September 14, 2018

Hurricane Florence: Stunningly Good Track Forecasts, Problems with Intensity

Hurricane Florence is has recently made landfall near Wilmington, North Carolina as a category one storm with sustained winds of approximately 85 mph (see latest radar image below).  It is now weakening rapidly.

Very few locations over land have experienced hurricane force winds (SUSTAINED winds of 74 mph or more) so far, but several coastal locations have received gusts of 70-90 mph (see max gust map below)

This storm is a "classic" on how well our forecast models have done:  superb track prediction and mediocre intensity forecasts.  And the U.S. models did very well on track--better than the European Center model.

To illustrate, here are the forecast tracks from several models initialized on September 9th (last Sunday) at 11 AM PDT.  Stunningly good prediction....aiming the hurricane landfall on the southern N. Carolina coast.  All the models were doing the same us some confidence in the track.
Absolutely marvelous. 
 Looking at the position errors of the storm (which includes the position along the track as well), shows increasing position errors in time for the various models, with the U.S. GFS model being the best.  Nice to see.  (US GFS is AVNO and the European Model is ECMF, plot produced by Professor Brian Tang, U. of Albany)

Clearly, track forecasts are very, very important.   Anyone with a bit of sense will prepare or evacuate (in vulnerable areas like the coast) if a hurricane is heading towards them.

But it would be nice to get intensity right as well.    For the same 120 hr period shown in the track figure above, here is the intensity (central pressure of Florence) that was observed (black line) and forecast (colored lines).  The observed storm strengthened rapidly during the early period, leveled off, and then weakened .  The operational global model (GFS, shown by the blue AVNO line) was initially too weak and was too strong at the end.  The new operational model (FV-3) was not deep enough for most of the storm.  The high resolution US hurricane model (HWRF, purple line) was way too deep more of the time, including at landfall.

As I have noted before, getting track right demands getting the large scale flow correct--which we can pretty much do today.  But getting intensity right demands we simulate the internal dynamics of the storm at high resolution...and do it correctly days ahead.  Very difficult to do...and may be impossible according to some research (including research done at the University of Washington by my colleague Greg Hakim and students.)

One final piece of interesting hurricane information.  Do you think the number of landfalling storms on the U.S. continental (mainly the SE US) is increasing, decreasing, or staying the same?   I suspect many would say increasing, based on the considerable hype in the media.

The truth?  A slow decline during the past few decades as shown by several papers in the peer-reviewed literature and government statistics.   Here is a plot (a) from a 2018 paper by Klotzbach et al. in the Bulletin of the American Meteorological Society for hurricane landfalls on the continental US (1900-2017).  Slight downward trend.  Similar trend for major (categories 3 and above) hurricanes.

The big question is what will global warming do to the frequency of US landfalling hurricanes. Some research suggests that the most intense hurricanes will get more frequent, but the total number will decline.   But what will happen to landfalling U.S. storms is something that is actively being studied.


  1. Is the case that accurate intensity models require more distal data than do track models? (available energy being more a matter of what's ...available?)

  2. I read study results that suggested there is no evidence yet of "more" storms in a climate-changed future, but there is evidence for bigger and wetter storms. Your comments?

  3. Very interesting information. I was wondering since we live in an electrical type universe, what if we look at hurricanes as electrically related and measure them as such and connect the dots to intensity of the storm etc. The amount of energy in a storm is more than what can be produced by warm ocean and heat from condensation etc. Can we research it from an electrical paradigm. Cliff I enjoy your blog and thank you for it. Best to you. Rick Ward

  4. Michael Mann, actual climate scientist writing on this.

  5. There is little doubt that forecasting hurricanes is approaching the reliability of forecasting for routine seasonal weather. This is an improvement of reliability of one aspect of risk - probability. It is calculated to happen as a probability of a certain intensity and location and time. From a perspective of risk communication this is an improvement of only one aspect of risk.

    Now consider how that improvement in probability forecasting can create a confidence in the public in their decisions to evacuate or not. The decision to evacuate is a function not just of probability. It is actually more a function of consequence. If there is any residual error margin of probability of either track or intensity of a hurricane , a catastrophically consequence will likely be the result.

    When the risk experts calculate / estimate the four factors of risk, probability is seldom the single factor to focus on, unless calculation of probability is supremely reliable, which it very rarely is. Exposure, vulnerability and consequence are usually more discernible and pertinent in terms of survivability.

    If one is really concerned with the problem of risk communication and the hazard in question is highly consequential then a far greater degree of communication should be directed to describing consequence, vulnerability and exposure.

    I say this because as confidence in probability forecasting increases, the inevitability of error recedes as a factor in decision making.

    Think about this - the risk framing of any hurricane - as a metaphor for how we regard the risk of climate change. We all focus on the probability of it happening or more specifically if it will happen to us and if there is any plausibility it won't, that determines our actions. For decades this is all we can argue about - the probabilities of the hazard happening. We don't spend a lot of energy contemplating the consequences if it does.

    and then acting accordingly.

    An expectation of accuracy in probability might just wind up biting us in the ass when the inevitable error occurs if we fail to consider consequence more seriously.

