Texas Supercomputer Upgrading the Hurricane Forecast

aarondubrow writes "Researchers used the Ranger supercomputer to test a new, high-resolution hurricane forecasting system that incorporates Doppler radar data from planes flying into the storm. The forecasts were shown to improve intensity predictions by an average of 20 to 40 percent over the official forecasts of the National Hurricane Center (NHC). The prediction system is being hailed as a breakthrough and is one of a handful being assessed by the NHC to become part of the operational forecasting system used in emergency situations."

Comments

Badly needed

[dtmos]

As TFA states, while the location predictions have been improving significantly, the best hurricane intensity predictions are only slightly better than what can be obtained from a Ouija board. (No offense intended to those in the field; I know it's a tough problem.)

Just defining "intensity" in a useful way can be difficult. For example, if Storm A has a region in the Northeast quadrant with 100 mph (161 km/h) winds, but elsewhere winds do not exceed 80 mph (129 km/h), and Storm B has 100 mph (161 km/h) winds in all four quadrants, both have the same max wind speed. Which is more intense? What if Storm B has 95 mph (153 km/h) winds in all four quadrants? What if the two storms have the same wind speeds, but are different sizes? If Storm C has lower wind speed than Storm D but, due to its slower forward speed or other reasons, drops five times as much rain, which one was the more intense storm?

When I counsel high school and college students, I always tell them to "work on important problems." Even though I make a point of saying that the definition of "important" is "what's important to you," I am always asked for examples of "important problems." Getting better hurricane intensity forecasts is one of the examples I always mention.

Re:Badly needed

[Captain Splendid]

Which is more intense?

No one cares. Honestly. Because anyone who lives in hurricane zones is already well acquainted with quadrants and which side of the hurricane they'd rather be on when it passes by.

As long as this is an overall step up quantifying and predicting hurricanes, those little details like that will only matter to the specialists and parsers such as yourself. Everybody on the ground will just be happy that things got a little more refined and predictable.

Re:Badly needed

[Yobgod Ababua]

"Just defining "intensity" in a useful way can be difficult."

It doesn't seem that bad. Most of the relevant disaster-planning data can be expressed by giving both the average windspeed and the maximum windspeed. That lets people know both what they're likely to expect and the worst case they should plan for.

I strongly suspect, however, due to the rotational nature of hurricaines, that windspeed does not actually vary over "quadrants" very much, but is instead strongly correlated with radial distance from the "eye". If this were not the case, the energies would disperse rather quickly in different directions. Since they -are- naturally symmetrical, that makes it easier to pick an arbitrary point to measure at for formal "intensity". I should go ask my climatologist friend... he flies into hurricaines.

Size, precipitation, and land speed are also useful things to know, but not particularly related to intensity itself.

Re:Badly needed

[icebike]

Everybody on the ground will just be happy that things got a little more refined and predictable.

But who on the ground even knows that things got more refined and predictable?

After all, everyone on the ground sees the same news media clowns struggling to stand upright in the same puddle of water getting lashed by
winds in areas especially selected for their wind tunnel effect. The media is constantly preaching of doom and gloom and great destruction from storms which are ALREADY predicted via current methods to be largely spent by time of arrival. Disaster theater.

When will you ever see a government agency suggesting people just baton down the hatches and ride a storm out, because our prediction model says it will not amount to much? It will Never Happen, post Katrina. And it leads to near universal disregard for official predictions.

Official predictions of "no big deal" only happen after the storm has made landfall and pretty much spent itself. The next state over, seeing the coverage of light damage from the prior state hunches their collective shoulders and says "Meh" long before the official predictions catch up.

It seems we crowd source hurricane predictions once it gets near to making landfall.

If this new modeling can tell more accurately predict intensity days before landfall, would government (let alone the press) act any differently?

Re:Badly needed

[dtmos]

No one cares.

C'mon -- you can do better than that. Things can't get "a little more refined and predictable" if the phenomenon we are predicting is incompletely defined, to the point that no two people have the same understanding of the concept.

Besides, anyone who lives in hurricane zones is also well acquainted with the fact that the present intensity forecasts are terrible, which leads to cycles of over- and under-preparation by the populace. It would be nice if we could tell, say, three days in advance whether a Category 1 or a Category 4 storm were going to hit, so that we could determine the appropriate evacuation zone. At the moment such forecasts are not the paragon of reliability.

Re:Badly needed

[VoiceOfSanity]

Before Hurricane Katrina made landfall, the storm surged in intensity from a Category 3 to a Category 5 storm. At that point, the National Weather Service in New Orleans issued out what has since been referred to as the "Doomsday" message, making it graphically clear what was facing the region when Katrina struck. It was the first time that the NWS went to this extreme a message, and made people realize just how real the threat was.

Having lived in the region at that time (panhandle of Florida) I can tell you that people took seriously the threat as we had evacuees streaming through the area.

Re:Badly needed

[Deus.1.01]

Lets see, sea vessels(fishing, shipping, etc), air travel oh...and a flash flood are really....flashy in a watery grave, infrastructure damaging sorta way.

