September 19, 2018

Time to Drop the Saffir-Simpson Hurricane Scale

One of the most familiar aspects of hurricane season is the constant talk about the strength of developing storms using the Saffir-Simpson scale, which ranges from Category 1 (a marginal hurricane) to Category 5 (a powerful monster).

Today's Saffir-Simpson scale is based only on the maximum SUSTAINED wind of a hurricane, the average wind over a few minutes and not the peak gusts.

As we shall discuss below, the S-S scale is an anachronism that poorly communicates the real threats accompanying hurricanes, and in fact can be quite deceptive, resulting in people being unaware of the real dangers that can threaten them.  It is also too hurricane centric, not highlighting major threats from "lesser" tropical storms and disturbances.

Hurricane Florence is a good example of the problem.  It reached the Carolina coast as a marginal category 1 storm, and few inland location experienced hurricane-force winds (74 mph or more).   But because the storm slowed to a crawl as it made landfall, very large rainfall amounts (as much as 30-40 inches over five days near the coast) occurred, producing very serious flooding.

So folks in vulnerable areas might have felt complacent when they heard that "only"  a category 1 storm was approaching.    And even weaker tropical storms, not even considered a hurricane, can produce similar levels of precipitation, such as Tropical Storm Allison in 2001 (41 inches in Beaumont, Texas).

Hurricanes can produce serious damage in a number of ways:

1.  From the effects of the strong sustained winds and gusts damaging buildings and other structures.
2.   From storm surge, in which the hurricane winds push water up on to coastal regions and into rivers open to the ocean.  Storm surge is generally the most damaging aspects of hurricanes.
3.  Heavy rain, which can result in flooding, both near rivers and in low areas.

Storm surge is the greatest killer in hurricanes

The problem is that the Saffir-Simpson scale only quantifies the winds.

But the deficiencies of the S-S scale are far worse than that:

1.  It does not quantify the size of hurricanes, which hugely impacts the potential to do damage.  Hurricanes vary substantially in size, with the big ones obviously able to cause far more extensive damage.
2.  It does not quantify the amount of storm surge, the real killer.
3.  It does not quantify the amount of rainfall over any period.
4.  It does not quantify the level of flooding in the interior.

And it is worse than that.  

The effects of identical hurricanes can be very different depending on their speed of motion.  Slow moving storms like Harvey or Florence, did major damage because they moved very slowly, allowing big rainfall accumulations.  If they had moved more quickly, as for most storms, their effects would have been radically less.

And worse than that.

The effects of a storm can vary by terrain and coastal bathymetry (variations of water depth offshore), allowing identical storms to have differing impacts depending on where they hit.

Replacing the Saffir-Simpson Scale

The S-S scale made some sense when atmospheric and hydrological sciences were in a primitive state; when we lacked the capabilities to model and forecast the details of tropical storms and hurricanes and their impacts.

But times have changed.  Our ability to forecast hurricane tracks out nearly a week is now stunningly good.  And closer in time, we have good skill in predicting wind, rainfall, flooding, storm surge and the like, including the geographical distribution of the threats.     That is why the WeatherChannel folks and others were pushing the "catastrophic" rainfall and flooding threats during the day before landfall.

So why not drop the problematic and often confusing measure of hurricane strength (and one that neglects often-dangerous tropical storms) expressed in the S-S scale and simply warn folks of the specific threats, such as high-wind warnings, heavy precipitation warnings, storm surge warnings, flooding warnings.  And as we develop better probabilistic and uncertainty information, that can be communicated.

Forecasters can simply say a tropical storm or hurricane is approaching ("HURRICANE WARNING")  and describe the specific forecast threats and how they vary in time and space.

Here on the west coast of the U.S., we often experience the landfall of Pacific cyclones that are the equivalent in damage potential to minor or even major hurricanes, but we do well with providing the specific threats, without any categories.    Categories for tornadoes (the Fujita scale) are probably fine, since tornadoes are only associated with one type of threat (wind damage).

Probably OK.
If we do drop the Saffir-Simpson, there is one group that will probably complain:


  1. Sorry - that doesn't make any sense to expect the science of meteorology to drop this... It's effective at measuring the central pressure and wind speeds of a hurricane. And we already have countless other tools for expressing storm surge, tornado threat, and flooding risks.

