After the storm

Last year's devastating US hurricane season hammered home the link between energy and weather. But some believe it may also hold a key to predicting this year's weather, says Todd Crawford of Weather Services International

The 2005 Atlantic tropical hurricane season set many historical records, with 25 named storms, 13 hurricanes and three category 5 hurricanes - Katrina, Rita and Wilma. With seven major hurricanes - category 3 and above - occurring, 2005 was the most active season since 1950, when there were eight. There were five named storms in July 2005 for the first time on record, and the earliest category 4 storm was recorded on 8 July with Hurricane Dennis.

Do the conditions that were favourable for this record-breaking season have any predictive value for US temperatures this winter? There has been some speculation within the meteorological community that active tropical seasons are followed by cold Northeastern winters, using as evidence the fact that the three years with the most named storms before 2005 were all followed by cold winters (1933, 1969, 1995). If a significant relationship does exist, this would be very important for those involved in trading energy. This study looks into the relationship between tropical activity and US winter temperatures.

Methodology

Temperature data from nine US regions (figure 1) for the previous 110 winters (defined here as December-February) was obtained from the National Climatic Data Center (NCDC) website. Four different measures of tropical activity were used as potential predictors in this study: (1) number of named storms; (2) number of major hurricanes (category 3 or greater); (3) number of hurricanes; and (4) the Accumulated Cyclone Energy (ACE) index.

The ACE index is calculated by summing the squares of the estimated sustained velocity of every active named storm at six-hourly intervals. Figure 2 shows yearly ACE values for tropical seasons from 1851-2004. At the time of writing, the ACE value is 225, which ranks as the sixth largest value on record.

Results

Since the winter temperatures in the Northeast have the most impact on national energy usage, the tropical-winter relationships in this area are looked at first. Figure 3 details the relationships in the form of scatter plots. In all the scatter plots, the winter temperature rank is plotted on the y-axis, where 1 represents the coldest winter in the data set and 110 represents the warmest winter.

An examination of the plots suggests that there is no strong relationship between any of the four tropical predictors and the following winter in the Northeast, and in fact a regression analysis confirms this (R2 0.01 for all four comparisons).

However, if only the most active tropical seasons are inspected, one potentially interesting relationship stands out. The three most active years, as measured by number of named storms (figure 3a), were all followed by cold winters. Is this significant? If we flip a coin three times, for example, the probability of obtaining the same result (either all heads or all tails) is 25%. So, the relationship in figure 3a would occur by chance 25% of the time. Further, the fact that the other three predictors do not show this relationship also reduces confidence in its future forecasting skill.

To emphasise the conclusion, remember that all three category 5 hurricanes in 2005 had female names (Katrina, Rita, Wilma). Do we expect the next category 5 hurricane to have a female name as well? We certainly would not use this knowledge to predict that the next category 5 hurricane will also have a female name, but rather recognise the fact that it is likely that the "feminine 5s" occurred due to random chance. Similarly, we should be somewhat sceptical about the 'three-for-three' cold winter relationship.

Table 1 details the strength of the correlations (as expressed in R2 values) for all nine regions and for all four tropical predictors. There are marginally significant correlations for cold winters following active tropical seasons in the East North Central region and for warm winters in the Southeast. Neither of these relationships are confirmed by all four predictors, however.

The strongest correlations are found in the South and Southwest, where active tropical seasons are typically followed by warm winters. This relationship also holds when we only look at the most active tropical seasons. In figure 4, we see that 15 of the top 17 ACE years were followed by warmer-than-normal winters in the Southwest. This is the strongest signal we found in the entire US dataset. A relationship this strong would occur only once in 250 times for a random process.

Similarly, in figure 5, 12 of the top 14 ACE years were followed by warmer-than-normal winters in the South. This is the second strongest signal in the dataset. A relationship this strong would occur only once in 90 times for a random process.

What is the reason for the warm winters in the South and Southwest after active tropical seasons? Many of the warmest winters in this region occur during La Nina events. The weaker tropical wind fields associated with La Nina events typically allows for more active Atlantic tropical seasons. So, it should not be too surprising that active tropical seasons are usually followed by warm winters in these regions. This may be a case where correlation exists completely independent of causation. In other words, both the active tropical season and the warm winters are both symptoms of the larger seasonal driver of tropical Pacific ocean temperatures, and may not be directly related to each other.

Summary

This research shows that the strongest relationships exist in the US South and Southwest, where active tropical seasons are almost always followed by warm winters. In the Northeast, the three previous tropical seasons with the most named systems were all followed by cold winters, but the other three tropical predictors did not confirm this relationship, nor is it statistically significant.

Dr Todd Crawford is the lead seasonal forecaster and a senior scientist in the Forecasting Research and Development group at WSI Corporation, a leading provider of weather-driven business solutions for professionals in the energy, aviation and media markets, and multiple federal and state government agencies.Email: [e-mail].
Web: www.wsi.com.