Home | Prediction -> Products -> Hurricane Forecasts

  July 2004 IRI Atlantic Hurricane Activity Forecast

There is an enhanced probability (approximately 40%) that the number of tropical cyclones in the Atlantic during the 2004 peak season (August to October) will be in the normal category, which is defined as between 6 to 9 named tropical cyclones. This probability is greater than the long-term average probability of 33%. There is also a 30% probability that the number of named storms in this period will be in the below normal (5 or less named tropical cyclones) category, and a 30% probability for the above normal (10 or more named tropical cyclones) category.

Fig. 1 - Probability Forecasts for Number of Tropical Cyclones

The accumulated cyclone energy (ACE*) index during these months has an enhanced probability (approximately 40%) of being in the  normal range, with a probability of 30% each of being in the below normal and above normal ranges. 

Fig. 2 - Probability Forecasts for Accumulated Cyclone Energy

This forecast is consistent with the warm tropical Pacific sea surface temperatures (SSTs), and the slightly above normal SSTs in the northern tropical Atlantic, as shown in our SST forecast.

Background Information

The mean number of observed Atlantic named tropical cyclones (1971-2001) in the peak season is 8.1 with a standard deviation of 3.2. The near normal category is between 6 and 9 named tropical cyclones in the August to October peak season. The lowest number of tropical cyclones in the peak season during this historical period was 3 and the maximum was 15. If the peak season climatological median ACE in the period 1971-2001 is defined as 100%, the normal range varies between 68% and 129%. The standard deviation of the ACE index is 71%, and in extreme years the index can exceed 300% or be less than 20%. The historical variability in the ACE index is proportionally larger than the variability of the number of named tropical cyclones, as it takes into account not only the number of tropical cyclones but also their intensity and duration.

This outlook was produced by tracking Atlantic hurricane-like systems in one of our operational atmospheric general circulation models (AGCMs), ECHAM4.5, forced with IRI's predicted sea surface temperatures. While low-resolution (approximately 2.8 degrees longitude and latitude) AGCMs are not adequate for forecasts of individual hurricanes, they can have significant skill in predicting the amount of tropical cyclone activity over specific basins, as is the case for the ECHAM4.5 over the North Atlantic. Model tropical cyclones are weaker and larger than observed, but have an identifiable signature with many observed tropical cyclone characteristics. The model skill is due to the variability of the tropical cyclone activity being mainly determined by large-scale variables that affect that activity, such as sea surface temperatures and vertical wind shear, which can be predicted using AGCMs. The temporal distribution of these model tropical cyclones in the North Atlantic is similar to that of observed tropical cyclones in the region. The amount tropical cyclone activity of both model and observations in the Atlantic is strongly influenced by ENSO.

The IRI July 2004 hurricane activity forecast is different from the NOAA 2004 Atlantic Hurricane Outlook, which calls for an above-normal hurricane season. The Tropical Storm Risk July 2004 statistical forecasts of Atlantic hurricanes for the calendar year 2004, also calls for a slightly above-normal season, both in the number of hurricanes and in the ACE index. An above-normal Atlantic tropical cyclone activity season is the Colorado State University May 2004 forecast.

* The ACE index is calculated by summing the squares of the estimated 6-hourly maximum sustained wind speed in knots for all periods in which the tropical cyclones are either of named storm or typhoon intensity. Reference: Bell, G.D., and Coauthors, 2000: Bull. Amer. Met. Soc., 81, S1-S50 (see also http://www.cpc.noaa.gov/products/outlooks/background_information.html).