Let's talk about forecast reflectivity and vorticity

[Rob H]

A lot of people look at forecast reflectivity and the day of a chase, you'll usually see people posting on Facebook "look at that huge discrete cell the WRF breaks out in my target area!"

Relying on a forecast graphic that abstracts away most of the "real" data can be very dangerous, although we're getting to the point where you can (almost) get away with it. I still thought it would be a good idea to try and figure out what's going on behind the scenes, so bear with me, and feel free to add anything to the conversation or answer questions

One thing I've noticed is that if you have enough moisture and instability, areas of forecast precipitation are usually co-located with areas of 500mb Absolute Vorticity:

500mb Absolute Vorticity:
http://i.imgur.com/icxzXzQ.png

Composite Reflectivity:
http://i.imgur.com/OxoeS3m.png

The conventional method of forecasting synoptic scale lift is to look for the vorticity maxima, and infer precipitation near and East of it, if I recall correctly. Looking at a chart that's not as noisy as the HRRR (we'll use Unisys' 500mb NAM/WRF). It looks like a vort max is in extreme northern MN, and very little vorticity is in the target area that the HRRR alludes to:
http://i.imgur.com/bVr95bS.gif

So the HRRR is obviously seeing things that the Unisys NAM/WRF can't. My initial questions about the images above:

1. With the other necessary ingredients present, are these tiny pockets of vorticity the lift, or "trigger", for severe convection as the images would suggest?
2. What causes these pockets of vorticity?
3. How can you forecast these pockets? Do you need to infer them in certain synoptic scenarios?
4. How can you nowcast these pockets? Will they appear on water vapor or visible satellite?

 

[Jeff Duda]

Rob, the small pockets of vorticity you see in the HRRR charts is just a fact of increased model/grid resolution. Absolute vorticity is defined as

abs. vorticity = f + dv/dx - du/dy

where f = 2*omega*sin(phi) is the planetary vorticity that is only a function of latitude, and dv/dx and du/dy are wind component derivatives in the east-west and north-south components. The vorticity pockets in HRRR charts are a result of the large-scale flow interacting with individual thunderstorms (probably the updraft portion) that block the flow and cause storm-scale flow anomalies. These anomalies result in increased wind gradients, which results in the appearance of vorticity. Quasigeostophy is not valid on the storm scales, so those little pockets are not a triggering mechanism for storms. As already indicated, they are actually the effect, not the cause. However, even if such a pocket were to exist without a nearby storm, the vorticity in the pocket would not increase the chances for thunderstorm initiation ahead of it.

As the resolution of the forecast models increase, the scale of features that can be resolved decreases. Since the atmosphere is continuous and motions exist on all scales right down to the molecular, this process of finding new smaller scale feature every time the model resolution is refined will continue.

Welcome to the next generation of weather forecast models. I've heard rumblings that the CONUS nest of the NMMB (currently run at 4 km) may be upgraded to 3 km next year.

 

[Patrick Marsh]

Great questions Rob. Here are my responses

Those vorticity "splotches" (highly-technical term!) are actually the result of the model thunderstorms. They aren't necessarily the cause of the thunderstorms. So you shouldn't really be forecasting for them, as they don't appear until the thunderstorms have already developed. That's why you see them in the HRRR and not the NAM. The NAM isn't developing thunderstorms intense enough to be manifest in the vorticity field in mid-levels.

<professor mode>
The old rule of thumb about vorticity maximums and "looking to the east" stems from Quasi-Geostrophic theory.

In QG-Theory, one expects rising motion whenever you have increasing cyclonic vorticity advection with increasing height. In general, we assume vorticity advection at the surface is ~0, thus, cyclonic vorticity advection at H500 implies increasing vorticity advection with increasing height. One interesting aspect to this is that if you have anticyclonic vorticity advection at H500, but have strong anticyclonic vorticity advection at the surface, you still have "increasing cyclonic vorticity advection with height", and thus still have rising motion. So it isn't always as simple as finding vorticity maxima! (Note, the proper way to state "near and East of [the vorticity maximum]" is "downstream of the vorticity maximum", because you never know the wind direction, and the vorticity advection may not be to the east!)

