (Hagemeyer 2006 http://www.srh.noaa.gov/mlb/enso/P2.4_18th_CLIVAR_AMS.pdf , Hagemeyer and Almeida 2005 http://www.srh.noaa.gov/mlb/enso/16th_climate.pdf ). The focus of this latest study is on improving the predictability of the most significantly impacting forecast and sociological challenges of the Florida dry season, excessive stormy periods, excessive rainy and dry periods, and cold weather outbreaks.

To better determine sub-seasonal or intra seasonal variability and improve predictability of these impacting weather events, the 6 month (November - April) predictand database and forecast methodology refined in Hagemeyer 2006 was subdivided in space and time into two three-month periods, November, December, January (NDJ) and February, March, and April (FMA) for the Florida region and for subregions of Florida and for selected single stations. Multiple linear regression and logistic regression results for the 6-month Florida dry season forecasts in Hagemeyer (2006) were recomputed for all new combinations in space and time in an attempt to provide more detailed seasonal forecasts for decision makers. The overall significance of these relationships were similar to Hagemeyer (2006). The new logistic regression results more clearly defined scenarios when the forecasts of extreme storminess, rainfall and cold outbreaks work well and when they don't, which is valuable information for decision makers. Narrowing down the dry season forecasts in space and time also closes the gap between climate and weather as extreme sub-dry season variability is generally a result of the accumulated passage of individual weather systems or can even be the result of the influence of one extreme weather system. However attempting to narrow the sub-seasonal periods too much gave poorer results, validating the original thesis in Hagemeyer (2001 and 2002) that to achieve an acceptable confidence interval on extreme variability, forecasts of approximately 3 month periods are optimum due to sampling problems with the historical extreme database for one or two month forecast periods. Nevertheless important insights into the predictability of extreme storminess, rainfall, and cold outbreaks were achieved as well as into the veracity of attribution of extreme events to phases of the major teleconnection indices.

Extremely stormy and/or wet periods were found to be almost exclusively related to El Nino, and the challenge of prediction during ENSO neutral conditions, which are most common, remains. The Arctic Oscillation (AO) was found to be superior to the NAO as a predictor of temperature and it is noteworthy that there has not been a devastating statewide freeze in Florida since 1989, a period of 16 years, while there were 6 major freezes in the 1980's causing great economic impact and societal changes. Negative AO was found to be most highly correlated with colder temps in Florida and while the AO has generally trended higher since the late 80s (warm phase) it does not appear possible to attribute individual extreme cold outbreaks or the lack thereof to the mean state of the AO.