By Jim Williams Sept 8th,2016
On August 18,2016 a tropical wave labeled invest 99L located S.W of the Cape Verde Islands began it’s long journey west and for the next 12 days would drive forecasters crazy. Early on some intensity models such as the SHIP and IVCN were showing potential for a category 1 hurricane in or near the Caribbean. The system 99L would not find a favorable environment for developing into a tropical storm until it would reach the Gulf Of Mexico. Some of the most reliable models even the ECMWF (European center for medium range forecasting) flip flopped on this becoming a named storm before reaching the gulf. At one point the ECMWF for at least 3 runs showed a potential hurricane hit for South Florida causing local TV mets to ratchet up coverage. I have to admit I did get caught up in the possibilities of this happening by checking supplies and alerting some business associates. How things can change as a couple of days earlier in my update video I mentioned how 99L could tangle with Hispaniola and not develop at all as the system was dragging anchor. Meanwhile the GFS (Global forecast system) American model showed limited development at best and seemed to have the best handle on 99L until it reached the gulf. Even the BAMS (Beta and Advection Model) shallow version for weaker systems pretty much nailed track & intensity until turning north in the gulf.
Up until 99L started developing in the gulf I thought the GFS model was finally going to win a big tropical tracking story in the Atlantic and get one on the ECMWF. I like to compare the GFS to the NFLs Buffalo Bills of the 90s always playing well getting to the Superbowl (projection of hurricane landfall) only to lose to the ECMWF . At the end of August 99L finds a more favorable pattern for development with a trough breaking the ridge to the north and nearby upper lows weakening plus the high to it’s south causing anchor dragging was finally letting go allowing 99L to become Hermine. Meanwhile the GFS model still insist that development will be minimal while most runs of the ECMWF and the HWRF (hurricane weather research model) show a hurricane at landfall in Apalachee Bay,oh no not again? Remember 2012 Hurricane Sandy GFS out to sea ECMWF N.E landfall. Funny thing is while all the comparisons of the GFS vs ECMWF get debated ad nauseum on Hermine quietly the UKMET (United Kingdom Met Office) model out performed both of these models in both track and intensity.
Notice the bottom line on both charts above OFCL (National Hurricane Center (NHC) official forecast in ATCF system and not from advisory data) did the best overall. With all of the talk about which model is the best the National Hurricane Center takes all this data and decided the forecast track and outperformed the models. The group of forecasters in Miami are very good at this while not always better than the model consensus are usually very close to the best. Often the best performing models will be displayed inside the NHC cone at hurricanecity.com at times reflecting the thinking of NHC forecasters.
Hurricane Hermine made landfall on Friday at 1AM just as predicted but with 80mph winds slightly higher than the projected 75mph point is outstanding job by NHC. Was the forecast perfect? no, if not for the stubborn GFS not showing development NHC may have upped the hurricane landfall chances much sooner. Forward to the threat to New England the cone above shows uncertainty off the coast which ended up happening as Hermine circled as a post tropical storm for days. I do think it is a great idea to have NHC continuing advisories on post tropical systems as the threat to lives and property resumes and can increase the expanse of threatened areas. The watches and warnings were necessary due to the expanding size of Hermine off New England with several models such as the HWRF and GFDL showing a track very near the coast,NHC could not take chances on the heels of Sandy in 2012. On a final note the NHC track forecast cone is the best tool for figuring out where a system will go especially short range but it is time to eliminate the poor performing models and start more funding for the better ones starting with the GFS.