5 day Tropical Weather Outlooks from the National Hurricane Center
July 31, 2013
A perhaps under-appreciated bit of progress made in computer modeling with regards to tropical cyclones is how accurate they’ve become in forecasting development of some types of disturbances well in advance. Over the past decade, dynamic models have forecast tropical cyclogenesis farther in the future, sometimes before the spawning disturbance is even in the Atlantic Ocean, and crucially, with fewer false positives. Tomorrow, the National Hurricane Center will avail itself in these advances when they release their first Tropical Weather Outlook that covers a five-day period. For the first time, the NHC’s four-times a day guidance on which disturbances are being monitored for potential tropical cyclone development will extend beyond 48 hours. It is a part of an evolution that’s been ongoing for just over a decade. The evolution has given greater clarity to the Hurricane Center’s thoughts on the probability of development.
Let us remind ourselves of the form that the Tropical Weather Outlook had for most of the decades of its existence. The following paragraphs are taken from the nominal 5 PM Outlook on Monday, August 22, 2005 and the forecaster was future NHC director Rick Knabb:
A LARGE TROPICAL WAVE IS LOCATED OVER THE EASTERN ATLANTIC OCEAN
ABOUT 525 MILES WEST OF THE CAPE VERDE ISLANDS. THE ASSOCIATED
SHOWER ACTIVITY REMAINS LIMITED… BUT THIS SYSTEM HAS SOME
POTENTIAL FOR SLOW DEVELOPMENT DURING THE NEXT DAY OR TWO AS
IT MOVES WESTWARD OR WEST-NORTHWESTWARD AT 10 TO 15 MPH.
DISORGANIZED CLOUDINESS AND SHOWERS EXTEND FROM EASTERN CUBA AND
HISPANIOLA ACROSS THE SOUTHEASTERN BAHAMAS AND THE TURKS AND CAICOS
ISLANDS… AND INTO THE ATLANTIC FOR SEVERAL HUNDRED MILES. THIS
ACTIVITY IS POSSIBLY ASSOCIATED WITH THE REMNANTS OF TROPICAL
DEPRESSION TEN… AND DEVELOPMENT DURING THE NEXT DAY OR TWO SHOULD
BE SLOW TO OCCUR AS THE SYSTEM MOVES WESTWARD OR WEST-
ELSEWHERE… TROPICAL STORM FORMATION IS NOT EXPECTED THROUGH
In this format, the outlooks made an effort to communicate what the forecaster’s thoughts were on the potential for development. Admittedly though, the wording could be “fuzzy” at times, and it required a close reading on a 6 hour basis to sniff out changes in wording that may have indicated the forecaster thinking more highly of the system’s future prospects. We’ll continue with the system mentioned in the third paragraph as an example on how the wording would change as development occurred:
CLOUDINESS AND SHOWERS EXTEND FROM EASTERN CUBA AND HISPANIOLA
ACROSS THE SOUTHEASTERN BAHAMAS AND THE TURKS AND CAICOS ISLANDS…
AND INTO THE ATLANTIC FOR A FEW HUNDRED MILES. THIS ACTIVITY IS
POSSIBLY ASSOCIATED WITH THE REMNANTS OF TROPICAL DEPRESSION TEN.
GRADUAL DEVELOPMENT OF THIS SYSTEM IS POSSIBLE DURING THE NEXT DAY
OR TWO AS IT MOVES WEST-NORTHWESTWARD AT ABOUT 10 MPH.
So, we’ve gone from “development should be slow to occur” to “gradual development is possible”. This is an example of a subtle change where it would be hard for the average person to discern whether this was merely re-wording or a slight change in thinking.
CLOUDINESS AND SHOWERS EXTEND FROM EASTERN CUBA AND HISPANIOLA
ACROSS THE SOUTHEASTERN BAHAMAS AND THE TURKS AND CAICOS ISLANDS…
AND INTO THE ATLANTIC FOR A FEW HUNDRED MILES. THIS ACTIVITY…
WHICH IS POSSIBLY ASSOCIATED WITH THE REMNANTS OF TROPICAL
DEPRESSION TEN…HAS BECOME MORE CONCENTRATED THIS MORNING NEAR THE
SOUTHEASTERN BAHAMAS. ADDITIONAL DEVELOPMENT OF THIS SYSTEM IS
POSSIBLE DURING THE NEXT DAY OR TWO AS IT MOVES WEST-NORTHWESTWARD
AT ABOUT 10 MPH. AN AIR FORCE RESERVE HURRICANE HUNTER AIRCRAFT IS
SCHEDULED TO INVESTIGATE THE SYSTEM THIS AFTERNOON…IF NECESSARY
Now, “additional development is possible”. Certainly an upgrade based on existing circumstances. Seems like we’re getting closer to something happening, but how close?
