8/4/2004 FCST: Upper Midwest

[Robert Dewey]

I'm trying to figure out exactly which features are suspected of over-amplification due to convective feedback. The low level wind fields?

What can happen is that when the forecast models induce convection (such as today, over IA), it can contaminate the rest of the downstream grids. This usually results in abnormally high QPF, strong wind fields, intense 500mb vorticity centers (compact), and compact SFC lows. The models are suffering from all of these symptoms, but I am not so sure that they arn't correct in doing so... So to over-amplify would mean the flow/500mb vorticity/and depth of the hghts...

 

[Robert Dewey]

It wraps up some VERY strong winds in the low-levels, and the GFS does the same but not with that intensity. SPC should not upgrade to MDT given the unlikeliness of this event.

- Rob

Yep, its unlikely, but still possible... The 00Z models will make the ultimate decision...

 

[Amos Magliocco]

This is interesting because we know that clusters of storms can produce convectively induced vorticity maxima. Now I'm less certain about this, but I think I recall an instance a few years ago of either ETA or RUC actually anticipating that. I remember it struck me as bizarre because of how unusual the entire phenomena is so I wasn't exepecting numerical modeling to see it.

I guess tomorrow is another case of why it makes sense to ditch the models on Day 1 and dig through real data before making serious chase decisions.

 

[Robert Dewey]

This is interesting because we know that clusters of storms can produce convectively induced vorticity maxima. Now I'm less certain about this, but I think I recall an instance a few years ago of either ETA or RUC actually anticipating that. I remember it struck me as bizarre because of how unusual the entire phenomena is so I wasn't exepecting numerical modeling to see it.

I guess tomorrow is another case of why it makes sense to ditch the models on Day 1 and dig through real data before making serious chase decisions.

One thing to do is watch the RUC forecasts... Right now, the 18Z RUC 12HR forecast for tomorrow is pretty much identical to the 12Z ETA 24HR forecast - Except the low is a *bit* further north at 850mb, on the RUC (the RUC is also a little more widespread with the +40KNT contour at 850mb as well). The more I watch the evolution of things, the less I see convective feedback. The 12Z GEM/12Z UKMET both show the same solution as the USA models, which makes me *almost* positive that its not convective feedback since all of these models use different convective schemes.

Anway, check out this plot from UWISC, this is supposedly SFC winds - Check out the 50knts SFC winds in MI, just north of the heavy precip core! Their model usualy WAY over-predicts SFC wind speeds, but I just thought this was interesting to show...

 

[Glen Romine]

One thing to do is watch the RUC forecasts... Right now, the 18Z RUC 12HR forecast for tomorrow is pretty much identical to the 12Z ETA 24HR forecast

A 12 hr forecast from the 18Z RUC would be valid 06Z, whereas the 12Z ETA 24hr hour forecast would be valid at 12Z - so this is really compairing apples with oranges.

Convective feedback is funny thing - generally only a problem when the explicit convective scheme fails and the much coarser convective paramterization scheme kicks in - generating convective overturning on much too large of a spatial scale - which can cause significant problems for the model solution usage in those areas impacted. However, upscale growth of convection should not be automatically attributed to convective feedback - the models are run at relatively high resolution these days and generally are capable of resolving mesoscale processes. Unfortunately, the operational ETA only has one shot at getting the mesoscale evolution right - so when you see these types of features you should automatically think this is one possible evolution subject to verification - but to literally say things will happen as shown, down to the county level, will generally leave you grossly disappointed. A good check is to look at the ensemble forecasts to see if they show a consistent signal, or if there is a large range of variability.

Glen

 

[rdale]

18Z Eta is more to the north - helicities of 400+ get up to a Grand Rapids to Flint line by 18Z with CAPE up to 1000. Best instability still over IN/OH but less directional shear.

Interesting over Iowa is that it whips up BOATLOADS of QPF overnight (7" of rain SW of DSM through 6Z tonight when there's not a drop falling now) and takes that max through Milwaukee (4" by 18Z) then towards Flint by 06Z tomorrow night. Another max develops near Toledo by 00Z tomorrow evening and extends up to Buffalo.

Local AFD's aren't saying squat other than "SVR STORMS POSSIBLE" at the most.

- Rob

 

[Robert Dewey]

A 12 hr forecast from the 18Z RUC would be valid 06Z, whereas the 12Z ETA 24hr hour forecast would be valid at 12Z - so this is really compairing apples with oranges.

Opps ! What I meant meant was the 12Z ETA 18HR forecast (that truly is what I meant )...

Anyway, I am still trying to understand convective feedback and model physics/etc. I thought that if the model falsely initiates convection (on either scale), that it would contaminate the downstream grids in several ways -
1) Thunderstorm outflow/clouds would cause temps to cool
2) Model soundings for the particular region would show almost complete saturation (since the model is "raining)
3) Not sure, but wouldn't it also cause intense VVs?