Re:Badly needed

[dtmos]

It doesn't seem that bad.

Um, sorry, but speak to your climatologist first. These storms are big, and the problem is that, when they come ashore, each location will experience a different wind profile. Thus there is no single "average" wind speed to forecast upon landfall -- every location will experience something different, and listing some kind of two-dimensional overall average of the storm isn't much help: Not only do the storms vary significantly in size, but the size of the eye varies significantly, too. To make matters still worse, there's a phenomenon called eyewall replacement, resulting in an effective eye diameter that varies in size over time (along with the maximum wind speed).

It is true that there will be a maximum wind speed in the storm, and that it will usually -- but not always -- be in the right side of the eye wall for storms in the northern hemisphere. However, at the moment there is no detailed forecast of the size of the area having those peak winds, so it's very difficult for emergency officials to plan for the devastated area.

Also speak to your climatologist about your misunderstandings about hurricane wind fields. They are rarely symmetrical, due to interactions with approaching weather fronts, other nearby storms, forward movement of the storm itself, and other factors. A truly symmetrical storm, like Katrina in the Gulf of Mexico, is notable because of its symmetry. The US Navy tropical storm forecasts include the radius of 34-, 50-, and 64-knot winds away from the center for each quadrant of the storm, and you will find differences of many tens of nautical miles between quadrants.

Re:Badly needed

[dtmos]

Yep. And you will notice that they did not predict the intensity surge from Cat 3 to Cat 5: They only issued the "Doomsday" message in response to the storm's increase in intensity. It would have been nice if the surge were forecast a day or two earlier, so that a more orderly evacuation could be made.

If it had made landfall as a weak Category 3 -- especially if the high wind field was over a small area -- and the storm made landfall in a sparsely populated area, e.g., Franklin County, everyone in the more populated areas would have been greatly annoyed at the "false alarm" sent by the Weather Service. Since the NWS' Intensity forecasts can't predict eyewall replacement cycles, it had no way of knowing in advance whether a Cat 3 or Cat 5 storm would make landfall.

Re:Badly needed

[Canjo]

But who on the ground even knows that things got more refined and predictable?

After all, everyone on the ground sees the same news media clowns struggling to stand upright in the same puddle of water getting lashed by winds in areas especially selected for their wind tunnel effect. The media is constantly preaching of doom and gloom and great destruction from storms which are ALREADY predicted via current methods to be largely spent by time of arrival. Disaster theater.

This is indeed a problem, but when you live in a Hurricane-prone area you typically aren't watching the national news, which is trying to make $$$ by making a spectacle. You're probably watching the local news stations which are relatively more informative. In fact, when Hurricane Gustav hit New Orleans in 2008, I remember the guy on the local Fox station explicitly told us NOT to watch Fox News, since it was just sensationalist and trying to scare people.

In any case, it doesn't hurt to err on the side of caution with these things.

Improve by 40%?

[Anonymous Coward]

...to improve intensity predictions by an average of 20 to 40 percent...

Can someone more knowledgeable in this area explain this in laymen's terms?
Can they now predict whether or not a specific tree will be uprooted? Or just give a 40% more accurate estimate for damage in an area?
Or are they now 40% less off when they predict the wind speeds of a hurricane?

Curious,

AC.

Upgraded to Guess from Wild Ass Guess.

[trout007]

Some more advances and we will be at Educated Guess.

Re:Upgraded to Guess from Wild Ass Guess.

[sakdoctor]

It never will, due to cumulative uncertainty in chaotic systems.

Re:Upgraded to Guess from Wild Ass Guess.

[mveloso]

I suspect that, at the Macro level, chaotic uncertainty is overwhelmed by statistical probabilities.

Re:Texas Supercomputer?

[bussdriver]

That is misleading; I've heard other groups say texas has been a loser state for longer. Depends on what one measures I suppose...

Federal indirect spending is probably not being counted by them; and many things leave out the super massive corporate welfare which is not an easy thing to total up (in part because it is corruption related that whole area benefits from obscurity) while all the social programs are well tracked. Corporate welfare beats out welfare by a huge margin (medicare and social security are not welfare programs; despite the movement to mislabel them and get people to think they are part of the general fund.)

Re:Texas Supercomputer?

[sycodon]

You believe whatever you have to believe to maintain that fragile sense of superiority ya have going on there.

Re:Automation, tok r jerbs

[icebike]

Why not replace the hurricane plane with a bunch of drones equipped for the task. Then you could have a half dozen flying around the hurricane at the same time. Yes that is just moving the pilot from the cockpit to command room (or where ever they sit), until they make them autonomous.

Given the loss of life record for hurricane hunter planes, the case for drones simply hasn't been made.

The last hurricane hunter lost was in 1974, the prior one in 1955. Far more of these planes were damaged on the ground than in the air.

Its doubtful drones would survive anywhere near as successfully, or produce anywhere near as much good data. Storms change slowly. An hour by hour measurement system is not warranted, and all of the tracking work (and an increasing amount of the measurement work) can be done by satellite.