A Case Study in Weather Forecasting:

Thunderstorms



This case is told by Mr. Robert Hollenbach, a senior civilian Forecaster at the NASP METOC. He has been forecasting since 1977, including service on the USS Independence, the USS Wisconsin, and the USS America. He served as Forecaster at NAS Kingsville, Staff Forecaster at Pearl Harbor, and as an Instructor at the USN Aerography School. He has logged over 47,000 hours at forecasting or forecasting-related tasks.



In the case described here, there was a question of whether severe weather would occur at all, in a typical scenario in which one would not expect severe weather to arise. A single key datum triggered a line of hypothetical reasoning that led to successful prediction of severe weather that other weather forecasting stations missed--until it was too late. This case also illustrates the need for the forecaster to possess an thorough knowledge of the needs of his clients.



OSA = Observation / Situation Assessment D = Decision A = Action

 

Event

OSA,

D, A

TIME

This was my first day of midwatch.

I had watched the weather channel before coming in.

This was a standard scenario.

When I walked in the guy at the FDO station was just two years out of school.

He said, "Front is through, NW winds, clear all day."

He wasn't looking at anything else.

This is what most people do.

The front goes through then they do not look upstream, and assume that nothing will happen.

I want to limit the phone calls and do not like to have the pilots standing around.

I want them in and out.

You end up repeating the same answer to the same questions.

This gets frustrating after about three times of explaining the same thing.

And you get multiple calls from the same squadron.

that's frustrating.

Why don't they share the information?

Just yesterday, four guys from VT-10 called about the same SIGMET.

When weather is happening the weather itself creates work--warnings, TAF amendments, etc.

If you get ahead of the game you can avoid the pile-ups of pilots in here.

Better early than later.

Better for them to call when it is not yet happening.

Also, in the PM we get calls about stuff being planned the next day. They have to do their schedules before they go home. If weather is happening in the PM (which is when it commonly does) then there is a big crunch.

 

OSA

February 2000

Wednesday

5:00 AM

 

Radar showed mid-level moisture--some spots of green in SATRAD in the cirrus region.

A front had gone through.

Winds were out of the northwest, no clouds.
(I can't recall if the 500 millibar chart showed a weak thermal high behind the front.)

 

OSA

 

Nothing was said at the turnover.

The forecast was for NW wind all day long.

 

OSA

 

The satellite loop showed that the cirrus was still there.

The 500 millibar analysis showed an area of turning winds.

This was a key clue.

The area of cirrus shown on satellite was atypical cloud cover.

The cirrus continued behind the ridge.

Usually the negative vorticity wipes the cirrus out.

This cirrus made it all the way across to an area of negative vorticity (downward vertical motion).

 

OSA

5:30 AM

My goal was to try to figure out what would happen.

If it kept moving through nothing would happen, but if it didn't, you get stung."

I've had that happen over the years.

If you pulled up the wrong charts or charts without plots, you would overlook the turning and not ask yourself why it was there.

You'd not keep the satellite loop going.

The wind shift implied that something was happening.

A novice would have missed it.

 

 

 

If I had not seen the weather channel that AM I would not have been concerned with it.

I would not have looked at the winds and found the vorticity maximum.

I like to come in one step ahead of the game.

The key is interpretation.

Not so much the vorticity max as the turning of the winds.

The vorticity max was at the top of the ridge so it would not show up so much.

The turning of the winds was enough to me.

Any turning in any amount should be considered significant.

Then you look for support and weigh all the factors.

Try to justify the turning, do not try to not justify it.

In this case, the association of the turning with the vorticity max keyed it off.

A vorticity of 6 is not enough for most people to consider significant. (The vorticity scale goes from zero to 22 in 2-unit increments, and can be positive or negative.).

If the front had not stalled and there were no light and variable winds to hint of a sea breeze... light and variable winds support land heating and thereby contribute to the formation of a sea breeze.

The sea breeze linked up with the front's tail moving back up.

The sea breeze nullified the gradient associated with the ridge and allowed the front to move back up.

If the front had not stalled there would have been no moisture for the vorticity to interact with. the cirrus would have de-enhanced and lost the mid-level moisture.

It was dry at NASP.

The humidity was only 48% and was going down over the AM hours. When the winds changed, the relative humidity increased to 78%.

If I had not been looking, the wind shift would have been a cue since

the prognoses called for a gradient and a small craft warning. The wind shift implied that something was happening.