All of which would be false, since there is no convection occuring (am I making sense? LOL)

As far as ensembles (mainly 09Z SREF), shows pretty much similar solution to the 12Z models...

 

[rdale]

Checking the WRF & 10km TAQ both have the same line of thinking with monstrous helicities over southern third of MI and extreme winds & CAPE in IN/OH.

- Rob

 

[Robert Dewey]

Checking the WRF & 10km TAQ both have the same line of thinking with monstrous helicities over southern third of MI and extreme winds & CAPE in IN/OH.

- Rob

I'm getting ready to run the WSETA on a 10KM grid using non-hydrostatic and no convective scheme (DTX suggested the no-convective scheme approach)... I am using the Tile 12 ETA to initialize) I could always use the GFS though...

 

[Glen Romine]

Anyway, I am still trying to understand convective feedback and model physics/etc. I thought that if the model falsely initiates convection (on either scale), that it would contaminate the downstream grids in several ways -
1) Thunderstorm outflow/clouds would cause temps to cool
2) Model soundings for the particular region would show almost complete saturation (since the model is "raining)
3) Not sure, but wouldn't it also cause intense VVs?

All of which would be false, since there is no convection occuring (am I making sense? LOL)

You are correct that all of these things will make changes in the model forecast - and if the evolution doesn't reasonably match the forecast, then the model output is progressively degraded with time. I'd probably call this convective contamination though - not convective feedback. The "feedback" is the key term here - once convection occurs on the oversized scale (for example grid scale convection on say an 18 km square grid - which is one big updraft!!!) will cause considerable latent heat release within the model - quickly generating a small, compact and intense disturbance with mucho precip - and the impact to surrounding areas is strong convergence leading to further grid scale convection. This occurs on a much smaller scale with any model generated precip - but appears like grossly amplified in the case of true convective feedback. The AVN model is notorious for convective feedback problems - often showing up as a ringed bullseye of precip from seemingly no apparent forcing. In this case - the forcing is clearly evident in the model - and the upscale growth is at least reasonable. Will it actually happen? Guess we'll find out soon enough.

Glen

 

[Blake Naftel]

Will await 0Z ETA for further discussion; however with a quick perusal of the 12Z ETA; it looks like Wednesday will be a north-central/central Indiana day. Favoring ETA rather than GFS; however like Amos will let real-time data and RUC forecasts make the final verdict if and when I may depart KZO following my midday shift.

As always; surface boundaries and the ongoing MCS will come into play; however a nice CAPE bullseye (via the 12Z ETA) points just north of Greentown/Kokomo. I wasn't aware of such strong earlier wording as noted in previous threads; but can see where it may be warranted with backed easternness; coupled with 850, 700, and 500 jet maximums close to/and or during peak heating. Agree with current SWODY2; but could foresee a possible upgrade to MDT.

I'll let the atmosphere dictate chase potential rather than a few forecast models as this is close to home.

..Blake..

 

[Jon Miller]

Indianapolis & Wilmington HWO's are now mentioning possibility of isolated tornadoes Wed afternoon: - but northern Indiana & Louisville don't. - looks like central to eastern Indiana is in the running for now.

 

[Glen Romine]

00Z ETA continues to promote the idea of a strong MCS maturing over IA tonight, then traversing IL during the early morning hours, into IN, and then off to OH by evening. Earlier runs suggested less of a cold push behind this feature, with a weaker vort max sliding in behind the stronger lead wave. The upper feature remains, but new ETA has stronger cold push, leaving the disturbance to traverse a rather stable air mass. Would expect SPC to go with moderate risk tomorrow - as deep layer shear aligns very favorably for a derecho type event. As for tornadoes - I think the risk is minimal - but might peak during the early afternoon as the lead disturbance appears to outrun the main instability axis by late evening. Of course, current obs trends do not agree with the above ETA solution, as it failed to maintain the convective cluster currently heading towards Chicago - and latest radar check shows this feature is alive and well. So, I think the moral of the story is wait until morning, and see what reality has to offer.

Glen

 

[Jim Bishop]

Someone's going to capture video of a great tornado tomorrow. I'm looking forward to seeing photos/videos of the event. I think tomorrow could be huge!

 

[Amos Magliocco]

I like the look of the 0Z ETA though the instability is less. The timing of the vortmax seems a little more favorable to me and I agree with the early afternoon timeframe, from 17 to 22Z I would guess.

What's not so encouraging is the system blasting through northern Indiana right now. I don't know how much that will turn things, but it sort of renders the model output even more tenuous than normal for me. Guess we'll see what's what in the morning.