EIIP Virtual Forum Presentation — May 11, 2011

April 2011 Tornadoes
A "Particularly Dangerous Situation"

Harold Brooks, Ph.D.
Research Meteorologist
Head of Modeling, Observation, and Analysis Team
National Severe Storms Laboratory (NSSL), NOAA

Amy Sebring
EIIP Moderator

The following has been prepared from a transcription of the recording. For ease of printing the session’s slide set (Adobe PDF) may be downloaded from http://www.emforum.org/vforum/NOAA/2011tornadoes.pdf .

[Welcome / Introduction]

Amy Sebring: Good morning/afternoon everyone and welcome once again to EMForum.org. I am Amy Sebring and will serve as your Moderator today. We are very glad you could join us. For our newcomers, we will be providing some instructions as we go along so you can relax and participate with us.

Today’s topic is the tornadoes that struck the Southeast during the end of April. For the three day period, April 25-28 alone, NOAA estimates a total of 305 tornadoes shattering the previous record set in 1974. And, according to NWS Storm Survey teams, there were over 24 killer tornadoes in six states--Arkansas, Mississippi, Alabama, Georgia, Tennessee and Virginia-- including three in the EF-5 category that caused over 300 fatalities.

We are making a recording, which should be available later this afternoon. And as I mentioned last time, we are making an audio only, MP3 version of our programs, which you can access directly from our Website, or by subscribing to a podcast now available from the iTunes Store.

[Slide 1]

Now it is my great pleasure to introduce today’s guest: Dr. Harold Brooks is a research meteorologist and head of the Modeling, Observation, and Analysis Team at the National Severe Storms Laboratory (NSSL) in Norman, Oklahoma.

In 2002, he received the United States Department of Commerce's Silver Medal for his work on the distribution of severe thunderstorms in the United States and, in 2007, he received the NOAA Administrator's Award for work on extreme weather and climate change.

He is currently on the steering committee of the European Conference on Severe Storms and was recently named a Fellow of the American Meteorological Society. Please see today’s Background Page for further information and related links.

Welcome Harold, and thank you very much for taking the time to be with us today. I now turn the floor over to you to start us off please.


Harold Brooks: You’re welcome. The month of April was a very active month for tornadoes in the United States, in particular the last week. I’m going to focus almost entirely on the April 27 portion of that, even though there were killer tornadoes on the 25th and 26th, almost all of them were on the 27th. We are estimating up to 350 people were killed on that one day alone, and there were nine on the other two days, the 25th and 26th. The 27th was the big day. We are still working a lot to try to figure out in terms of the casualties.

I want to talk a little about the forecasting of the day and if you had been watching in the days ahead what you might have seen.

[Slide 2]

In experimental forecasts that the Storm Prediction Center put out the week before indicated that Tuesday and Wednesday would be active days in particular. The first official forecast that looked at that came out on Monday morning. We’re showing the probability of a severe thunderstorm within 25 miles of any location.

The peak probabilities are 45% on this day, which for the outlook that is put out more than 48 hours in advance—that is about as high as the Storm Prediction Center ever goes. In fact, when they put their product on that Monday, they had a moderate risk for severe thunderstorms on Monday, Tuesday and Wednesday.

That was the first time they had ever had a moderate risk on all three days of the main outlook since they’ve gone to three day outlooks about ten years ago. The hatch theory in there—the little diagonal line is indicative of what is called significant severe thunderstorms—2 inch or larger hail, F2 and greater tornadoes or hurricane force winds. That would be highlighting northern Mississippi, eastern Tennessee, and up into Kentucky.

[Slide 3]

As the day went on the next day, the outlook expanded a little further north. In the text product that was associated with it, they discussed that the northern end of it may not be realized. They were concerned about the convection from the thunderstorms from the 26th. If they occurred, that might actually limit the northern extent of the really severe thunderstorms. In fact, that is what occurred.

So it is possible that if the thunderstorms overnight on the 26th and the early morning of the 27th had not occurred, we might have had a larger area on the 27th than we actually ended up having.

[Slide 4]

On the morning of the 27th, the conductive outlook—the first product the Storm Prediction Center puts out, in the upper left-hand corner uses categories of slight, moderate, and high risk of severe thunderstorms. The high risk area highlighted northern Mississippi and Alabama, southern Tennessee, and northwestern Georgia.

The tornado probabilities in the bottom right-hand corner—30% probability in the hatch area now there is at least a 10% chance of a F2 or greater tornado within 25 miles of the location. This forecast would have come out at 2:00 AM EST, 1:00 CST. You can see the major area that was highlighted was northern Alabama and the surrounding areas.