 

 

 





I put a statement in the Temporary Condition line in the Dash-1s that I was doing.

This was to let the pilots know the potential was there through if one were to skywatch here at the station, they would see clear skies.

 

D

5:30 AM

So I knew there was something to look at.

There was no time pressure at this point.

The stuff was still a few hours away.

 

D

 

Satellite loop showed no change in the cirrus enhancement, implying that the cirrus was not something to be concerned with.

Typically, as it cones down a ridge it should dive and dissipate due to the negative vorticity.

It was not enhancing, but it also was NOT going away.

It had turning and cirrus.

There must have been something balancing the positive vorticity (or divergence with the trough) and the negative vorticity as it was going down the ridge.

I could have done things differently.

I do not like doing other people's jobs, like giving brief explanations of synoptic situations and stuff like that.

We needed to amend the five-day forecast on the Home page, but with two forecasters working here we probably could have done that.

 

OSA

 

Normally, when I come in after two or three days off watch, I do not question the forecast at changeover.

If I see something and ask about it, they often haven't looked anyway!

 

D

 

I just decided to keep a look on this over the next few hours, and wait for the next model output charts at 9:00 AM.

I always glance at the PUP and LPATS as I walk in and out to go skywatch.

If you do not know what to look at you continue to not look.

Once you get burned and bust a forecast no one lets you forget it.

Even when you are right--forecasting bad weather, they complain at you for saying they can't fly.

 

D

 

I went back to the 00Z analysis and thought he saw some turning over southwest Missouri, but not a lot since it was at the top of the ridge axis.

There was moisture at 500 millibars.

The depression was 2.

The winds were faster ahead than behind, implying divergence.

Water vapor imagery showed no dry slot, implying the air was homogeneous and there was no instability.

I was seeing that there was potential here.

Divergence on top of a front will kick stuff off.

Something would happen.

The timing was the issue.

 

OSA

7:00 - 7:30 AM

No one else was going for anything.

There was nothing in the forecasts.

 

OSA

 

I decided to continue waiting and watching.

The error would be to ignore it until it was too late to let anyone know.

Thunderstorms would kick off.

Pilots would get stuck out there.

 

D

 

The new 500 millibar analysis arrived.

He saw rotation, enough to call it something--a disturbance.

The satellite enhancement showed that the cirrus was not changing--but it was also not dissipating--and was still moving to the south.

Being in this situation a couple of times makes you want to look at everything so you do not get stung.

You have to be prepared to explain why something did NOT happen in a standard scenario.

 

OSA

9:00 -9:15 AM

 

I decided there was something out there, and that it could kick off.

 

D

 

Across the loop, the cirrus stayed the same size and had no moisture associated with it, implying that it was not a thing to worry about.

It would be localized and not widespread.

 

D

 

I decided to look at the prognoses--the charts.

I wanted to look for any tendencies to lessen the disturbance or keep it intact, or deepen it.

 

D

 

None of the charts showed instability.

But the vorticity was there.

300 millibar chart showed a little increase in winds.

Some of the models tend to smooth out the little stuff, but the vorticity analysis does not do that.

 

OSA

9:45 AM

Vorticity in the same area as the increased winds implied divergence at 300 millibars, but not really a jet max.

It was not big enough to be called a jet max.

It is called a "minor short wave" since it is usually only at one level and is not as strong as a jet max.

If the vorticity maximum was in the same area as the winds, that would imply that the disturbance would not deepen.

A novice would not be looking.

They are trained in school but tend to not do what they're taught, and they get stung.

After that, you always ask "why."

 

D

 

He had to write a TAF in 30 minutes--at 10:00AM

 

A

 

He decided to keep watching the satellite imagery and saw no change.

It implied that there was no dissipation, but the disturbance might yet have fallen apart.

Still, I had to decide what to put in the TAFs.

A novice would not be focusing on the satellite loop.

 

D

 

I decided to put a thunderstorm warning in the TAF.

I did it to cover my rear.

It would have been an error to not out it out.

It is better to have them laugh at you because you were wrong (forecast rain, but no rain) than complain to you because you were wrong (no rain forecast but it rained).

A novice would have missed it altogether.

You are not ever supposed to have to put out a T-1 and a T-2 at the same time, since if you do that implies that you missed something and by the time you have to put it out it is already there.

