Through August, the 2004 hurricane season has had 86 percent of the NTC activity of the average hurricane season. No tropical cyclone activity occurred in June or July. As of 1 September, eight named storms, four hurricanes and three major (Cat. 3-4-5) hurricanes have developed. The average number of named storms through August is 4.2, the average number of hurricanes is 2.4, and the average number of major hurricanes is 0.7. This August has thus been much more active than the average August, with these numbers being 8, 4, and 3.

1.1  September Forecast

      Table 2 and Fig. 1 list the nine September-only predictors with their location and sign for higher TC activity. Table 3 shows which of the nine predictors are used for each forecast parameter of NS, NSD, H, etc.

Table 4 lists the value of each September-only predictor and whether its 2004 value indicates above or below average September-only TC activity. Six out of nine predictors indicate a below-average month; however, the September statistical prediction calls for about average activity, due to the significant number of predictors that are just slightly negative for storm activity and the greatly enhancing Atlantic SST pattern presently in place.

Our early September update of September TC activity has been increased somewhat (see Table 5). This is due to the unexpected considerable amount of tropical cyclone activity that has occurred in the deep tropics during August. When August is very active in the deep tropics (as this month has been), September activity tends to also be enhanced.

1.2  Independent October-only Statistical Forecast

       We are issuing our second monthly forecast for October-only tropical cyclone activity this year. Through examination of the NCEP/NCAR reanalysis, we have discovered four predictors that in combination explain about 60 percent of the October cross-validated variance in Net Tropical Cyclone activity (at 1 September) for the hindcast period of 1950-2001. We are currently unable to find combinations of predictors that explain large amounts of variance for the individual tropical cyclone parameters (i.e., named storms, hurricane days, etc.). Therefore, our October forecast consists of predicting NTC and consequently increasing or decreasing October's values for the other parameters accordingly. For example, if October NTC was 150 percent of normal, and a typical October had two named storms, we would forecast three named storms for October.

Table 6 and Fig. 2 list the four October-only predictors with their location and sign for enhanced TC activity.

Table 7 lists the value of each October-only predictor and whether its 2004 value indicates above or below average October-only TC activity. A majority of the October predictors call for below-average activity.

Three out of the four October predictors are negative. However, the one positive predictor (August SST off the coast of Japan), is quite positive, and therefore we are increasing our October forecast somewhat from what we predicted in early August. We are now calling for a slightly below-average October with an NTC of about 80 percent of the climatological average (Table 8). In round numbers, we are forecasting 3 named storms, 1 hurricane, 0 intense hurricanes and an NTC of 15 for October. This forecast of below-average October activity is due primarily to the very warm values of sea surface temperature anomalies (SSTA) in the Niño 4 and Niño 3.4 regions of the equatorial Pacific. Warmth in these regions has been shown to reduce Atlantic basin late-season tropical cyclone activity.

Table 9 displays a summary of this year's hurricane activity through August and our projection for the rest of the season. We expect September activity to be much above the average September and October activity to be slightly below normal. We assume no activity in November.

Due to the unusual amount of August Atlantic basin tropical activity, we are raising our full season 2004 forecast to 16 NS (earlier forecasts 14 and 13); to 8 hurricanes (earlier 8 and 7); and to 5 intense (Cat. 3-4-5) or major hurricanes (earlier forecast was 3).

2  U.S. Landfall Probability Forecast for September and October

       We have recently developed a methodology for calculating the probability of hurricane landfall along the entire U.S. coastline for the months of September and October. Based on a 1950-2001 dataset, the probability of various intensity classes of tropical cyclones making landfall in September is based on September NTC values and North Atlantic sea surface temperatures. For our forecast of a September 2004 NTC of 85 our landfall probabilities are given in Table 10. Table 11 displays the landfall probabilities for October. Landfall probabilities for the U.S. are above average for this September based on a above average Atlantic basin September NTC forecast. October landfall probability is expected to be below average based on a lower than average forecast of October NTC.

3  August 2004's Unusually High Hurricane Activity

       Since the start of Atlantic basin aircraft reconnaissance flights into hurricanes at the end of World War II there has never been a previous August to have three intense or major (Cat. 3-4-5) hurricanes (Alex, Charley, and Frances) develop in the Atlantic basin. Only August 1995 (5) saw more Atlantic basin hurricanes develop. Of this August's four hurricanes (Alex, Charley, Danielle, Frances) all but Danielle impacted the United States. Although we forecast an above-average amount of tropical cyclone activity for August we could not have anticipated the unusually high amounts of storm activity that occurred (8 named storms). This will be a topic for much future research.

An important enhancing factor (that we appear to have underestimated)was the influence of the unusually strong positive sea-level temperature anomaly (SSTA) patterns in the eastern Atlantic. A very similar warm Atlantic SSTA pattern occurred in 1995 when August also had one of the highest recorded named storm totals (7).

August 2004 hurricane and troipcal storm activity has been close to an August record. We expect the remainder of the 2004 hurricane season to be more active than the average September-October period. We foresee above-average September activity but below average activity for October and no activity for November.