A BROAD SURFACE LOW PRESSURE AREA IS PRODUCING WIDESPREAD CLOUDINESS
AND THUNDERSTORMS FROM EASTERN CUBA AND HISPANIOLA NORTHWARD ACROSS
THE SOUTHEASTERN BAHAMAS…THE TURKS AND CAICOS ISLANDS… AND INTO
THE ATLANTIC FOR A FEW HUNDRED MILES. THIS ACTIVITY HAS BECOME A
LITTLE BETTER ORGANIZED OVER THE SOUTHEASTERN BAHAMAS… AND A
TROPICAL DEPRESSION COULD FORM LATER TODAY OR ON WEDNESDAY AS THE
SYSTEM MOVES TO THE WEST-NORTHWEST OR NORTHWEST AT 5 TO 10 MPH. AN
AIR FORCE RESERVE UNIT RECONNAISSANCE AIRCRAFT IS SCHEDULED TO
INVESTIGATE THE SYSTEM THIS AFTERNOON. INTERESTS IN THE BAHAMAS…
THE NORTH COAST OF CUBA…AND SOUTHERN FLORIDA SHOULD MONITOR THE
PROGRESS OF THIS SYSTEM.
Now it’s clear that we’re real close. No ambiguity in “a tropical depression could form later today…” And that was indeed the case
THE NATIONAL HURRICANE CENTER HAS INITIATED ADVISORIES ON TROPICAL
DEPRESSION TWELVE… LOCATED OVER THE CENTRAL BAHAMAS ABOUT 175
MILES SOUTHEAST OF NASSAU.
(For those wondering about the fate of the other disturbance mentioned in the first Outlook from which I quoted, it would go on to be mentioned in Outlooks for the entire lifespan of Katrina, with the last mention coming almost exactly a week later stating that associated shower activity had diminished. It ran nearly the entire gamut of phrasing.)
The Outlooks quoted above were part of a study conducted to assess the feasibility of incorporating categorical probability (low, medium, high chance) of development into the Tropical Weather Outlook. As detailed in a paper presented at the American Meteorological Society’s 2008 Hurricane Conference, “Verification of the National Hurricane Center’s Experimental Probabilistic Tropical Cyclone Genesis Forecasts“, the study showed potential for the idea and it was tested in-house during the 2007 season, (which happened to be the first year of the Graphical Tropical Weather Outlook, then in experimental form). The testing demonstrated that the idea would work and it was subsequently incorporated into the 2008 Outlooks and continued their form in 2009.
Over time. in-house forecasting using explicit 10% increments (rather than low, medium, and high categories) showed enough refinement that the 2010 Outlooks used the percentages. As explained in the 2010 season verification report (section four), forecasts were well calibrated for the high and low end percentages, but there was an inversion in the middle range (a greater percentage of systems with a 40% chance of development actually developed than those with a 70% chance). The summary of 2007-2010 forecasts reflected some of that calibration issue, but overall the forecasts matched up well.
So, the forecasters were making progress with the product they had in its current scope, but it was becoming clear that potential existed to broaden that scope. Hurricane Julia of 2010 provided an example of forecast models providing accurate guidance outside of the bounds of the Tropical Weather Outlook;
The genesis of Julia was not well anticipated. The disturbance that became Julia was introduced into the Tropical Weather Outlook (TWO) with a medium (30%) chance of formation
only 18 h before the system became a tropical cyclone. This prediction was raised to a high (70%) chance 6 h before genesis occurred. The lateness in mentioning the disturbance that spawned Julia may have been due to the typical operational practice of not introducing an African wave disturbance into the TWO until it reaches the Atlantic Ocean. In this case, the disturbance developed into a tropical cyclone only about one day after leaving the West African coast. It is of note that many of the global model forecasts successfully called for the genesis of Julia up to several days in advance.