[Slide 5]

Several hours later when that forecast was updated, they increased the probabilities of significant tornadoes and moved things a little bit back to the southwest perhaps for the emphasis. This was a remarkably good forecast. This came out at 11:30 on Wednesday morning, or 10:30 Central Time.

This highlighted the main action. This ended up being the main killer tornadoes, including the Birmingham-Tuscaloosa storm, which was down in here, the Hackelberg and Phil Campbell storm that went up into Tennessee, and the Smithville, Mississippi tornado that was over in this region—those were the biggest three killer tornadoes.

There was another one up in the northwest part of Alabama through Dekalb County, and a fourth one that was later in the evening that came out of northwest Georgia and went up through southeast Tennessee. There were really five major killer tornadoes that killed more than 20 people each.

[Slide 6]

As the day went along, tornado watches were issued. These are four of watches. There were a total of 16 issued during the day. I have highlighted the four in the main area—they are in time order of issue from the upper left hand corner to upper right hand corner, and to the lower left hand corner and the bottom right hand corner. This display shows the counties that were in the tornado watches and the radar pattern at the beginning when the watch was first issued.

The times the watches were issued are below them. You can see in the upper left hand watch—this line here of strong storms is the remnant of the overnight convection that lead to a few deaths in northeast Mississippi and later on into fatalities off the screen in northeast Alabama. There have been two sets of storms the night before.

Now the new storms are starting to form back along the Mississippi River. This is one of the lines of storms that would lead to fatalities. You can see the faint hint of radar echo that is associated with another line of storms and in between the beginning of what will be the third line.

These are the three primary lines of storms during the daytime. The second watch that came out covered most of Alabama. This was issued almost three hours later. You can see the storms that have already formed in Mississippi that are going to become killer tornadoes. A couple of hours later a watch was issued to cover middle and eastern Tennessee.

At this time you can see down in this area the storm that produced the fatalities in Chickasaw and Monroe Counties in Mississippi. This is the storm that produces the Hackelberg and Phil Campbell tornado and the beginning that comes up through Dekalb County in northeast Alabama.

We are beginning to see the storm that produces the tornado at Tuscaloosa. This is the storm right there in its early stages that produced all the dramatic video. In the final watch that is issued, that was really relevant to most of the major fatalities is the one that covered most of Georgia and into southwest North Carolina. This was primarily valid into the late night hours.

Those were the four main watches. There were a total of 16 that covered most of the rest of the severe weather on the day. These were the four where most of the damage occurred.

[Slide 7]

Victor Gensini, a graduate student and the University of Georgia put together this very nice map of where all the tornado warnings were issued during the day. This is a nice visual of where action occurred. There were tornadoes as far north as New York State.

The fatalities in the Virginia were in the southwestern part, in this line of storms. You can almost pick out the lines of storms as the move northeast. We get a real picture that north central Alabama was a major target area where a lot of things occurred. When you get into east Tennessee where all the polygons associated with the warning cover it where you can’t see the map in the background.

You also see one thing that perhaps lessened the impact of the event where it could have been even more catastrophic—you see storms that went south of Atlanta and north of Atlanta, but the metro area wasn’t impacted much by the storms. It was almost split on the way going around.

There were 600 tornado warnings issued during the day. It was a large number covering a huge area in the eastern part of the United States.

[Slide 8]

In the initial reports of that day, these are referred to as the preliminary reports of the day, there were 292 tornado reports. It is important to know that many of those are duplicate reports. A tornado on the ground for 60 miles may have 10-20 or more reports associated with it—different people seeing it and reporting it into the system.

That gives you an idea of some of the long track events down here in Mississippi. You can see this long line here that is probably associated with one tornado. There are another couple that come out through northwest Alabama and northeast Mississippi.

The tornado that hit Hackelberg, which is in Marion County, Alabama, actually went all the way up into southern Tennessee and was on the ground for about 130 miles. There are a lot of reports from that particular one.

The blue dots are wind reports and the green dots are hail reports. One thing we have noticed in the past is there are very few hail reports that make it into the database of severe weather reports. When there are large tornadoes on the ground in the vicinity, people tend to concentrate on the tornadoes.

This was initially the 292 preliminary reports. We’re probably going to end up for the one day with somewhere in between 150-170 final reports, as some of these multiple reports get put into one tornado. As damage surveys are done, they sometimes find tracks of additional tornadoes as they were going along.