 

D

 

I determined that it would reach the coast between 3:00 and 4:00 PM.

The decision was based on the movement I saw in a four-hour satellite loop.

On that, and on two runs of the analysis, I could do a simple calculation in my head of when the disturbance would get here--where it was four hours ago, where it is now, the approximate distance covered, then just divide.

The pilots get little training in weather.

Their mind set is due to the basic training they get in weather:

After a front goes through the weather is good and flying is good.

You have to convince them otherwise.

In this case I had to get out a T-1 to force them to ask and then get an explanation.

Pilots should get better weather training--not more training but more savvy training.

 

D

 

I included the 3:00 - 4:00 time in the TAF.

I called for thunderstorms beginning about 2:00 or 3:00 and out through sunset.

Usually if they do crop up in the PM they do not die down until after sunset.

Round-robin flights out of here going north on VR, by the time they land up there, eat lunch, and then come back down, that is just when stuff is popping up down here, the late afternoon.

That is the key time for them to get back in.

I wanted the pilots to be aware that if weather would happen it would be at about the time they'd be coming back in.

So, they would take a longer lunch and wait a while before leaving to come back here.

If a novice were at the FDO desk, they would never glance at the PUP.

SRF's get in a mind-set.

After they've done their fields forecasts every two hours and their dash-1s they just sit.

The NEXRAD should be seen from the FDO desk, not the SRF desk.

 

A

 

I decided to wait for the rest of the model run to come in.

I knew to keep an eye on the surface winds on station.

(see entry for 10:30 - 12:30 PM, below)

The error here would be to miss stuff.

It was always a possibility that the later model runs would show a decrease in intensity implying that the deepening had been counteracted by the ridge.

Or, the models could show an increase, implying a change from isolated thunderstorms to severe thunderstorms.

Also, the models runs would include a revised lifting index, and if the lifted Index increased, that would imply severe weather ant not isolated storms.

Severe weather has implications for the Charlie areas--I have to put out special warnings.

So, missing this would be a big error.

 

D

 

Saw some pink in the radar.

It was at the edge of the Mobile radar and showed pink,

 

OSA

10:30

This implied that there was moisture at the high level, which fit with the cirrus to the east.

This all was fitting the standard scenario.

 

D

 

The cirrus continued to move south and he noticed on GOES that it was enhancing.

There seemed to be some low and mid-level stuff building up.

This implied that storms were going to happen.

The disturbance was starting to pick up low-level moisture.

The pink in the radar was not the precipitation (yet)--it was showing the moisture being picked up.

 

OSA

 

New prognoses, NGM outputs, etc. arrived. They still showed nothing.

The runs included no increase in the Lifted Index.

No change in anything.

So, the biggest problem again was that the models tend to smooth things out.

In this case, the GOES data was critical, and radar too.

Nothing else was helpful.

There are not a lot of stations out in Mississippi and rural Alabama, so there is a paucity of ground observations.

Sparse data.

I've lived (and died) by satellite data more times than not.

 

OSA

10:45 -

11:00 AM

Winds at NASP shifted from the northwest at 12 knots to the east-northeast at 3 knots.

This suggested that the front was passing to the south now.

The front had stalled out and gotten closer to the station.

A novice may not have noticed it when the popcorn cumulus began showing up in the GOES imagery, at about 1:15 PM.

 

OSA

10:15 - 12:30 PM

I kept an eye on the local surface winds--watching the front.

OSA

10:15-12:30

PM

I hypothesized that either the front had stalled and backed up or the land heating has counteracted any gradient that was out there.

f the front moved down, divergence would hit it over the water and nothing would happen here at NASP.

If the front stalled, divergence would hit it over land and pull it back up.

Land heating is rare for that time of year--not enough heating to cause a sea breeze.

And besides, you still need the light and variables winds, even in the winter scenario.

So I figured that the front had just stalled.

With stationary fronts, as the upper-level system hits the front, there is a bit of turning on part of the front. Part of the front turns back up north and part turns down southward.

In this case, the southward-turning part of the front was down in Mississippi--they had the north-west winds there and to Mobile.

East of Mobile to here we had East-northeast winds.

So the center of the turning was between Mobile and New Orleans, approximately.

 

D

 

I knew they would either have northwest winds (steady gradient) or light and variable winds out of the east.

