8/19/09 DISC: IL/MN/OK/KS

[CZandbergen]

In the end, I was too concerned with storm potential to sit around and be completely sure that storms would form before issuing a watch for OK/KS. It's fun to speculate about what went wrong. However, until I see something I can use that will give me supreme confidence to lay off that environment, or I get a call from somebody with a better forecast *and* an explanation in real time, I'll continue to have to issue the bust watch you saw yesterday.

Rich,

Thanks for jumping in here and giving your point of view. It is really neat to have a forum where we can hear from the experts that have to make those calls. I started this thread to see what I and or we could learn from yesterday. If someone had the magic bullet for forecasting this stuff there wouldn't be as much research left to do in this area. I don't chase watch boxes but I will admit it was nice to hear the watch being issued as I was crossing I35 headed west. Just lets me know that the experts came to a similar conclusion as me.

I have learned quite a bit just reading the discussion thus far and appreciate everyones comments. One thing I will throw out there is that while I was driving back to Tulsa last night, I was able to watch the debris from the Enid cell ride the upper air currents due east. What was interesting to me is that it looked pretty much like what was happening as the cell began to die and is depicted in the picture in this thread. I keep coming back to the lack of directional shear above 2km. Everything from 2km up was pretty much due east. From the soundings and from what we saw yesterday it didn't look (to me at least) like we had a cap issue. All that said, I also think the discussion about the interaction with the boundary is interesting as well.

Good stuff, if anyone else has any ideas I would like to hear them.

 

[Jeff Snyder]

For the most part, it certainly looks to me like northwestern Oklahoma was more in the left entrance (i.e. subsident) region of the jetstreak on the eastern side of the trough than anything else. However, it's pretty much in the trough axis, so it's tough to say.

I know there's the conceptual models for tornadic supercells events (that is, primarily, downstream of a trough aloft, perhaps in the diffluent and divergent left-exit region of an embedded jet streak, etc). However, it just seems odd the some synoptic-scale subsidence (with magnitudes often the order of centimeters per second) could inhibit deep, moist convective updrafts with magnitudes of 20-50+ meters per second (~3 orders of magnitude larger!). There appears to have been decent surface convergence along the OFB yesterday, which, with little to no CINH, makes me wonder HOW the relatively weak synoptic-scale downward motion can "overwhelm" the very intense storm-scale convective upward motion associated with such extreme buoyancy. I know that we see this time and again, though.

FWIW, often the negative aspect of subsidence is warming aloft and the formation of subsidence inversion that can act as a cap. However, as seen in the above LMN sounding, this was not the case. There was no cap (or very weak capping if you consider a 100mb ML "parcel"), so the thermodynamic manifestation of subsidence (e.g. often midlevel warming) doesn't appear to have been a factor.

As Rich noted, many indices must be interpreted in light of conditional convective presence. Heck, we know that we can have 5000+ j/kg MLCAPE without any supercells, just as we can have 500 m2/s2 0-1km SRH, 60 kt 0-6km shear magnitude, etc., all without any tornadic supercells occurring. For the most part, none of these tell us anything about convective initiation, and they are typically only useful as convective threat predictors when the condition of initiation is met. As we've seen, as well, perhaps the "presence of discrete convection" condition is not exactly binary; perhaps the utility of such parameters is more complex than "yes" or "no" in terms of a convective updraft. Like the 6/7/07 central Oklahoma bust (which featured very favorable parameters for tornadic supercells AND saw a couple of short-lived thunderstorms, similar to yesterday in northern Oklahoma), I imagine that the residence time of parcels within the zone of maximum convergence was not enough to establish a healthy updraft before the particular storm moved away from the surface convergence. In other words, the storms needed a little longer time in the zone of strong low-level convergence in order to better establish themselves before moving off the OFB (or dryline as on 6/7/07). There are no parameters that I know of that can help much in this regard, unfortunately. Moisture convergence helps, but we're talking more about Lagrangian motions (i.e. trajectories)...

 

[Rich Thompson]

For the most part, it certainly looks to me like northwestern Oklahoma was more in the left entrance (i.e. subsident) region of the jetstreak on the eastern side of the trough than anything else. However, it's pretty much in the trough axis, so it's tough to say.