[Slide 9]

The killer tornadoes on that day—right now we believe there were 25 tornadoes that killed people. This graphic was put together showing those tracks. We’ve gotten a little more information that has lowered the death toll at the moment to about 305, although we are still trying to calculate some things.

There was probably an additional death in northwest Georgia that is not in the database yet. The primary killer tornadoes—the numbers you see—there are numbers that indicate the time in UTC time—subtract five hours to get central daylight time, or four hours to get eastern daylight time.

This was 22:20 to 00:00, so that was 6:20 PM to 8:00 PM, and the black numbers are the numbers of deaths in a particular county. For instance, there were 39 deaths in Tuscaloosa County, and 20 in Jefferson County which is where Birmingham is. This is the Tuscaloosa-Birmingham tornado. Fifty-nine deaths are associated with it right now.

The very long track tornado that comes out of Marion County that starts with 24 deaths there and 26 in Franklin County. Right now I think there are 75 deaths are what we are estimating right now. In another tornado, there were 26 fatalities in Dekalb County.

These are the paths of the tornadoes. Killer tornadoes went up as far as Virginia. It does not include the stuff that happened in the very early morning of the 27th or on the 25th or 26th—there were killer tornadoes as far back as central Arkansas. That is a picture of what happened on the day in terms of the impact.

[Slide 10]

I want to do a little bit of meteorology and show you what things looked like. This is an image of a couple of different tornadic thunderstorms. The one of the left is the May 3, 1999 Oklahoma City tornado that went through Moore and Bridge Creek and Oklahoma City, very near me. The one of the right is the Tuscaloosa tornado as it is going through Tuscaloosa.

You can see where the radar is located—the little black dot. These are essentially the same size. The horizontal scale is the same on both maps, so you see the storms are about the same size. You see the characteristic classic hook echo. Down in the part of the hook, you’ll see if you look very carefully, there are pieces of purple—that is actually debris.

In the Oklahoma, this is the west side of Moore, Oklahoma getting hit. In the Tuscaloosa, this is Tuscaloosa getting hit. Pieces of building materials look bright on radar—they are very good reflectors, so they show up. The rest of the colors—the reds are heavy rain and we go down to lighter rain with the yellows and the greens.

These two storms look remarkably similar. We see the remnants of a northern storm in Tuscaloosa and the beginnings of another storm back off to the west, but they look very similar.

[Slide 11]

If we look on the larger scale, a product that has recent been developed at the Severe Storms Lab—it’s what we call rotation tracks. You look at velocity data out of the radar and you see where velocity data says we ought to have something that is rotating. It doesn’t identify a tornado, but it identifies rotating thunderstorms.

We can give you an idea of the scope of the outbreak last week. All of the yellowish background lines—you see on the right hand side of the map—this goes from the 25th to the 27th. Each of these lines is an individual rotating thunderstorm.

This storm that I’m drawing line along has a rotating component to it starting in Mississippi and all the way across Alabama. The one to its north continues all the way up through Georgia and into southwest North Carolina. There is another one if you follow it all the way up into Virginia with strong rotation over about a 300 plus mile path.

To put this into comparison, the green part over on the left side is the rotation track associated with the May 3, 1999 outbreak. It is missing a storm because we don’t have the data off the radar that was in Wichita at the time. It is missing a rotation track up in Wichita. It was a very large impact tornado outbreak with over 60 tornadoes on the day—and just for comparison, the geographic area that is covered by the rotation tracks from last week is an amazing thing.

Last week was a certainly a very large outbreak, probably one of the six largest outbreaks in U.S. history and the largest one since 1974. There were tornado tracks that right now they will probably end up with 2,000 miles in total path links of tornadoes. That compares to the 2,600 miles on the April 3, 1974 super outbreak that actually hit some of the same counties.

In southeast Tennessee, there were fatalities on the same street on April 3, 1974 and last week’s tornadoes. Some of the same places hit, but there were more damaging tornadoes on April 3 up into Ohio. That is sort of a large scale picture.

[Slide 12]

This damage picture is out of Hackelberg, Alabama, a city where there were I believe right now 18 fatalities. In the background were steps that led up to a church. There is nothing left of the foundation other than the steps going up to it.

[Slide 13]

This is the Wrangler jeans distribution facility in Hackelberg. One of the long term impacts is that this was the primary employer in the city and it is obviously not in very good shape anymore. The parent company of Wrangler is trying to figure out where they can put a facility close to Hackelberg, at least temporarily, so people can go back to work soon.