 

D

 

Light and variable winds out of the east would be atypical because they would imply that the front had stalled and was being pulled back up to the north on top of this region.

 

D

 

Winds were light and variable out of the east.

There were still no clouds associated with the tail of the front.

But there was the possibility that clouds could increase in the low level along the front.

If the winds had stayed north-west I would have amended the TAF--there would be no way we would get low-level moisture here.

 

OSA

About 12:30

PM

 

Winds were southeast at 10 knots.

This implied that the front was right on top of us or just to the north and NASP was back on the warm side of the front in the low-level moisture.

 

OSA

1:00 - 1:15 PM

 

I went outside and skywatched.

I expected to see low-level clouds.

If they were not there, I'd have known that I'd blown the forecast.

I also looked for enhancement--clouds going up and not just stratifying.

Stratification would mean capping and subsidence can cap the development on the cold side of the front.

But on the warm side, the inversion can be broken and the air goes unstable.

 

A,

OSA

 

 

I could see low clouds scattered and flowing up to the north

 

OSA

 

Over this next hour there was rapid development.

Satellite showed clouds down in the Alabama-Mississippi border region into southeast Mississippi.

There was a line of popcorn cumulus at the I-10 corridor.

Showers began to show in radar.

I knew I'd nailed the forecast.

What was happening was what I had told them was going to happen.

 

OSA

1:15 - 2:15 PM





Thunderstorms and the vorticity max expanded along the cumulus line.

Everyone else had to change their forecasts.

Norfolk did a SIGMET but after stuff started to develop.

I had not talked to any of the other stations because I was busy with pilots.

Everyone thought it would be a good flying day--lots of people were flying.

the National Weather Service waited until stuff popped then they issued numerous amendments.

 

OSA

2:30 PM





Stuff popped up.

 

OSA

3:45 PM

The front moved through NASP and things cleared out.

Winds shifted to the northwest.

None of my guys got stuck.

They left for their return legs later than they would have and arrived at NASP later in the PM.

So no one had to divert, no one out of about 14 flights.

 

OSA

4:00 - 4:30

PM

 


 

Decision Requirements

 

Cues and

Variables

 

 

 

 

         Although satellite imagery is often said to be the prime data source for getting the bit picture, the satellite data can be critical in supporting on-going situation assessment through long-term monitoring of the loops. Using a loop one can readily calculate the direction and rate of motion of developing systems and thereby forecast the onset time of severe weather. this can be the only way to determine onsets for smaller-scale weather events that are sometimes glossed over by the computer models.

"The runs included no increase in the Lifted Index. No change in anything. So, the biggest problem again was that the models tend to smooth things out. In this case, the GOES data was critical, and radar too. Nothing else was helpful. There are not a lot of stations out in Mississippi and rural Alabama, so there is a paucity of ground observations. Sparse data. I've lived (and died) by satellite data more times than not."

 

Needed

information

 

 

 

 

         Upper-air data on troughs and vorticity can be critical in forecasting rapidly-developing severe weather.

         Awareness of overall weather situation at the beginning of a watch is critical, but as pass-down very little is typically said.

"Nothing was said at the turnover. The forecast was for NW wind all day long."

"Normally, when I come in after two or three days off watch, I do not question the forecast at changeover. If I see something and ask about it, they often haven't looked anyway!"

         The layout of the operations floor at the METOC clearly calls for changes, including making the NEXRAD PUP directly viewable from the FDO desk rather than the SRF desk.

         The key data in severe outbreaks of this sort seem to be moisture at the low levels, the amount of lifting, convergence-divergence across levels, and an energy source such as an upper-level trough running over a front.

         Computer models sometimes gloss over smaller-scale events, with the exception of the vorticity charts. the forecaster needs to be alert to weather events that the models can miss, and then look to data types that can be informative.

         Sometimes the decision to explore a possibility hinges on seeing a single key clue:

"The satellite loop showed that the cirrus was still there. The 500 millibar analysis showed an area of turning winds. This was a key clue. My goal was to try to figure out what would happen. If it kept moving through nothing would happen, but if it didn't, you get caught stung. I've had that happen over the years."

         The experienced forecaster never relies exclusively on the data provided through mediated systems. They always directly observe the atmosphere and clouds at the field. This dovetails on the notion that the METOC facility and layout should be re-designed. Skywatching the region south of the field is possible through easy exit from the operations floor, but skywatching toward the field and north is possible only by a considerable traverse of the building. Ideally, the operations floor should look out onto the field through a panoramic window.