(Discovery of a Master’s thesis on the genesis of Hurricane Julia helped provide this example.)
Julia was not a solitary case. While there were certainly instances of “quick-draw” development not foreseen by the models, they were proving to be adept at the “classical” mode of development, the Cape Verde storm. The NHC was aware of this and had started performance of 5-day Outlooks that were kept in house and results were reported at the 2012 AMS Hurricane Conference (“Extended-range genesis forecasts at the National Hurricane Center” recorded presentation). The results start being shown about five minutes into the presentation and the presenter, NHC hurricane specialist Eric Blake explained that one issue they saw with the extended part of the forecasts (days 3-5) was that many systems that were given a mid-range probability of developing during that time period actually developed earlier (in the 1-2 day timeframe). 25% of systems in that 40%-60% range failed to ever develop. However, they were all systems that would go on to be in the short-range outlook; they weren’t phantom systems. Once upon a time, the prevalence of “boguscanes” in the models was such that this would have been a difficult feat to achieve.
The presentation went on to show the results for the eastern Pacific basin, interestingly, even though forecasters were aware of their tendency to under-estimate the probability of development (and Blake said that he had started to take that tendency into consideration and adjusting his forecasts accordingly) there was still an under-forecasting bias in that basin. At the 9:45 mark he shows a comparison of the short-term and extended term forecasts for each storm in the 2010 season. One sees that the in-house Outlooks containing Julia picked up on the potential for development four days in advance and that Lisa was another storm seen well in advance.
The work continued and before this season began, the NHC had announced that 5 day Outlooks would become public at some point in this season. A presentation of the 5 day outlooks was made at the 2013 National Hurricane Conference. The outlooks will give the usual write-up and 0-48 hour percentage chance of development for existing systems and have a separate section for systems that exist/are only in the extended time-frame and have a near 0% chance of developing in the short term. Blake’s presentation touched on some of the issues with a graphical presentation (e.g. how to handle systems that overlap) and it appears those have not been fully resolved yet; the announcement of the August 1 debut of the extended Tropical Weather Outlook states that the accompanying graphic is under development and “may become available later this season”.
This new product represents a great deal of progress that has been made in forecasting tropical cyclogenesis. A great deal remains, however. Two customers that expressed interest in these extended outlooks, the oil services industry and the US Navy, might be disappointed, in the near term, at least. While the models have shown increased skill in anticipating storms in the deep tropics, the Gulf of Mexico and sub-tropics remain a trouble spot. The former, of course, being the location of interest for the oil business and the latter being relevant due to the Navy owing to the massive Norfolk Naval Station. It is doubtful that either will find great value in having extra notice of a storm that’s ten days to two weeks out at sea.
The National Hurricane Center and researchers will work aggressively on that problem, of course. This is but one part of their efforts to improve and expand the scope of forecasting. 7-day forecasts, with the accuracy of 4-5 day forecasts from a decade ago are another example of in-house work that is being prepared for future public consumption.