One thing about QG-Theory is that you really should not apply it to storm-scale models. QG-theory is really geared toward coarse model resolutions (~80 km and greater). As you move toward the higher resolution models you begin to get "noisy" vorticity minima and maxima fields that make the use of QG-Theory extremely (in particular the advection of cyclonic vorticity advection) difficult.

When you are using high resolution numerical guidance, you really do not need to use QG-Theory. All QG-Theory gives you is an inference of where ascent may or may not occur. With the high resolution models, often times you can investigate that explicitly.

At the end of the day, large scale ascent identified by QG-Theory really doesn't help you with forecasting exact location of storms. If you think about it, large scale ascent identified by QG-Theory is on the order of a few centimeters per second, whereas thunderstorm updrafts are on the order of meters per second. What the large scale ascent does is lift the CAP/warm layers aloft, which actually cools the CAP/warm layer. (Note, you don't want to remove the CAP via cold air advection. This results in subsidence/warming and actually strengthens the CAP. You remove CAPS via ASCENT!)
</professor mode>


Lastly, I use the explicit thunderstorms in high-resolution guidance as an indication that the model believes the cap has weakened sufficiently to allow storms to develop. Furthermore, the mode of the thunderstorms in the model gives some indication as to how he model is evolving the CAP (i.e., weakens it everywhere [lots of storms] or if it only breaks in isolated areas [isolated storms]).

 

[Patrick Marsh]

I've heard rumblings that the CONUS nest of the NMMB (currently run at 4 km) may be upgraded to 3 km next year.

NMMB is currently expected to remain at 4KM through FY14. It may increase in FY15.

 

[Bob Hartig]

I just noticed this thread. It's a great discussion! Thanks to Rob for stating the question so well and to Jeff and Patrick for answering it so explicitly. I had wondered the same thing about PVA--to wit, how the heck do I use it? Absolute vorticity was the first forecast map I ever became aware of, and the GFS was the first model I ever used--exclusively for quite a while; I knew no better--and all I understood about that combination was that all those colors sure looked pretty.

I came to an understanding, similar to Rob's, that I should look for convection downstream from enhanced vorticity. Sometimes that seemed to work; sometimes I felt clueless. Then along came the hi-res stuff, and it seemed to me that the vorticity was so linked with actual forecast reflectivity that it had no real forecasting value. It appeared to be a chicken/egg phenomenon, and I did in fact deduce that it was probably an effect rather than a cause of forecast convection. So I was right! That's a small but sweet victory for a guy who often feels like he's still learning how to tie his shoes when it comes to understanding some of this stuff.

What I don't get, though, is Pat's professorial comment:

Note, you don't want to remove the CAP via cold air advection. This results in subsidence/warming and actually strengthens the CAP. You remove CAPS via ASCENT!

I had thought that one of the values of the 700 mb map was to indicate the removal of the cap via cold air advection. If I observed cooler air moving toward a capped area, typically between 850 and 700 mbs, then I assumed that capping would begin to weaken. Is my reasoning wrong? How does cold air advection result in subsidence and in strengthening rather than erosion of the cap?

 

[Paul Schmit]

Bob,

I'm sure Patrick will be able to answer this with more of a commanding authority of the fundamentals and nomenclature, but here's my two cents. One great way to think about atmospheric potential instability is looking for negative vertical gradients in the equivalent potential temperature (Theta_e). This negative gradient is exacerbated by warm, moist air at the lower levels, combined with decreasing temperatures and decreasing moisture as one travels up the air column. The implication of a negative gradient in Theta_e is that synoptic-scale lifting tends to increase absolute vertical lapse rates, which can cause a cascade into absolute convective instability. Basically, as the entire air column rises, the warm moist air near the surface only cools a little (something like the wet adiabatic lapse rate), while the drier air above cools much more substantially (at the dry adiabatic lapse rate), and so whatever lapse rate was there to begin with winds up even steeper once the broad-scale lift sets in.