I know there's the conceptual models for tornadic supercells events (that is, primarily, downstream of a trough aloft, perhaps in the diffluent and divergent left-exit region of an embedded jet streak, etc). However, it just seems odd the some synoptic-scale subsidence (with magnitudes often the order of centimeters per second) could inhibit deep, moist convective updrafts with magnitudes of 20-50+ meters per second (~3 orders of magnitude larger!). There appears to have been decent surface convergence along the OFB yesterday, which, with little to no CINH, makes me wonder HOW the relatively weak synoptic-scale downward motion can "overwhelm" the very intense storm-scale convective upward motion associated with such extreme buoyancy. I know that we see this time and again, though.

FWIW, often the negative aspect of subsidence is warming aloft and the formation of subsidence inversion that can act as a cap. However, as seen in the above LMN sounding, this was not the case. There was no cap (or very weak capping if you consider a 100mb ML "parcel"), so the thermodynamic manifestation of subsidence (e.g. often midlevel warming) doesn't appear to have been a factor.

As Rich noted, many indices must be interpreted in light of conditional convective presence. Heck, we know that we can have 5000+ j/kg MLCAPE without any supercells, just as we can have 500 m2/s2 0-1km SRH, 60 kt 0-6km shear magnitude, etc., all without any tornadic supercells occurring. For the most part, none of these tell us anything about convective initiation, and they are typically only useful as convective threat predictors when the condition of initiation is met. As we've seen, as well, perhaps the "presence of discrete convection" condition is not exactly binary; perhaps the utility of such parameters is more complex than "yes" or "no" in terms of a convective updraft. Like the 6/7/07 central Oklahoma bust (which featured very favorable parameters for tornadic supercells AND saw a couple of short-lived thunderstorms, similar to yesterday in northern Oklahoma), I imagine that the residence time of parcels within the zone of maximum convergence was not enough to establish a healthy updraft before the particular storm moved away from the surface convergence. In other words, the storms needed a little longer time in the zone of strong low-level convergence in order to better establish themselves before moving off the OFB (or dryline as on 6/7/07). There are no parameters that I know of that can help much in this regard, unfortunately. Moisture convergence helps, but we're talking more about Lagrangian motions (i.e. trajectories)...

Nice post, Jeff. The synoptic-scale subsidence, if it was present, was not clearly manifest in the observed data (other than a lack of clouds on the cool side of the OFB, noted by Jon D.). The jet quadrant arguments change when you have curved flow, such that the RF quadrant near the jet core is not necessarily subsident. I do, however, think that a lack of larger-scale ascent can influence the mesoscale vertical motion profiles along the OFB. It really does come down to parcel residence time in the zone of ascent, and the specific details regarding boundary depth, trajectories, etc. This won't be an easy problem to solve anytime soon.

 

[Rich Thompson]

But that is the point. There were supercells in Kansas. Yet, no tornadoes with a sky-high index value. I can't remember a day without tornadoes when things looked so good and supercells formed.

None of this should be taken personally or as a criticism of SPC.

My point with the previous statement is that a combination of 4-5 ingredients (STP) can't possibly explain all of the variability observed in the atmosphere. Sure, I'd call yesterday a failure of the index, and every other index/parameter out there. However, the storm E of ICT formed late and on the cool side of the boundary, which suggests low-level warm advection as a culprit. That isn't the "standard" way for a tornadic storm to form, and I wonder if that cluster was every fully surface-based.

More importantly, the presence of high STP values gives some indication of the probability of a sigtor given a supercell. From a sample of 835 supercell proximity soundings, STP > ~6 are associated with sigtors about 75% of the time. That still leaves 25% that do not produce sigtors (or any tors) for various reasons, many of which are unknown. I understand that you're searching for answers in this case, and that the parameters/indices are far from perfect. Still, what would you propose using that's better? It's still an inexact science, as suggested by the presence of "indices" in the first place.

 

[Chad Cowan]

No offense intended, Chad, but the 300 mb height chart was rather unexemplary to show your case. Looking at it closely, I agree with Mike: with the exception of far eastern/southeastern OK, the entire state was under BOTH the left entrance and right exit regions of the jet streaks around, thus implying quite a bit of downward motion under the kissing jets there.

No need for the disclaimer; I'm still learning and like to know when I'm wrong. I have a couple questions though. First of all, the jet streak quadrant map that I posted is assuming zonal flow with no curvature in the streak. I thought that in order to map the entrance quadrants with a curved jet streak the line should start at the apex of the curve and bisect the highest winds to the tip of the streak, like this: http://chasethestorms.smugmug.com/gallery/7046675_dqQQV#625707245_homXM What is the right way to do this?