There was one fatality in this building when it collapsed. You can see how walls have fallen and cars have been thrown into the walls and the walls have fallen onto them in the parking lot. There was only the one fatality, but in terms of economic impact, if they can’t get people back to work, this was by far the largest employer in the town.

[Slide 14]

Phil Campbell, a small town just northeast of Hackelberg in the next county—this is a house that has lost all the flooring and taken off of the foundation. You can also see a lot of tree damage. This is indicative of at least F4 damage.

I’m not sure if we looked at F5 for this building, but you can see where in the middle the carpet has gotten rolled up by the tornado. It managed to stay attached on one end, but on the other end, the carpet was actually detached from the flooring and has been rolled up in the middle of the house.

[Slide 15]

This is an image from Smithville, Mississippi where there were 15 fatalities, lots of damage and lots of debris. This is one of the images that may become one of the images of the damages—they found a Ford Explorer that had blue paint on it and they couldn’t find any blue paint in the vicinity. Someone noticed that the water tower had been hit by the Ford Explorer. It was thrown up in to the big thing on top of tank of the water tower.

There is actually a little bit of paint missing at the top of the water tower. Someone went up there and they’ve done some chemical analysis and indeed, this Ford Explorer hit the tank of the water tower and bounced away. It was unoccupied at the time, fortunately. That gives you an idea of the kind of power associated with these. Smithville was another place that was rated with EF5 damage.

[Slide 16]

This is a housing complex in Tuscaloosa. You can see a wide gradation of damage here. The buildings on the south side are F2 damage at most—roofing material has been lost off of the one on the right end, but survivable without significant injuries in most of the bottom buildings, up to where you have all the roofing material gone, which is indicative of F3 damage.

There are a few vehicles that have apparently been thrown, perhaps into the mess of the housing complex. It is likely that people in here, if they had been in small rooms or shelters, unless something like a car fell on them, would have survived. There would have been injuries, but this was certainly a survivable event.

This is what a lot of Tuscaloosa looked like over a large scale.

[Slide 17]

I’ve attempted to piece together some of the aerial images that were taken by NOAA from the coastal geodetic surveys and this is actually a path through Tuscaloosa. You can see the Bryant-Denny Stadium where the University of Alabama plays football up on the left. The tornado path is mostly south of the campus area.

You can see the track of the tornado—virtually this line that goes through. I know there were fatalities at the overpass here on the highway on the southwest side of town. In the light area in this line there is a lot of damage and piling up of debris along the center line. The coastal geodetic survey has done over flights over several of the paths and this is a composite of four of their images that are put together.

They covered the entire damaged path on the Tuscaloosa storm, over Smithville and over several of the other tornadoes—some really amazing images that are high quality and high resolution. These are actually the low resolution images that I put together. I had to scale them down to get them to fit.

In high resolution, I believe the resolution is less than a meter on the images they offer at this point. There are lots of data that can be provided with this information.

[Slide 18]

That is an overview of last week’s event. Amy asked me for a little bit on what might be happening in the future that might help us. One of the things we do a lot here at the Severe Storms Lab is looking at radar—Doppler radar that was deployed into the field in the early 1990s.

It played a huge role in quality warning. Over 90% of the tornadoes and all the killer tornadoes had warnings issued prior to the tornado occurring. On average, the time between the warning and the tornado was 24 minutes of the entire outbreak.

In the future, radar will continue to improve. One of the things that is going on now is polarimetric radar. The big benefit out of polarimetric radar is it improves precipitation estimates. It allows us to discriminate between different precipitation types—heavy rain and hail. There are two beams out of the radar, and you can do all kinds of wonderful things with that.

It turns out it makes it more obvious to see tornado debris. Tornado debris doesn’t look like hail or rain. The polarimetric radar signature of debris is strong and may work when you’re not seeing the debris ball—the purple dot that I showed you earlier.

[Slide 19]

The other radars I’ll show you examples from are the phased array radar and the gap filling radar. I’ll show you some images from storms that occurred a year ago yesterday very near me. One of them touched down 300 yards south of my office.

This is an image from an experimental phased array radar we have here that shows reflectivity on the right. In this case, we are zoomed way in. We are basically looking at the tip of the hook from the other storm. You can see the hook coming around and this really bright debris ball that is a few kilometers across.

On the left side is the velocity signature, so you’ve got winds blowing away, indicated by the red, and winds towards the radar indicated by the green. This is the rotating signature associated with larger scale rotation. In this small area there is an enhanced signature that is associated with the tornado itself. We are almost looking inside the tornado.