 

Hypotheticals

 

 

 

         Forecasting involves not just determining what will happen and when, but sometimes involves determining why something that is expected in a given typical scenario is NOT happening.

"Typically, as it cones down a ridge it should dive and dissipate due to the negative vorticity. It was not enhancing, but it also was NOT going away. It had turning and cirrus."

         There must have been something balancing the positive vorticity (or divergence with the trough) and the negative vorticity as it was going down the ridge.

         Hypothetical reasoning about severe weather outbreaks involves a number of possible alternative scenarios. the highly experienced forecaster engages in a great deal of hypothetical reasoning, sometimes in spite of the fact that incoming data cut against a favored hypothesis or make one or another scenario less likely.

"But there was the possibility that clouds could increase in the low level along the front. If the winds had stayed north-west I would have amended the TAF--there would be no way we would get low-level moisture here."

 

Options

 

 

 

 

         It is important for the forecaster to know the usual activities and routines of the clients and not just the bare flight plan (e.g., time of departure to determine such things as ceiling and visibility at the airfield). For example, pilots doing round-robin flights will often eat lunch at their way-point then depart for the return to NASP, bringing them into NASP in the PM hours just when summer regime storms are developing. If earlier in the day the forecaster anticipates the possibility for afternoon severe weather, alerting the pilots to this before they depart NASP will let them know to delay their departure for the return leg.

 

Goals

 

 

 

       Issuing a thunderstorm warning sometimes constitutes justified over-forecasting. If severe weather does occur and a warning had not been put out, that would be a worse error than putting out a warning and then the bad weather does not happen. " It is better to have them laugh at you because you were wrong (forecast rain, but no rain) than complain to you because you were wrong (no rain forecast but it rained).)

 

Situation

Assessment

 

         Training in the Schoolhouse on the standard scenarios should include gaming in which the students are set-up to "get burned," just as was possible in this case. The learning in the Schoolhouse is said to transfer only after lived experiences in which errors are made, teaching the student to look for things to explain why expected events do NOT occur and why the unexpected CAN occur. The information you need depends on your knowing where to look.

         Novices are said to be weak in terms of picking up subtle clues for small-scale events that the computer models gloss over.

"If you pulled up the wrong charts or charts without plots, you would overlook the turning and not ask yourself why it was there. You'd not keep the satellite loop going."

         Novices are said to fall prey to mindsets, such as the mindset that says that after frontal passage the weather will be clear. Hence, they expect nothing to happen and do not look to see if anything is happening. Schoolhouse exercises should help them break through such mind-sets. "They are trained in school but tend to not do what they're taught, and they get stung."

 


 

Time/effort

 

 

         Pilots do not appreciate the emotional impact of their casual behavior, and through their own lack of understanding of the forecaster's activities, they cause an unnecessary increase in forecaster workload. the emotional impact of the client's behaviors on the forecaster becomes salient:

"Even when you are right--forecasting bad weather, they complain (euphemism) at you for saying they can't fly."

"It is better to have them laugh at you because you were wrong (forecast rain, but no rain) than complain (euphemism) at you because you were wrong (no rain forecast but it rained)."

It seems ironic that the forecasters want their clients (the pilots) to leave them alone. But there is a reason:

"I want to limit the phone calls and do not like to have the pilots standing around. I want them in and out. You end up repeating the same answer to the same questions. This gets frustrating after about three times of explaining the same thing. And you get multiple calls from the same squadron.

that's frustrating. Why don't they share the information? Just yesterday, four guys from VT-10 called about the same SIGMET. When weather is happening the weather itself creates work--warnings, TAF amendments, etc."

         Pilots and their wings should do a better job of sharing weather explanations. Weather information presented to pilots should be explanatory and not merely descriptive.

         The training pilots receive in weather needs to be better, and not just more, but more savvy:

"The pilots get little training in weather. Their mind set is due to the basic training they get in weather: After a front goes through the weather is good and flying is good.

You have to convince them otherwise. In this case I had to get out a T-1 to force them to ask and then get an explanation."

         It is important to possess the willingness and fortitude to inspect incoming data stream frame by frame for long periods of time.

 

Concept Mapping Toolkit
Insitute for Human and Machine Cognition
The University of West Florida