UPDATED 8 AM AUGUST 1
For the historical record, here are the first 5 day Outlooks
TROPICAL WEATHER OUTLOOK NWS NATIONAL HURRICANE CENTER MIAMI FL 500 AM PDT THU AUG 1 2013 FOR THE EASTERN NORTH PACIFIC...EAST OF 140 DEGREES WEST LONGITUDE.. THE NATIONAL HURRICANE CENTER IS ISSUING ADVISORIES ON HURRICANE GIL...LOCATED WELL TO THE SOUTHWEST OF THE SOUTHERN TIP OF THE BAJA CALIFORNIA PENINSULA. 1. SHOWERS AND THUNDERSTORMS ASSOCIATED WITH A LOW PRESSURE SYSTEM LOCATED ABOUT 775 MILES SOUTH-SOUTHWEST OF THE SOUTHERN TIP OF THE BAJA CALIFORNIA PENINSULA HAVE CHANGED LITTLE IN ORGANIZATION DURING THE PAST SEVERAL HOURS. THIS SYSTEM IS MOVING WESTWARD AT 10 MPH AND CONTINUES TO HAVE A MEDIUM CHANCE...50 PERCENT...OF BECOMING A TROPICAL CYCLONE DURING THE NEXT 48 HOURS. CONDITIONS ARE EXPECTED TO REMAIN MARGINALLY CONDUCIVE FOR DEVELOPMENT AFTER THAT...AND THIS SYSTEM HAS A HIGH CHANCE...60 PERCENT...OF BECOMING A TROPICAL CYCLONE DURING THE NEXT 5 DAYS. FIVE-DAY FORMATION PROBABILITIES ARE EXPERIMENTAL IN 2013. COMMENTS ON THE EXPERIMENTAL FORECASTS CAN BE PROVIDED AT...USE LOWER CASE... http://WWW.NWS.NOAA.GOV/SURVEY/NWS-SURVEY.PHP?CODE=ETWO FORECASTER AVILA ABNT20 KNHC 011132 TWOAT TROPICAL WEATHER OUTLOOK NWS NATIONAL HURRICANE CENTER MIAMI FL 800 AM EDT THU AUG 1 2013 FOR THE NORTH ATLANTIC...CARIBBEAN SEA AND THE GULF OF MEXICO... TROPICAL CYCLONE FORMATION IS NOT EXPECTED DURING THE NEXT FIVE DAYS. && FIVE-DAY FORMATION PROBABILITIES ARE EXPERIMENTAL IN 2013. COMMENTS ON THE EXPERIMENTAL FORECASTS CAN BE PROVIDED AT...USE LOWER CASE... http://WWW.NWS.NOAA.GOV/SURVEY/NWS-SURVEY.PHP?CODE=ETWO $$ FORECASTER FRANKLIN/ZELINSKY
The disturbance mentioned in the Pacific Outlook is Invest 90, which is east-southeast of Hurricane Gil.
The Atlantic is quiet at this time and, as indicated by the Outlook, apparently for the foreseeable future. Note that there is a large outbreak of dust from Africa spreading across the tropics. But even before the dust storm, there wasn’t much of anything in the tropics. Go back three or four days and there was the remnants of Dorian and… not much of anything else. That’s not to say “and the usual assortment of tropical waves”. There was one wave just west of the Cape Verde Islands that was already sparse on convection before it was consumed by the dust surge. Now, there isn’t anything in the path of the dust to mow down or suppress. As Michael Watkins said in his tweets that were included in Brendan Loy’s post, the dust surges are annual occurrences and their suppressing effects are temporal. Furthermore, the suppressing effects of the Saharan Air Layer are a matter of debate. From the conclusion of “Reevaluating the Role of the Saharan Air Layer in Atlantic Tropical Cyclogenesis and Evolution“, a paper published in the June 2010 issue of the Monthly Weather Review:
The results of this study suggest that the SAL has perhaps been overemphasized by some in the
research community as a major negative inﬂuence on tropical cyclone genesis and evolution.
In fact, the evidence appears to be more to the contrary in that the Sahara is the
source of the AEJ, which acts as both a source of energy for AEWs and a source of strong
background cyclonic vorticity, and there is evidence of a positive inﬂuence through an induced
vertical circulation associated with the AEJ. To the extent that the SAL may be a negative
inﬂuence on storm evolution, one must recognize that the SAL is just one of many factors
inﬂuencing tropical cyclogenesis and evolution in the Atlantic. Each storm must be examined
carefully within the context of the larger-scale wind and thermodynamic ﬁelds (either from
global analyses or satellite data), particularly in terms of other sources of vertical wind shear
and dry air (i.e.,subsidence drying versus warming over the Sahara).
As such, it’s fair to say that it’s quiet in the Atlantic and that the SAL is the predominant feature at the moment. It may be a stretch, however, to say it is quiet because of the SAL.
UPDATED 11:00 AM August 3
Early verification of the first 5 day outlooks… the disturbance mentioned in the Pacific Outlook became Tropical Depression Eight, at this time. In the Atlantic, the remnants of Dorian were first mentioned in the 2:00 PM Outlook on August 1, with a 20% chance of development. Subsequent Outlooks took it up to 30%, then 40%, back to 30%, then jumped up to 50% in the 8:00 PM Outlook on Friday and then to 60% in the following forecast. At 5:00 AM today, the National Hurricane Center resumed advisories on Dorian as a tropical depression. The revival is expected to be quite brief, with the depression forecast to become a remnant low tonight.