4  August Versus September Association

       For those August periods which have had twice or more of the average amounts of Atlantic basin August tropical cyclone activity (or August NTC) it is observed that the following September NTC is also above average (see Table 12). Active September NTC amounts usually follows high values of August NTC. In hurricane seasons having August NTC values over 50 (i.e., twice as much as the average August seasons, September NTC usually also above average. August 2004 NTC activity was 86. On this basis alone we should expect above average NTC this September.

By contrast, it is observed that an inactive August does not mean that September will also be inactive. Table 13 lists the 10 most active Septembers since 1950 which followed inactive Augusts.

Very active Augusts are usually followed by active Septembers. But an inactive August says little about how active September will be.

Table 14 shows those 10 years with a low values of NTC activity in both August and September.

5  United States Impact

       As of early September the US has been impacted by two tropical storms (Bonnie and Gaston) and two major hurricanes (Charley and expected major hurricane Frances). Charley will likely be judged to be the second most destructive U.S. hurricane behind Andrew (1992). And, Frances will likely bring more destruction than Charley did and perhaps rival Andrew's massive economic loss.

The global climate signals since 1995 have been similar to the global climate signals of the mid-1920s to the mid-1960s when many more major hurricanes struck the US East Coast and Florida. Since August 1995 the Atlantic basin has experienced 35 intense or major (Cat. 3-4-5) hurricanes of which only five have made US landfall: Opal (1995), Fran (1996), Bret (1999), Charley (2004) and now Frances (2004). This amounts to one U.S. landfall major hurricane for every seven Atlantic basin major hurricanes. The long term average is about one in three. Even though the U.S. has experienced two major hurricane strikes so far this year (Charley and Frances), in terms of the number of U.S. major hurricane strikes the last 10 years (only 5 of 35 versus one of three for the century long average) we have been, from this perspective, very fortunate.

6  Florida's Recent Decades of Major Hurricane Avoidance

       The Florida landfall of Major Hurricanes Charley and Frances, although very unfortunate, should not be taken as indicating that Florida is experiencing anything different than what it has seen in the past. The real surprise should be that Florida (excluding the panhandle) has experienced so few major hurricane strikes since 1965 as compared with earlier periods. These differences are shown in Table 15. During the 40-year period of 1926-1965 the Florida Peninsula received 14 major hurricane landfalls while during the recent 38-year period of 1966-2003 only Andrew came ashore. Part of this difference is due to the difference in global climate signals between these two periods (see the previous discussion on our web site of the changing strength of the Atlantic Thermohaline Circulation and its influence on U.S. East Coast major hurricane landfall). Another factor in Florida's avoidance of major hurricane landfall has been just plain luck. Many major hurricanes have moved close to Florida but just have not come ashore. For example, major hurricane Floyd (1999) came very near the southeast Florida coast before veering to the North.

So, Floridians should, in general, feel lucky that they have not received more landfalling major systems in recent decades. Major (Cat 3-4-5) hurricanes make up only about a quarter of all named storms but when normalized by inflation, coastal population, and wealth per capita they cause, on average, about 80-85 percent of all hurricane destruction (Pielke and Landsea 1998). One should not invoke global warming or any other special or unique climate change processes to explain the two recent Florida landfalls of major Hurricane Charley and Frances occurring only three weeks apart. These two Florida landfalling hurricanes should be thought of as the normal adjustment to the long period climatological average. Florida is located at a latitude and longitude where the long term records of the last century-and-a-half show that major hurricane in Florida (excluding the Panhandle) account for about a third all U.S. landfalling major hurricanes of which the U.S. had 73 in the last century. In the last 38 years (1966-2003) Florida major hurricane landfall numbers (1) have been anomalously low. The Florida landfall of major hurricanes Charley and Frances in the last three weeks should be thought of as a rare but natural occurrence and an expected statistical adjustment to the century long climatological average. The real surprise is that Peninsula Florida went nearly 39 years (1965 to 2004) with only one major hurricane making landfall (Andrew, 1992).

7  Forecast Verification

      We will issue a seasonal update and a revised October-only forecast on Friday, 1 October 2004. All our forecasts for 2004 will be verified in late November 2004. Our first seasonal hurricane forecast for the 2005 season will be issued on Friday, 3 December 2004. All forecasts and verifications are available at our web address given on the front cover (http://tropical.atmos.colostate.edu/ forecasts/index.html). Our individual county, subregion and region tropical cyclone landfall probabilities are available online at http://www.e-transit.org/hurricane.

8  Papers Which Document Our August-only and September-only Atlantic Basin Forecasts

Blake, E. S., 2002: Prediction of August Atlantic basin hurricane activity. Dept. of Atmos. Sci. Paper No. 719, Colo. State Univ., Ft. Collins, CO, 80 pp.

Blake, E. S., and W. M. Gray, 2004: Prediction of August Atlantic basin hurricane activity. In Press. Wea. and Forecasting

Klotzbach, P. J., 2002: Forecasting September Atlantic basin tropical cyclone activity at zero and one month lead times. Dept. of Atmos. Sci. Paper No. 723, Colo. State Univ., Ft. Collins, CO, 91 pp.

Klotzbach, P. J., and W. M. Gray, 2003: Forecasting September Atlantic basin tropical cyclone activity. Wea. Forecasting, 18, 1109-1128.

Pielke, Jr. R. A., and C. W. Landsea, 1998: Normalized Atlantic hurricane damage, 1925-1995. Wea. Forecasting, 13, 621-631.