Phased array radar is a military technology that has been around for a number of years. Current, conventional radars have a dish that spins and collects data as it spins. It makes a circle, you tilt it up a little bit, it makes another circle, you tilt it up and it makes another circle—and it takes about five or six minutes to collect an entire volume of radar data information.

Phased array radars, instead of being mechanically steered are electronically steered, and it has a lot of beams that come out of it. You essentially can point them almost instantly at what you want to look at it and collect an entire volume of data. To process it takes less than a minute.

You can lots of other things with the radar while you are doing that. You can tell a few of the elements of the radar to look in a particular area. If there was a tornado, while you still want to know what is going on in the rest of the areas so that you didn’t get caught by surprise, you could spend a little bit of time in the region of the tornado and perhaps get updates on it every ten to fifteen seconds.

That is incredibly valuable. It also has the capability of doing non-meteorological things very well. It can be used as an aircraft tracking radar, which is what the Navy on the ships use it for—looking for friendly or unfriendly aircraft approaching.

As a result, it is possible that the phased array radars could be in place in about ten years and might replace about a set of seven radar networks that exist in this country for weather and aviation purposes, and be able to do both at the same time, which would be a major cost saving in the long run and provide much better data and information so people may make better warning decisions when they do that.

[Slide 20]

Another image from a second phased array radar that is experimental in the Norman area—you can see three different circulations. The left panel is velocity signatures off of the terminal Doppler radar in Oklahoma City which is intended primarily for warning of aviation hazards at the Oklahoma City Airport.

You can see three velocity signatures—one very large one. This is the one we were just looking at from a different angle and at a slightly different time. There are two more down here on the bottom left. Associated with those are the reflectivity signatures from another phased array radar. To the right you see the hook echo from the main part of the storm and another circulation east of that associated with this velocity signature.

This part is moving northeast and produced a relatively weak tornado. The western signature continued to move east and killed one person. Another tornado killed someone else later on during the night. My office is right about there, so this was one that was relatively close to us.

The reason I couldn’t show an earlier image was because the people who run the reflectivity radar for some reason decided to abandon their posts as the circulation passed just south of them. They turned off the radar for a minute and didn’t collect data for a few minutes, and then they went back to it.

We can see really spectacular images and this uptake time of a minute is a really interesting concept that should add a lot of value.

[Slide 21]

The other kind of radar that exists that is being tested here in the area is called CASA. It’s an adaptive sensing system where a set of radars—right now there are four in the network—are used in conjunction with each other.

Instead of just having one radar looking where it ought to look, the four radars intelligently communicate with each other and decide, based on a set of rules, what the most interesting weather event is and focus resources there and get a picture within an area. This is being compared to current Doppler radars.

These small radars are gap filling radars that aren’t very strong, but you can put them in between where current radar coverage is and because right now one of the problems we have is since the earth curves, the bottom of the radar beams are well above the ground when you get only 30 or 40 miles away from the radar.

Putting these small ‘cell phone tower’ radars, you can actually look in real detail at the various lowest levels. You can also make them look very quickly because of their adaptive scanning strategies. What we have here is a set of images over a ten minute period of a small tornadic storm that is only about fifty miles from the radar or less from the NEXRAD radar.

The signature associated with it in the NEXRAD is right here—it is actually an anti-cyclonic storm—it was rotating clockwise instead of counter clockwise like most tornadoes. In the bottom panels, we have four minute updates from the NEXRAD radars. The lowest scan is probably half a mile above the ground.

Here’s the tornadic signature and again over here, and it is the right direction this time. Down here a few minutes later—you’re looking well above and you don’t have much resolution—this is a small tornado and it doesn’t look like very much is happening.

With the CASA network, with the ability to update every minute, you see the signature occurring here, and it stays right with us another minute, and another minute, and so on, until it intensifies and becomes a bigger signature as we go along.

With this one minute update, this is the kind of thing it is likely that while this initial point right here might have gotten someone’s attention in the forecast office, it was still pretty small and they might not have been able to see very much of it. They might not have known things. It’s not clear that the signature here on this one would have attracted much attention at all.

With this very close to it radar, we are able to see things over a great period of time. That might be a benefit for helping near metropolitan areas where you might want more information in terms of improving the warning systems.

[Slide 22]

Beyond radar, one of the things we are looking at is how to improve forecasting events. One of them is to run an ensemble based on a model based on radar input, and run it for just a few hours. Most forecast models are run for days now—also, to generate probability of environmental conditions.

[Slide 23]

An image of that is a storm that is developing. We can take the radar information and put it into the model and we can make the estimate of in an hour from now, there will be a tornado that will be occurring, and this will be the probability that the tornado will be occurring, and this will be the most likely path.