  2. As you point out, Cliff, there are mid-latitude cyclones, such as the Columbus Day Storm in 1962, and "Ali" that just went through Ireland and North Britain today, that approximate hurricanes in damage but are dealt with by forecasting. Even with 1962's technology my parents and my school district had some inkling of the Columbus Day storm in advance of it hitting. Would it have helped if it had been characterized a Category 2 or 3 storm? Not likely. Those in Ireland certainly were aware of all of Ali's dangers without categorization. And did it benefit those in Northern Luzon to have the focus on "Mangkhut's" categorization as opposed to it's ability to drop an enormous amount of water? You make a good argument with the evidence.

  3. Good point about dropping the current measurements, but I'm not so sure the categorical warning would have made much difference to those who stubbornly refuse to bug out when they're told to. The Governor and other authorities made it crystal clear innumerable times for those in the expected flood zone areas that they absolutely had to get out of there, or else call for emergency assistance in evacuation ahead of the storm. Not much else they could have done.

    1. I don’t think most people understand the nuances of things like how fast the storm is traveling, how it’s path can enhance or mitigate these effects, or much of the details of a hurricane really.

      People hear the rating, and since it’s been ingrained in east coast folks, they unconsciously perceive it as a threat or not.... thinking things like “I’ve ridden out a cat 3, a cat one will be a cakewalk”, etc.

      Even storms around here run into issues, an early season storm is usually motivated by low snow pack, low river/reservoir levels, drier/sturdier soil, yet you have issues like foliage on the trees still (more drag) and drier wood making it more prone to snapping.

      But as we often see, a storm can become a perfect storm of sorts depending on the weather leading up to it. Saturated soils, low elevation snow, full reservoirs/rivers, wind direction/lack of recent storms, can all effect the storms impact.

      I think we’d be fine getting rid of the hurricane categorization, but the public likes things bite size. We would need to come up with a different scale.

      Also, as long as social media is as big as it is, it needs to be utilized. We have the ability to warn people based off geographic location. It would take some cooperation with Facebook, but weather alerts based on location should absolutely be a thing. You could have a pop up on someone’s feed with the warning and specific info to their area.

      Or, we could just let nature and Darwin take its course...

  4. I would not change it. We (meteorologists) always overreact to events when people die. The forecasts were spectacular, from track to the expected amount of rain. Flooding is a local phenomena away from the coast in these systems, so trying to rate expected impacts is not possible. Rainfall impacts on a large scale was strongly communicated. The answer to flooding is inundation mapping.

  5. Perhaps something akin to earthquake intensity with the Richter scale which is based on peak acceleration values being replaced with Moment Magnitude which measures the amount of energy released. For hurricanes this would be potential energy (air mass volume integrated with velocity gradients) I would assume? Since hurricanes are slowing I would think this should be coupled with localized rainfall estimates.

  6. This is off topic, but I am frequently finding that when I make a comment on a topic, it is being posted under a different topic. (This post is under the Saffir-Simpson Hurricane Scale.)

    Sorry for the distraction, but did not know any other way to bring this to Cliff's attention.

  7. Ian had good points. My take is that a Hurricane's category is what it is at any given point and forecasters give the hurricane's category along its projected path. Its destructive potential is not factored into category levels. At this point in time, there is no real ability to express in a single "category number" the destructive potential of a storm. A lot of individual factors, additional to just the wind speed and pressure, have to be forecast for, as enumerated in these comments and the article, and merged into one forecast number. Hopefully, in the future, a synthesis of all of the factors could be modeled. We're just not there yet.

  8. Water is the real killer, not wind. So I agree that it seems a bit odd to focus only on wind in determining storm strength.

    The risks of inland flooding, in particular, aren't well understood by many people. Given that hydrology, terrain and other aspects of geography are all digitized, at very high resolution, it seems to me that combining that data with rainfall forecasts in a simulation model could lead to flooding forecasts that are very granular and could be queried by street address.

    Telling me a 100-year flood is likely in a river that's a mile away is a lot different than giving me the probabilities for various depths of flooding at My House. But that requires a multi-disciplinary approach to building a flooding simulation model, and not just a weather forecasting problem.

  9. Perhaps instead of dropping the SS scale it could be enhanced by appending additional scales. Maybe A-F for moisture content?

  10. The problem would be comparability. The minute you drop the scale, the global warming cult will be calling every new storm the worst ever.

  11. Smart people must be working on this question. Have alternatives been proposed? An index based on winds, total energy, size, and moisture content perhaps. Put a grad student to work, Cliff!

  12. I would advocate for a "Modified Saffir-Simpson Scale" wherein a few more columns denoting storm map speed (projected), storm surge, rainfall per hour are added. This could then also be calculated to the topographical lay in the pathway models.


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...