Forecasting Svr. Weather...something extra for your tool box

[QVectorman]

Here are 2 papers explaining the of using vertical velocities from models to help identify enhanced areas of risk for severe wx/tornadoes and a LITTLE better (haha to say the least) explanation of the logic behind it.

The first article mentions specifically and infers that when forecasting svr. storms and tornadoes and the Kain-Fritsch convective scheme is used in the model, optimal results from this particular scheme and in forecasting said weather is achieved when the vv's process is applied on top of the Kain-Fritsch convective scheme. Possibly explaining why Fritsch endorsed this forecasting process.

Also notes more eloquently than I did at conveying the message that the explicit timing, intensity, and location of the updrafts hence, svr. storms and tornadoes should be considered as qualitative, rather than as quantitative, forecast guidance. And that the process should not be used in an attempt to target forecast meso/microscale events but rather identify geographical regions that favored such.

http://wwwtw.vub.ac....PhDDDehenau.pdf

"The author thinks the vertical velocity (VV) at the 500 hPa level is also important

when searching for areas where large updrafts at grid scale is important, especially if

a mass flux convection scheme, like Kain-Fritsch, is used. The more negative VV, the

stronger the updraft and the thunderstorm. Strong updrafts suggest severe turbulence

and a possibility of strong gusts. In some cases presented hereafter, it was a very

useful tool for diagnosing severe thunderstorms."

"Vertical velocity at 500 hPa in high resolution models gives a clear indication of

updrafts in supercells but also in other weather systems, such as fronts. As we try to

calculate the risk of supercell tornadoes as well as strong convective gusts, vertical

velocity is a good indicator of strong updrafts. Without updrafts, tornadoes are

impossible (Brooks, 1994a). It needs to be stressed the vertical velocity is strongly

correlated with the convection scheme. The vertical velocity is resolved on the

model’s grid scale. The role of a convection scheme is to remove instability. If too

much instability remains, convection on a grid scale occurs, causing updrafts in a

small area. If the vertical velocity remains large, less instability is removed by the

convection scheme. It implies the results of the STP-approach, influenced by grid

scale vertical velocity, are strongly related to the type of convection scheme and the

resolution of the model."

"A good indicator in this case is the 500 hPa vertical velocity VV, giving an idea of the

updrafts on a meso scale. The more negative VV is, the stronger the updraft and the

thunderstorm. Strong updrafts suggest severe turbulence and a possibility of strong

gusts. The maximum updraft is in the vicinity of Herne, where a strong tornado

caused extensive damage"

http://journals.amet...0.1175/WAF907.1

"Given these complicating factors, model vertical velocity and quantitative precipitation forecasts (QPFs)can aid in the assessment of band potential, providing evidence to support or refute the hypothesis of band formation. The hypothesis of band formation is supported when a narrow (relative to the resolution of the model) region of forecast ascent and precipitation corresponds to the forecast location of midlevel frontogenesis and weak moist symmetric stability (Schumacher 2003). Although the explicit timing, intensity, and placement of a model forecast precipitation band may be in question, the fact that the model dynamics are producing banded precipitation suggests the presence of a favorable environment. Precipitation fields from high-resolution models can provide additional insight into the nature (i.e., movement, intensity, and dissipation) of the precipitation event given the fine horizontal resolution and hourly accumulation periods of these fields. Similar benefits of using high-resolution model precipitation fields have been demonstrated by Roebber et al. (2002) for convective precipitation events.However, at 12–24-h forecast projections, the explicit timing, intensity, and location of the band should be considered as qualitative, rather than as quantitative, forecast guidance (Roebber et al. 2004)."

This is the 21z 700 mb VV forecast from the NAM where I noted the enhanced areas of +VV over the OH River Valley where at 21z multiple tornado warnings were being issued. Note the thin narrow band of max lift paralleling the OH River...it's shape length and wide mimics the SPC reports of a thin narrow band of tornado reports in the same region that were reported pretty much + -1 hr 21z. Other areas of enhanced +VV on the map did not enhance convection in those regions because either CAPE, sfc forcing or shear were not ample or not present together. Where all 3 are present and in combination with progged +VV enhanced threats of svr. wx and tornadoes is present.

[QVectorman]

I will also add this for non met crowd.

METEOROLOGIST JEFF HABY

(1) Study the 700mb vertical velocity progs on the NAM, and GFS, paying close attention to your forecast area. Take note whether thevertical velocity is upwards or downwards. Upward motion (+UVV's) are caused by low level convergence and/or upper level divergence. A UVV of 6 to 9 is moderate, 10+ is large.