If everything works out right, if the storm ends up right in the middle of where we said it would be, this is the kind of thing that is being developed. We’re hoping we can be working on this kind of technology that will be available seven to nine years from now.

It would be the kind of thing that rather than—currently, what we do is wait for radar to see that there is a rotation signature, or we wait from a report from a storm spotter. This will be the kind of thing that might be able to provide information that an hour or two from now it is very likely that this storm that is just beginning or hasn’t even begun yet will be producing a tornado in an area.

[Slide 24]

Essentially what you do is run several kinds of models, you add observations, that information comes back and it changes, you rerun the model again and get more observations, and you are eventually able to produce this probabilistic forecast system.

This is what is actually done for printing forecasts out for say, next week, when the current system is attempting to lots of information. In fact, at the European Center for Median Range Weather Forecasting, they actually run an ensemble that has 100 members in it that help provide really great information on what is going on.

We are hoping to try to do the same thing on the really short time scale within the decade—looking at maybe an hour or so ahead.

[Slide 25]

When we put the phased array radar into that kind of system, what we find is if we used the phased array radar as an input into it, the top two panels are the reflectivity and vertical vorticity, which is the measure or rotation associated with a tornadic thunderstorm.

You’ll see that there is lots of strong rotation (that’s the red) associated with the phased array radar data. If we put in NEXRAD data covering the same time period there are very few hints of very strong rotation occurring in that. By bringing in the data more often from the phased array radar, you are able to produce a much more realistic picture of the circulation.

At the time this is made, there is a strong tornado on the ground associated with this storm. That is a big deal in terms of improving our ability to make the forecast.

[Slide 26]

Another thing that happens is that we can also make very short term forecasts of the conditions within an environment. This is with the CASA network. These arrows are wind barbs. The brighter colors are reflectivity. This is from the May 10 storm last year.

We get incredible detail over the total of one hour long for this simulation. Every time it animates it is another five minutes. Actually the data is coming in a lot faster than that. It gives a lot of detailed information on how the winds are changing the environment and what is going on with the storm.

It should be helpful for warning purposes, but it is also helpful for a context we have never really considered. Even without the storms, things like the wind’s energy community are very interested in being able to predict on short time ranges the probability of major changes in the wind. This may provide a great deal of value for them.

[Slide 27]

There are some challenges involved in it, though. When you start updating data every minute or five minutes, you have to be able to figure out if the data are any good right away. Unfortunately one of the things we deal with in meteorology all the time is that really important extraordinary changes that occur frequently look like bad data because they are not what you expect to happen.

We have to learn a lot of things—errors in the models themselves, how sensitive they are to various kinds of errors, technical issues on how to bring data into the model, and how we can make effective use of ensembles on things that look at thunderstorms themselves.

Last week was an extraordinary event, to summarize. We have some changes that are ongoing over the next decade that will hopefully dramatically improve our ability to observe and forecast what is happening.

That is what I have, and I would be happy to take questions.

Amy Sebring: Thank you very much Harold. That was excellent. Now, to proceed to our Q&A. .

[Audience Questions & Answers]

Amy Sebring: What is the outlook in terms of the federal budget and the challenges that is going to present in implementing some of these strategies you have just described to us?

Harold Brooks: We have a new lab director who just arrived on April 24, so this happened right after he came in. We had our first meeting with him yesterday. We recognized that the budget doesn’t look very good for next year.

That forecasting technique is something that currently we have on the order of one and a half million to two million dollars a year into it, and it’s a much bigger project than that. That is a real danger—the budget issues will make this difficult to implement.

I work in the research part of NOAA. How things get implemented on the Weather Service side is also a challenge, and the Weather Service has historically paid for a lot of that research. Their budget doesn’t look very good for paying for that.

The phased array radar work is ongoing. One of the big issues is even if you build it, is there money to implement what is going on? The polarimetric upgrades to the radars got delayed because of budget issues several times. They should have been out in the field four years ago.

We continue to do the work we can. We’ll have to see what happens with the budget picture that goes with it.

Amy Sebring: Support of the emergency management community would not hurt.

Harold Brooks: That’s right. I didn’t mention one of the things we’re also working on—a lot of it is being done by graduate students—one of the big questions that occurs after a big event is, as best we can tell from the Weather Services side is there wasn’t a whole lot more they could do in terms of forecasting.

The forecasts were accurate. The warnings were out in plenty of time and there were still over 300 people killed. The question of—do we need to find better ways to communicate the information, or provide better ways for people to shelter? It is perhaps a big of a challenge as the meteorological side.