Secondly, you said that being under both the left entrance and right exit regions implies quite a bit of downward motion under the kissing jets. I thought that both of those quadrants suggested vertical motion and the other two quadrants suggest downward motion. Can someone please un-confuse me? (EDIT: I got the entrance and exit terms mixed up here, disregard my noob-ishness)

What I've noticed from studying the 300mb chart more is that, regardless of the quadrant placements, there was a LOT of upper-level convergence over Oklahoma where the the second jet streak was plowing into the first.

 

[Mike Hollingshead]

You've got your quadrants backwards, BTW Mike, northwest OK yesterday was in the right-entrance region of the upper jet streak.

I was saying left entrance talking about the lead jet streak, then mentioned the other jet diving down the back of the trough. That should be right. And you may have meant right EXIT?

 

[Chad Cowan]

I was saying left entrance talking about the lead jet streak, then mentioned the other jet diving down the back of the trough. That should be right. And you may have meant right EXIT?

Yeah, you're right. This is why I prefer using the left-front, right-rear terms. The whole entrance in the exit and exit in the entrance stuff gets me turned around.

 

[Mike Hollingshead]

It's interesting to me that the right entrance, left exit(both obviously) and even right exit can work, but man it's hard to come up with anything nice in that left entrance area. One of those things you see and think right away, ok nothing else matters as that just never works out. Surely it has though, but I can't come up with any examples(from my limited "database" of looking at things on chase days for a few years).

 

[Jeff Duda]

First of all, the jet streak quadrant map that I posted is assuming zonal flow with no curvature in the streak. I thought that in order to map the entrance quadrants with a curved jet streak the line should start at the apex of the curve and bisect the highest winds to the tip of the streak, like this: http://chasethestorms.smugmug.com/gallery/7046675_dqQQV#625707245_homXM What is the right way to do this?

Yes, you marked that correctly! on a side note, although I think Jeff Haby's site (www.theweatherprediciton.com) is a very good site for beginners in meteorology, I don't really think much of it for anything above intermediate level meteorology. I've taken his 500 question quiz and run across at least 3 or 4 questions which he marked the answer as being totally wrong. Regarding his page about jet streak theory, I don't agree with his explanation at all. Case in point:

(2) Parcel enters region of higher wind speed. This increases the Pressure gradient force at the same time the Coriolis has not been changed much.

An air parcel is accelerated by the pressure gradient force, which must be there first. He has the implication backward. Plus, once the air accelerates, the Coriolis force on it will increase because the Coriolis force equation is COR = f*v, where f is the Coriolis parameter 2*(omega)*sin(phi), where omega is the rotation constant = 7.27 x 10^-5 / s and phi is the latitude in degrees, and v is the magnitude of the wind speed. Thus, whenever air parcels accelerate (i.e., wind speed increases) the Coriolis force will increase on it. This is not what Jeff's site says.

The easiest explanation for me for vertical motion in a jet streak is to use the PVA/NVA argument. Since you have a jet streak, or wind maximum, then there should be a tightening of the height gradient at the streak (and a relaxing of it before and after the streak). Thus, by writing out the resulting pattern of heights, you should see cyclonic curvature on the poleward side of the streak and anticyclonic curvature on the equatorward side. Thus, you would have a vort max on the poleward side of the middle of the streak, and a vort min on the equatorward side. Thus you would have PVA in the right entrance and left exit regions -> upward motion by QG theory, and NVA in the other two quadrants -> downward motion by QG theory. Granted, this explanation only works when the heights are as I assumed, but it serves as an easy to understand model for me. When jet streaks are not linear, throw most of that theory out the window, I've learned. There is a COMET module about this, but it is a toughie (level 3). It is at this link.

Secondly, you said that being under both the left entrance and right exit regions implies quite a bit of downward motion under the kissing jets. I thought that both of those quadrants suggested vertical motion and the other two quadrants suggest downward motion.

As you said it, vertical motion can be in two directions, up and down. I think you may have just typed that wrong.

What I've noticed from studying the 300mb chart more is that, regardless of the quadrant placements, there was a LOT of upper-level convergence over Oklahoma where the the second jet streak was plowing into the first.

