Richard Dixon looks at the factors affecting the intensity of this year's Atlantic windstorm activity.

The first day of June marked the official start of the 2003 hurricane season, though there had already been a tropical storm noted in April - the earliest on record. The 2003 season follows three years of fairly quiet activity, from the perspective of landfalling major hurricanes (Saffir Simpson Category 3 or higher). The last major hurricane to make landfall was Hurricane Bret in 1999, which passed into a relatively unpopulated area of Texas as a Category (Cat) 3 hurricane.

The 2002 experience
The most notable of the six systems that made landfall in 2002 was Hurricane Lili, which, with 12 hours remaining to landfall, was a Cat 4 system bearing down on Louisiana. It was at this time that Lili caused damage to offshore oil platforms in the Gulf of Mexico. Around 800 platforms and rigs were subjected to Cat 4 strength winds, which resulted in repairs lasting several months. Latest estimates of insured loss to offshore platforms are around $700m. Lili highlighted the enormous concentration of insured values that has accumulated offshore in the Gulf of Mexico, as well as how significant offshore energy losses can correlate with onshore losses.

The rapid decay of Lili as it approached land led to the system being downgraded to a Cat 1 system that eventually caused only $400m insured loss on land. However, nearly one million people were evacuated from their homes. Lili had previously tracked through the Caribbean, damaging ] homes in Barbados, Haiti, St Vincent and the Grenadines and also Jamaica, where four people were killed through flooding and landslides. Damage to homes also occurred in western Cuba, where an estimated 100,000 people were relocated to other parts of the island.

Other tropical systems in 2002 also caused some limited amount of damage. Hurricane Kyle, which skirted the coast of South Carolina, blew down power lines and damaged some homes in Georgetown. Tropical Storm Fay brought 13.6in of rainfall to Freeport, Texas, and Tropical Storm Hanna led to some precautionary evacuations of Dauphin Island, Alabama, with power outages also reported in both Florida and Alabama. The other landfalling system, Tropical Storm Edouard, caused minor flooding in a few counties in Florida.

Another interesting hurricane from last year's season was Hurricane Isidore, which headed through the Caribbean, affecting western Cuba as a Cat 3 hurricane (in a path similar to Hurricane Lili the following week), and causing extensive damage to housing and agriculture. Isidore moved onshore over the Yucatan Peninsular and began to dissipate over land, before heading north and making landfall in eastern Louisiana as a tropical storm.

Although already a weak system close to landfall, satellite imagery showed that the standard circular shape of the hurricane had become more elongated and asymmetric after it moved inland. Isidore was, in fact, transforming into an extratropical storm, becoming a 'transitioning storm'. Although Isidore was less notable for the damage it caused, heavy rains of up to 8in extended far inland on the left-hand side of Isidore's track as it made landfall and began transitioning, ending long-term droughts in the states of Kentucky, Illinois and Tennessee.

Transformations of hurricanes such as Isidore are not uncommon. Recent studies show that around 40-50% of tropical systems undergo extratropical transition at some point in their lifecycle. Famous hurricanes such as Hugo, Floyd and the 1938 'Long Island Express' in the Northeast all underwent extratropical transition before or soon after landfall.

Transitioning storms have two dynamical differences to hurricanes. First, the highest winds on the right-hand side of the track move away from the storm track, out to radii far higher than a typical tropical cyclone. The high windspeeds on the left-hand side of the track are correspondingly reduced. The second difference is that for a given central pressure, windspeeds are 20-30% lower in a transitioning storm than in a hurricane. Both these characteristics are typical of intense extra tropical cyclones found in Western Europe, such as 1990's Daria or 1999's Lothar. Furthermore, most rainfall in transitioning storms is located to the left-hand side of the track, rather than uniformly around the centre in tropical storms.

The last significant transitioning storm event along the US coast was Hurricane Floyd, in September 1999. From being a large Cat 4 hurricane grazing the edge of the Bahamas with southern Florida in its sights, Floyd recurved to the north and began to lose its spiral structure. Even after landfall in the Carolinas, as the storm tracked along the coast of Virginia and New Jersey, the windspeeds onshore to the left of the track were only 60-70% of those modeled for a typical hurricane windfield, leaving an insurance bill less than half that anticipated for even a Cat 2 hurricane with these central pressures and track.

Floyd transitioned as it made landfall, pushing the strongest of its winds out to sea on the right-hand side of the track, and reducing the winds and associated damage over land. If Floyd had kept the same path and remained as a hurricane, insured losses would have been closer to $5bn rather than the reported $2bn. Not all transitioning storms fall apart in the manner of Floyd. For example, two famous transitioning storms, Hugo in 1989, and the 'Long Island Express' of 1938, decayed more slowly over land than if they had remained as pure hurricanes.