We have started partnering as much as we can with social scientists who can address some of those behavioral questions and some of the communication questions to improve the way we use the forecasts that are made.

Avagene Moore: We are almost halfway through May which is normally our most severe month for storms. Are there any long-range forecasts for the rest of the month to give us some idea what may be coming?

Harold Brooks: Unfortunately, we can’t go more than a few days or a week in advance. We’re dependent on what the long range pattern looks like. It is rare when we can see the things like last week, when it was a case of last week looked great all out—what we refer to as optically evident—evident from just looking at it.

Right now there is nothing that looks big next week that there is a lot of confidence in. It has been a quiet May so far. Today at one point, it looked like today might be a bigger day than it looks to be now. We really can’t say much beyond that now. We don’t have tools that allow us to do seasonal forecasting like our friends at the hurricane community do.

We know that historically having a busy or slow part at the beginning of the season tells us almost nothing about what the rest of the year is going to be like. So, you’re on your own.

Joe Sukaskas: With the relatively large number of tornadoes occurring, how effective were public notification and warning systems? Did public awareness get saturated by the probably large number of warnings?

Harold Brooks: This is one of the really interesting questions. In the background, there is a question that I really wish I knew the answer to, that is I don’t know what the public understands by the word "warning". The National Weather Service has a very technical meaning for the word "warning" and it is very precise.

I don’t know if there is any human being in the country that uses the word the way the Weather Service does. If you look at on average—and last week was obviously an extraordinary week—but if you look on average, the place in the country that historically has gotten the most tornado warnings is Harris County, Texas (Houston).

They are under tornado warnings on the average of six hours per year. The question of saturation is really hard for me, because I know that by the formal meeaning—if six hours is a lot where I live, in central Oklahoma (we average one hour per year)—if that saturates, it’s a huge problem.

I suspect that a lot of people have some sort of spatial scale, and they think that—here, we get the little map on our television screens if there is severe weather in the area—I wonder if people think that if the map shows up, they’ve been warned.

On the other side, almost after every event you’ll hear people say, "I didn’t get a warning". I think a lot of that has to do with the fact that people don’t use what we use in a technical phrase. They use it in a very different way.

If we want people to respond, we need to find out what they understand. Last week there were some issues with power outages that occurred in the morning. The systems had not gotten back up by the afternoon.

While I think this was not a very big problem, there was at least one couple in some of the newspaper stories the families said that they were unaware of the warning because the power had been out and they weren’t getting any information in at all.

We know that this wasn’t incredibly common, because there were a lot of other stories where people were clearly aware of what was going on and attempted to take shelter, or in one case, a young man decided he would rather take a nap than go with his girlfriend to the storm shelter. He slept his way through the tornado and was killed.

Ed McDonough: How valuable will these tools be in areas where the topography affects tornado activity, such as the Middle Atlantic area, where the rolling topography often leads to shorter duration tornados that come and go very quickly?

Harold Brooks: There are some advantages to some of the new radar technology in that we can see lower. Right now even if you’re in flat terrain, the lowest radar elevation tilts up over the ground. With the new technologies we can turn them to a lower angle. Also, the gap filling radars may be really helpful in that case.

If you’ve got a particular area of concern, you can presumably put one or two of those low powered radars in that area. You are only looking at 20-30 miles distance at the most. Those radars would be able to see those kinds of things and wouldn’t be as affected by the terrain. That’s a real promising area that can be used in, I think.

James A. Eberwine: With the tremendous advance in technology with advance warning time and the rapid news updates from the cable industry, why do you think so many people died? You answered part of my question, but were these communities hardest hit, there in 1974?

Harold Brooks: I think the big question is why there were so many fatalities. There are a couple of things. One is that these were violent tornadoes, as strong as tornados get, moving through populated areas. That is just at the beginning and that’s not necessarily a good combination.

A large fraction of the deaths occurred in mobile homes. What I’ve been working on the past two weeks and it seems like all the time, is to figure out where people were when they were killed. Right now we have gotten about 40% of the locations figured out. Two-thirds of the ones in homes were in mobile homes at this point.

There was one person I was reading about, that from the property record, the home was built in 1920 and was described as being in poor condition, and had a metal corrugated roof. My image of that was that it probably wasn’t a very sturdy structure, either.

One of the problems we have in the southeastern United States is the high mobile home fraction. Overall in the nation it is about seven percent of housing units, and in the southeast US, it is over fifteen percent. In parts of South Carolina, over twenty percent of the housing units in the state are mobile homes.