In my synoptic class I remember being taught that you can't exactly eyeball convergence/divergence in the upper levels. The big point there was that confluence does not necessarily imply convergence, and diffluence doesn't necessarily imply divergence. Typically when you have diffluence, the height gradient will relax and you'll get slowing of the winds such that, speedwise, convergence would occur, and when you have confluence, the height gradient tightens, thus giving speed divergence. You have to add the magnitudes of those two terms to determine if there is truly any divergence or convergence. Moreover, in a perfect world, divergence of the geostrophic winds is always zero. At 300 mb, you'd expect to find nothing but geostrophic winds. However, that is not the case, as you get ageostrophic flow (supergeostrophic and subgeostrophic) when the flow is curved, such as around a trough or ridge axis (as is in this case). Therefore, you can actually get nonzero divergence at this level. But still, you'd have to actually calculate it to be sure.

 

[Jeff Snyder]

(2) Parcel enters region of higher wind speed. This increases the Pressure gradient force at the same time the Coriolis has not been changed much.

An air parcel is accelerated by the pressure gradient force, which must be there first. He has the implication backward.

Jeff - I think Haby was looking at it in a Lagrangian perspective. In other words, as an air parcel enters a jet streak, it encounters an increasing pressure gradient (i.e. increasing PGF). Yes, the increased pressure (or height) gradient is the driver behind the jet streak in the first place, but, from the perspective of an air parcel embedded in the flow approaching a jet streak, it does encounter greater PGF with time (assuming flow through the jet streak is greater than the downstream movement of the jet streak). There is a time lag in the response of the coriolis force, and this force imbalance (and associated ageostrophic flow) is at the root of jet streak dynamics and the transverse circulation. This is getting a bit off-topic, and I'm relatively sure you're aware of this, but I figured it was worth noting anyway.

 

[Matthew Piechota]

the tornado that went from williamsville to beason IL (which is the tornado we saw) was rated an EF-3 with max wind of 140mph.

It was on the ground for 24.5 miles from 3:18 to 4:02

 

[Jeff Duda]

Ah okay, the way you said it makes more sense. Thanks for clarifying.

 

[John Farley]

Damage survey results

So NWS offices have confirmed five tornadoes each in Illinois, Iowa, and Minnesota, and one each in Wisconsin and Indiana,

for a total of 17. I am sure additional tornadoes will be confirmed by the Lincoln, IL office as investigations continue,

and possibly by other offices.


Chicago NWS Office:

Two tornadoes in Kane County - 1 EF1 with a path length of 1 mile and another EF0 with a path length of 3/4 mile.

Details - http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=lot&storyid=30615&source=0

One tornado in Chesterton, IN, EF2 with path length of 4 miles.

Details - http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=lot&storyid=30613&source=0


Lincoln NWS office:

One EF1 tornado which began in Greene County (NWS St. Louis CWA) but after a mile or so crossed into Scott County and

eventually Morgan County. Path length: 18 miles.

One EF2 tornado in Morgan and Sangamon Counties, path length 16 miles.

One EF3 tornado in Sangamon and Logan Counties, path length 24.5 miles.

Additional possible tornadoes still under investigation.

Details - http://www.crh.noaa.gov/ilx/?n=19aug09


Quad Cities NWS office:

One EF1 tornado in Buchanan County, IA, path length 2 miles.

One EF0 tornado in Jones County, IA, path length 1/10 mile.

Details - http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=dvn&storyid=30601&source=2


Minneapolis NWS office:

One EF0 tornado in the city of Minneapolis, path length 4.5 miles.

One EF1 tornado in Cottage Grove, MN, path length 1200 yards.

One EF0 tornado near Hudson, WI, path length 600 yards.

One EF0 tornado in North Branch, MN, path length 1 mile.

Other tornadoes touched down briefly in Brown and Blue Earth Counties, MN, but stayed in open country and caused no damage.

Details - http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=mpx&storyid=30471&source=0


La Crosse, WI NWS Office:

One EF0 tornado near Fredricksburg, IA, path length 1/2 mile.

One EF0 tornado near West Union, IA, path length 1.7 miles.

One EF0 tornado near Calmar, IA, path length 4 miles.

Details: http://www.crh.noaa.gov/news/display_cmsstory.php?wfo=arx&storyid=30595&source=0

 

[L Kimbrel]

Great thread, guys. I've read it twice thru just to figure out what was said, and it'll take a couple more read-throughs to understand it. A mini-meteorology course all in itself!

 

[Chad Cowan]

I have posted the HRRR (hi-res rapid refresh) model's forecast precip maps on my site here: http://chasethestorms.smugmug.com/gallery/7046675_dqQQV#628449808_RPYqH

The forecast time is 0z for all 11 images, and the model performed quite well. I hadn't used the HRRR much, but I'm definitely going to pay more attention to its output when I'm in the field in the future.