In some cases, former tropical storms reintensify as they are transitioning into extratropical storms. Hurricane Keith in 1989, a modest tropical storm as it crossed Florida, reintensified over the western Atlantic as it transitioned to become an extratropical storm with deep (Cat 3) central pressure of 945mb.

In the tropical conditions of southern Florida, nearly all hurricanes are pure tropical cyclones. Along the Gulf coast a small number of hurricanes start to transition at landfall, such as Isidore last season. North of Cape Hatteras, the tropical waters of the Gulf Stream move offshore and the nearshore Atlantic is not warm enough to sustain a hurricane central pressure. In this region, transitioning starts to become the principal factor in determining the behaviour of the most intense and damaging storms. For a system to retain its strength in order to make landfall in the Northeast, it must move quickly over the cooler waters of the North Atlantic or else it will lose its energy; a warm sea is the source of hurricane's strength. However, the faster a storm is moving, the more likely it is to undergo extratropical transition. Thus, damaging storms making landfall in the Northeast are more often than not undergoing extratropical transition.

Revisiting all the known intense historical New England 'hurricanes', back at least to the 1860s, shows that all of them were transitioning at landfall.

Reconstructed damage patterns from these storms are consistent with the asymmetric nature of transitioning storms: the main damage (and thus strongest winds) is located in a broad swath to the right of the track.

El Niño effect
Last year's hurricane season was less potent than average partly because of moderate El Niño conditions. El Niño is the warm phase of ENSO, the El Niño Southern Oscillation ENSO. When it occurs, the westerly shear across the tropical Atlantic is stronger, which acts to suppress hurricane formation across the region. Examination of the phase of ENSO and the activity of the hurricane season is very enlightening. Shown in Figure 1 is a plot of the average number of tropical systems for each phase of ENSO. As it shows, the presence of El Niño acts to suppress the number of storms markedly. There are approximately two storms less per year than in a neutral ENSO phase (years in which ENSO is neither in its cold or warm phase, and close to three storms less when compared to the La Niña phase, which is the cold phase of ENSO.

This information would lead us naturally to ask whether the moderate El Niño changed in the run-up to this year's hurricane season. A study of weekly sea-surface temperatures (SSTs), and their departure from the mean for late May in the Pacific Ocean, including near coastal Peru where ENSO is measured, shows that while warm sea-surface temperatures and anomalies linger in parts of the eastern Pacific, the El Niño has diminished, a trend that is expected to continue.

This year's forecasts
The 2003 hurricane season got off to an early start. Tropical Storm Ana formed during the week of 21 April, making it the only named tropical storm ever to form during April since records began. In the past, tropical storms have been recorded in January, February and March. However, since sea-surface temperatures reach an annual minimum in the middle of March, they are actually part of the prior calendar year's hurricane season. So what does this presage for the season ahead?

It may seem surprising, but there is no correlation between early and late season tropical cyclone activity. Active and quiet early seasons have each been followed by both quiet and active late seasons. In an average season, we would expect to see approximately ten tropical storms, of which six would become hurricanes. Typically, about two of these would become major hurricanes (Cat 3 or above).

There are a number of research groups that issue tropical cyclone seasonal outlooks for the North Atlantic each year. How do the seasonal outlooks compare, not just with each other but also with an average season?

According to the latest forecast from Colorado State University, issued on 30 May, the 2003 tropical storm season will feature above average activity. The forecast predicts 14 named storms, eight hurricanes and three major hurricanes, which represents a slight increase in activity from the long-range forecast issued last December. Net tropical cyclone activity, a measure of frequency and duration, is 40% higher than the long-term average. The forecast also predicts a 69% chance of a landfall anywhere on the US coastline, higher than the century average of 52%.

The National Ocean and Atmospheric Administration (NOAA) forecast, like the Colorado State University forecast, predicts an above average season. It suggests a 55% probability of an above-average season, a 35% probability of a near-normal season, and only a 10% chance of a below-normal season. It is forecasting between 11 and 15 tropical storms, of which between six and nine will reach hurricane status, and two to four will attain major hurricane status.

The latest forecast for this season from the Benfield Hazard Research Centre, published on 6 May, predicts 12.4 named storms, 7.0 hurricanes, and 2.8 major hurricanes, which is higher than the climatological mean. Overall, according to the forecast ACE index, a measure of windspeed energy, activity will be above both the ten-year and 30-year means.

By Richard Dixon

Richard Dixon is a senior modeler with Global Risk Modelling at RMS.