Friday, January 21, 2005

2004 Hurricane season in review Yesterday I attended an excellent seminar on the 2004 hurricane season.  Here's the highlights: The lecturer was Jack Beven of the National Hurricane Center.  He got his bachelor's degree at LSU and his master's and PHD from Florida State and has been with the NHC since 1993. He started off by comparing conditions in the hurricane formation areas to the long term average.  Sea Surface temperatures were 1-2 degrees (C) warmer than average, which contributed to the number of storms.  He then showed what the average upper air pattern was and compared it to 1995, which was a season that had a large number of storms,  but few making landfall (and the ones that made landfall did so in the Gulf of Mexico).  The key difference in the patterns was that the persistent high pressure was much farther east in 1995 and there was a trough of low pressure immediately off the Atlantic coast.  In 2004 the high was near the coast and there was no intervening trough.  This provided the steering necessary for storms to landfall on the Atlantic coast (Florida in particular).  The other contrast with 1995 was that the activity of 2004 fell neatly into the historical average with regards to when peak activity was.  Nearly all of the activity was in August and September.  The 1995 season had an earlier start and stayed busy through October. He then went into the 'highlights' of the season.  Jeanne was far and away the deadliest storm (3000+ killed) due to the disaster in Haiti.  Ivan claimed 94 lives, Charley 15, and Frances 8.   Total fatalities in the U.S. from  hurricanes was 62. In terms of damage, Ivan is currently rated as the costliest with 14.2 billion dollars of damage and Jeanne is a close second at 14 billion.  Frances' damage is rated at 8.9 billion and Jeanne's at 6.9 billion.  Total damage by all tropical systems was 44 billion dollars.  Note that these numbers are still in motion.  Nevertheless, these four storms all make the top six in terms of damage. There were 15 named storms, nine of which became hurricanes; six of them major.   The  eight tropical storms in August set a record. Another record set was by Hurricane Ivan, which had the highest Accumulated Cyclone Energy of any storm that the NHC has requisite data for (going back to 1899).  I had never heard of this index before.  It essentialy gives a number that indicates how strong and persistent a storm was.   Abrief  web search yielded this description:  The Accumulated Cyclone Energy (ACE) Index is calculated by summing the squares of the estimated 6-hourly maximum sustained wind speed in knots (Vmax2) for all periods in which the tropical cyclone is a tropical storm or greater intensity. For a complete description of the ACE Index see Bell et al. (2000), Bulletin of the American Meteorological Society (81) S1-S50. The season had the most landfalling hurricanes since 1985.  Due to the special circumstances of Gaston (which was rated as a tropical storm at landfall, and only later deemed to be a hurricane at landfall), this fact wasn't noticed immediately. This season had a record low for track forecast errors for each 12 hour period up to 72 hours before landfall.  The best model overall (for both track and intensity forecasting) was the FSU Superensemble.  He noted that no model has been the top one for two years in a row since the 80's. He then started to review the storms individually. Alex developed into a major hurricane further north than any storm since 1973 (Ellen apparently.  I'm not 100% sure that I wrote down the year correctly) Charley will be long remembered for his compactness.  The eye was never much larger than 10 miles and shrunk to 5-7 miles before landfall.  Due to the small size, there was not the storm surge that one would expect from a category four hurricane.  While the media controversy in the wake of the storm was over the track forecast, the error was within what the NHC considers to be acceptable and also, the landfall area was within the warning area.  The issue of concern to meteorologists is the intensity forecast, which completely missed the rapid intensification that ocurred in the final hours before landfall.  The central pressure fell from 964 millibars to 931 millibar in a time span of 4 hours 35 minutes.  This bumped Charley from category two to category four.  There is still no real explanation of why the strengthening ocurred.  The small size of the storm  certainly helped.  There is speculation that the storm interacted with an upper level jet streak, but that doesn't yield an explanation either.  Therefore, the mystery remains.  Beven semi-jokingly asked for volunteers from the graduate students to do research on the storm and explain it all.  In comparing to Frances to Charley, Beven said that Charley was a blitzkrieg and Frances was a siege.  He showed a graphic similar in concept to the one offered by WeatherBlog that showed how small Charley was in comparsion to Frances.  As I mentioned earlier Gaston was considered to be a tropical storm when it made landfall.  While during post-storm analysis of Doppler radar and surface obsevations, it was found that Gaston was a bare minimum hurricane. He showed two pictures of damge from Ivan, one being the I-10 bridge, the other being his beach house at Gulf Shores.  The lower floor of the two story house was devestated.  The house was part of the post-1979 (i.e. Hurricane David) building boom.  He stated that he was mildly amazed that the house made it this long without significant damage.  Ivan set off a significant number of tornadoes.  At one point on the afternoon of landfall, the National Weather Service office in Tallahassee had 13 tornado warnings out simultaneously.  On September 17, there were 39 tornado reports in the Northern Virginia/DC area.  In discussing Jeanne he reminded us that she never went through Haiti.  He apparently had to do a fair bit of searching to find a satellite image showing the heavy rain that caused catastrophic loss of life in that country.  It appears that a thunderstorm somehow got hung up in the mountains, which caused torrential rainfall.  He also showed the dilemma facing forecasters on the morning of September 19.  The 12Z model runs had an incredible amount of divergence in the forecast track of Jeanne (The UKMET model placed it at 80 degrees west, while the GFS placed it at 62 degrees west after five days; a difference of more than 1000 miles).  The GFS, of course, was the one that was most wrong, its five day error on that run was 1400 miles.  The FSU Superensemble and the GUNS consensus model handled the situation well, and that is pretty much what the NHC went with in its forecast. In mentioning Matthew, he noted that the storm had a storm surge that was disproportionate to its strength (i.e. unexpectedly significant).  His final slide was a quick look ahead for the 2005 season.  He noted that the skill of seasonal forecasts made in December/January is near zero (i.e. you could almost do as well by picking numbers out of thin air).  However, there currently are no apparent inhibiting factors present at this time that would supress hurricanes in the 2005 season; the El Nino Seasonal Oscillation (which can be quite a hurricane killer when strong) is fairly weak.  Therefore, it is reasonable to expect an average or above average season again.  In the question and answer session, someone continued the discussion on seasonal forecasting by mentioning that the forecasts issued by Dr William Gray only look at the number of storms, and doesn't really look at the potential for landfalling storms (which is what we ultimately care about).  It was stated that there is a paper forthcoming in Nature that describes a possible scheme for predicting the season's potential for landfalling hurricanes in advance.   Going back to the comparision of  the 1995 and 2004 seasons, it is clear that if one could anticipate the upper level pattern in advance, then it would be reasonably straight-forwardto make a decent forecast of the probability of  storms making landfall.  All in all it was an hour and a half well spent. Weather comment []1:49:19 PM   trackback []