Not very strong tornadoes and mobile homes are bad, but a very strong tornado in there is really a big problem.

Isabel McCurdy: Harold, these images look pretty to me. Do the folk who live in tornado region get training to understand what these images mean?

Harold Brooks: I think so. We have a PhD student here at the University of Oklahoma and one of the things she is going to be looking at is people’s understanding of those kinds of images. I know that I hang with a skewed distribution of people, but everyone around here I talk to knows what a hook echo is. You get a good sense of what that means.

On our televisions stations, they show velocity images where people are developing a pretty good feel for that. I’m not sure how widespread that view is, especially as you move further away.

Last week was radar 101. They were signatures you get taught if you take a radar/meteorology class. They aren’t always that clean and easy to see. One of the big questions that has not been getting enough attention is what the public understanding of things like that is. Perhaps we should be showing that all the time.

People will see that raw radar image and they’ll know this is really bad. Perhaps to a large number of people that looks like a red blob and they don’t understand that the southwest corner of the blob that is actually important to them.

It may be confusing them. There may be people who think they are in much more danger than they are, and other people who think because it is not raining very hard where they are on the radar, they must be okay, when actually they are in the worst possible spot to be.

Amy Sebring: When I was working in emergency management locally, our local weather forecast office worked closely with at least the emergency management community in explaining those signatures and so forth.

Harold Brooks: In the forecast offices of the National Weather Services, and there are about 120 offices around the country—there is a person that that is their job, is to help public understanding. I think emergency managers get a lot of training in that.

There is probably another set of people who ought to get more training. If you were the manager of a large hospital or something where you have a lot of vulnerable people in your care, those kind of people would also be useful—if you had to move everybody out of some part of a hospital and that is going to take a long time, you may need to be more aware of what the weather is doing than the average citizen.

Amy Sebring: Plus you may need to gear up for injuries.

Harold Brooks: On the May 3, 1999 tornado, I had a lot of discussions with our public health people after it hit Oklahoma City, and they had a really good plan for how to deal with mass casualties, and that is what they had implemented in the Murrell bombing in 1995.

When the tornado occurred, one thing they hadn’t thought about is that the tornado is one mile wide in places and you can’t cross the path that night. The tornado is off-center of the metropolitan area, and there is only one major hospital to the south of it. Half of the victims ended up going to that hospital because they couldn’t go across the damage path. Every hospital was supposed to get 50 patients but the one hospital gets a few hundred.

Amy Sebring: We do have a link to your YouTube interview and I will refer folks to that for a little bit more. Could you mention the long term historical trends with la Niña, what you think about global warming and so forth?

Harold Brooks: There is some evidence that la Niña has an impact on tornado occurrence. The really big thing is the el Niño is unfavorable for tornadoes in most of the central United States. It doesn’t completely eliminate them. Maybe la Niña appears to be the kind of thing that gives you a few extra percent. Maybe you’re flipping a coin with 55% rather than 50%.

Although we have seen some of our biggest outbreaks in la Niña years. That is still an area of research. As far as climate change is concerned, we have a couple of problems in doing identification. The recording base has changed over the years to where you really can’t look historically without doing a lot of adjustments to the record and at that point, it is hard to pick up signals.

Our expectations as the planet has warmed over the last 40-50 years, and it will continue to warm over the next century—what will happen, we expect, is that one of the ingredients that is important for significant tornadoes and significant thunderstorms is going to become more favorable. Another primary ingredient is going to become less favorable.

The balance between those two is very hard to pick out. We really can’t see much in the historical record that says much has happened. That is really an unknown question. Doing some really simple simulations of data, you can show that even if there is a fairly large affect, a five to ten percent change, a big number physically, over the next century—the fact the some years are big and some are small anyway—it will take 50-60 years of data collection done consistently to actually see any change.

It will be a long time before we can tell if global warming has any impact on tornadoes.


Amy Sebring: Time to wrap for today. Thank you so much Harold. It was a wonderful presentation. We appreciate your taking the time to be with us and appreciate what you and your colleagues do.

The slide contains Harold’s contact information if you need it: [email protected]

Now for a few announcements:

EMForum.org now has a page on Facebook! Please stop by when you have a chance.

Our next program will take place Wednesday, May 25th. Our Topic will be the U.S. Geological Survey’s efforts to support the emergency management with its natural hazards mission area, and this will be an opportunity to provide feedback as they develop a strategy for the future. Please make plans to join us at that time.

Until then, thanks to everyone for participating today and for the great questions. Have a great afternoon. We are adjourned.