So, considering cold-air advection near the CAP layer, here's how I interpret Patrick's statement. The CAP would be a layer of air where the local vertical gradients in Theta_e are far less negative, if not positive, and the same could be said about the absolute lapse rate. If we have localized CAA in the vicinity of the cap, the tendency is for the air column to sink, so now we must consider the implications for Theta_e and the lapse rate. If the adiabatic thermodynamics of a slowly rising air column in the presence of negative Theta_e tends to result in more intense absolute lapse rates after the column is lifted, then we can reverse that picture in the presence of CAA and say that sinking of the entire air column will compress the layers of air comprising the CAP and reinforce the convectively prohibitive lapse rates within that layer. So, if the CAP is characterized by a weak, but still negative, vertical gradient in Theta_e, we must get lots and lots of lift in order to make the overall lapse rates favorable for convection, and localized CAA through some layer will only reinforce the CAP above and below that layer through the vertical sinking of the atmospheric column.

Might be some flaws in my logic, but I think this is the gist of what Patrick was indicating!

 

[Jeff Duda]

Don't let that statement confuse you into thinking that CAA does absolutely nothing to weaken a cap. That's not true. CAA can and does weaken caps. However, since CAA occurs with sinking air on the synoptic scale, and sinking air heats by compression, there is compensating warming to counter the CAA. However, I'm pretty sure that this warming from subsidence in association with the CAA is not enough to completely negate the cooling due to advection. CAA will still result in cooling at a given level.

You can see this for yourself. Find a rawinsonde site that regularly sends balloons. Find a case where there is notable CAA occurring (like with the recent trough that just went through the CONUS). Calculate the temperature change over a 12-hour period due to advection (the formula is V_ (grad)T in vector form, which in component form is -(u-wind component)*deltaT/deltax -(v-wind component)*deltaT/deltay. Multiply that by 12 hours (make sure you convert to the same time units so they cancel), and add it to the earlier observation time. You will find that the temperature at the later observation time (at the end of the 12-hour period) is warmer than your estimate using just CAA. You can find out how much impact the compressional warming due to subsidence had by looking at the difference between your estimate and the actual observed value. It helps if you find a station that is in a steady-state of temperature advection and with no microphysics processes occurring to influence temperature change through water phase changes. You won't find a perfect scenario, but your estimate won't be that far off from what's really going on.

Looking at the bigger picture, this is why I prefer isentropic analysis. I made a post about this a few years ago (http://www.stormtrack.org/forum/showthread.php?28489-Isentropic-surfaces). While the case study focused on a winter event, the argument still applies to warm-season events as well.

 

[Patrick Marsh]

At a given level, CAA will result in a weakening of the CAP, as by definition the temperature at that level is cooling. However, one needs to think in the vertical dimension. Due to the associated subsidence, the CAP will subside to a higher pressure level (closer to the ground), resulting in a warming of the temperature trace. (You can see the effects of this around H500 in the wake of many shortwave troughs!)

Now, the resulting strength of the CAP depends on the actual environmental temperature trace and also on the strength of the CAA.

 

[Bob Hartig]

Gentlemen, thank you all. My head hurts from some of it, but I think I get the gist. Basically, because colder air is heavier air, CAA compresses the air column, forcing the capping inversion closer to the ground while simultaneously warming it up. Have I got that right? So while cooler air advecting in may drop the temperature by, say, 3 degrees at H7, erasing the cap at that pressure level, yet below it, the temperature at H8 may in fact heat up somewhat.

Do I rightly assume that forecast models factor this in--and if so, which ones do a better job of it? Or must I second-guess the models?

 

[Patrick Marsh]

All models do factor all of this in via some way, shape, or form. However, not all models do it the same way. Generally speaking, it is my opinion that if you're looking for convective initiation, then storm-scale models integrate this the best as they are explicitly resolving the generation of a thunderstorm updraft...not some parameterized approach to initiation.

However...no model will be perfect 100% of the time. Hence, the presence/absence of CI does not mean thunderstorms will/will not occur. =)

 

[Bob Hartig]

Thanks, Patrick. It's helpful to know that.