  6. I would suggest that the evidence for bigger and wetter storms happening currently and over the past few years is inconclusive at best. Certainly we have rain gages in numbers that are orders of magnitude greater than what we had 40-60 years ago, so that now we can actually know in much more detail how much rain has fallen and how it was distributed. Who can know what actually happened in some of the great storms of the past couple of hundred years when there were little or no data available for analysis? Remote sensing capabilities (satellite and radar) and numbers of surface observations are light years ahead of networks that were primitive or nonexistent half a century or more in the past. I will leave the question of what storms will be like in a climate-changed future to those who are much more qualified than I am to address that. First, let's make sure we can make meaningful and accurate comparisons between the big storms that occur today and comparable ones in the historic record.

  7. I "survived" Irma (trop force winds where i was); each time the forecasters and their breathless hype weather reporter cousins "over predict" the intensity....that means the next time a percentage of the population will remember the error and lower their preparations. "They were wrong" last time.

    The NWS info in Irma seemed to comingle weather predictions with "the winds/water could get to this intensity". They seemed to include disaster advice on how bad it could be with the scientific prediction of actual intensity. Thus every report seemed to err on the "bad or high" side "prediction inflation"?

  8. A nice visual of the wind on the east coast:

  9. Ahhhh, an even better visual here:,-79.940,7,m:ezNadVV

  10. Carl Baker,
    What Michael Mann is saying makes no sense at all. The number of landfalling storms --of any intensity--has DECLINED during the past decades. Is this due to global warming? There is no evidence of slowing flow during the past decades, nor a connection with GW. ..cliff

  11. Cliff, seems like there is publish research showing a slowing trend, comment?

  12. What about the issue of fake news.

    ON NBC nightly news Tue, Al Roker said Florence would be "at least a category 4
    if not category 5 hurricane when it hits!". He also showed a graphic indicating 40 inches of rain over Wilmington, NC.

    Well, Florence was only a category 1 hurricane and rainfall as of noon EDT tda in Wilmington, NC is a mere 12.37 inches. Most rainfalls since the storm hit all across North and South Carolina are in the teens or less. Myrtle Beach 6.74". Good for the golf coarse, really.

    In fact, if you looked at the Palmer Index chart (drought chart), northern SC has drought conditions worse than Ca. Needed rainfall is 9-12 inches to bring conditions back to normal. Even Wilmington, NC was below normal.

    5 deaths attributed to the hurricane are a tragedy. Yet, On average, 115 people die per day per state. Hurricanes can save lives by keeping people safely indoors away from driving and other absurdities. (When reporting deaths from storms, they should subtract out how many people would have died normally.)

    Like most weather events, the news media over hyped and overplayed the event.
    Do you think Al Roker and Lester Holt will do a retraction of their broadcast?

  13. Just saying, Florence is stalled right over NC and will continue to deliver rain to the area for at least another 24 hours. 40 inched doesn’t really seem that off, though the intensity was over predicted.

  14. hey, weatherguy,

    I suggest you move to North Dakota. Your kind lives there.

  15. Res warnings: If there was one chance in a hundred that your plane would crash just about no one would fly. For some reason those facing hurricane risks seem to think that a miss means we should all ignore hurricane warnings.

    I was in Chehalis for the Hanukkah Day (actually night) storm. Forecasts were that it would be the strongest wind storm in 10 years. People ignored ignored the warnning in droves. Not smart. And as they returned late in the evening over rural roads blocked by fallen trees even smart people wondered at why they were so heedless. We slightly overheated our house assuming that we would lose electricity, and got out the LED headlamps. A big grove of trees went down, followed immediately by the power.

  16. weather guy - NC got 30 inches of rain in 24 hours and its got another 24 hours to go before the fat lady sings.

    Better wait a bit before you start yelling fake news. As far as what I could see, most forecasts have nailed it pretty good, including the stalling out and the down grade as it hit shore. I continue to be amazed at the remarkable advances in weather forecasting in general and this is no exception.

    These guys have been on the money so far up to 5 days out:

  17. During a discussion on the Weather Channel, one of their Hurricane experts Dr. Nabb I believe, said that basically Florence traded wind velocity for size. So while the wind velocity dropped the actually energy of the hurricane increased dramatically due to how large it became.

  18. Cliff et al,

    I'm in SE China in the midst of Typhoon Mangkhut. My first tropical cyclone experience.

    From the 13th floor of my hotel it seems that some of the precipitation is coming down as snow. Is that possible? I didn't believe my eyes, but have been observing occasional snow flakes for the past 4 hours--it's 3PM here. I can't open the window to check.

    Thanks in advance to anyone that can help me understand this phenomena.

  19. wetherguy - Which is better, news that has things wrong, or no news? After much searching, I finally found the answer.. just turn the whole thing off! Go outside for a nice long walk and enjoy the beautiful Autumn weather.

  20. Weather guy has a point. I've seen the same thing over and over again. Sensational and emotional exaggerated stories about bad weather draw viewers and internet clicks. The media really hasn't changed much since the days of yellow journalism in the 19th century. The last 3 years have shown them to be nakedly partisan as well.

  21. A number of sources estimate that economic losses from Florence will place it in the top 10 hurricanes in inflation-adjusted dollars. I guess you can decide if that is a catastrophe, economically speaking. It appears that except for one storm, all of these have occurred since 2000.

    While hyperbole is not a good thing, perhaps there is some benefit in terms of evacuation response and minimizing loss of life.


Please make sure your comments are civil. Name calling and personal attacks are not appropriate.

The Other Type of Mountain Wave Cloud

 Folks love to talk about lenticular clouds , which are generally produced by air moving up (and down) downstream of a mountain barrier (see...