(2) Study the 850mb progs and see if your forecast area will be experiencing warm air advection, cold air advection or neutral advection. Low level warm air advection leads to upward vertical velocity while low level cold air advection leads to downward vertical velocity. Neutral thermal advection will neither inhibit nor enhance upward vertical velocity. Examine 1000-mb prog for low level convergence resulting from fronts, topography, low pressure, WAA and moisture advection.

(3) Study the 500mb vorticity panels and see if your forecast area will be experiencing positivevorticity advection or negative vorticity advection. PVA will lead to upward vertical velocity while NVA will lead to downward vertical velocity. Small values of vorticity will neither inhibit nor enhance upward vertical velocity.

(4) Study 300mb panels. See if any jet streaks will influence your forecast region. The divergence sector of a jet streak will lead to upward vertical velocities. These sectors are the right rear and left front quadrants. If the jet is in a highly curved flow, then the divergence occurs anywhere just north of the jet axis in the Northern Hemisphere. If the upper level winds are weak, this will neither inhibit nor enhance upward vertical velocity.

(5) Study thermodynamic diagrams to assess the potential for convective precipitation and precipitation type. Pay attention to moist/dry layers, wind profile, and indices.

(6) Ask yourself which phenomena (WAA, PVA, Streaks) are leading to upward vertical velocity and which phenomena (CAA, NVA, Streaks) are leading to downward vertical velocity.

(7) Take note of the variation between the graphical forecast models on the forecasting of

synoptic scale precipitation.

If vertical velocities are positive, and no precipitation is progged, then there is likely not enough available moisture. Check RH panels to see if uplift is enough to saturate atmosphere. PBLdewpoints can be used to assess how much moisture can be lifted.

(8) Decide whether precipitation is likely or unlikely and decide if severe weather is likely or unlikely. Decide on the following characteristics of the precipitation:

*Convective or stratiform

*Heavy or light

*Widespread or numerous

*Severe or non-severe

*Long lasting or brief

*Wintry or non-wintry

(9) Read NWS convective discussion, zones forecast, state forecast discussion and state forecast.

(10) Examine MOS precipitation data from several synoptic scale forecast models

(11) Look at national and local satellite and radar data as well as a current surface map. Note the features on each of these three sources and ask yourself how they will move through time.

(12) Write out your precipitation forecast, and the expected character of the precipitation if you expect precipitation to occur. Forecast should include amount, duration, type, intensity and other unique characteristics of the precipitation given in (8).

NOTES:

*If vertical velocity is significantly upwards and PVA and jet streaks are not present, then precipitation development is expected to be thermodynamic in origin (low level buoyancy, low level forcing) or the result of low level convergence.

*If vertical velocity is significantly upwards and little WAA or low level convergence is present, then precipitation development is expected to be dynamic in origin (upper level forcing by PVA and / or jet streak).

*If all three +UVV mechanisms are significantly present in the same vacinity (WAA, PVA, Streak), then expect severe storms or heavy precipitation. (must also have low level moisture, deep instability)

*If none of these UVV mechanisms are present, then precipitation can only occur by a mesoscale process (outflow boundaries, sea breeze, orographic lifting, shallow fronts, air mass thermodynamic thunderstorms).

*All forecasting rules of thumbs have exceptions. Add your own intuition to determine precipitation potential.

[icebreaker5221]

I think it's a good thing to play devil's advocate once in a while and ask yourself "what can go wrong", and you certainly played that role these last couple days. However, I'm not sure this was the right setup to bang the subsidence drum.

Certainly strong subsidence can be an issue. I once had a complete bust storm chase in OK back in 07. There was a tornado watch up, strong low-level jet, extremely high CAPE, and even a surface boundary (might have been a dryline but can't remember for sure), yet we got nothing. The problem was that a ridge was building in, and we were under negative vorticity advection, and hence, quite a bit a subsidence.

However, on a day like yesterday where you had both isentropic lift and positive vorticity advection, it doesn't even make sense to look for subsidence. You might be able to find it in small pockets here and there, but it's definitely not going to be an issue on the large scale. IMHO.

[CUmet]

There may have been some misunderstandings earlier or currently, so if I have this wrong than please correct me. From what I gather over the last few days of posts, there were two overarching issues:

1. The importance of synoptic scale lift/subsidence on severe convection.

2. The utility of using global model vertical velocity output in severe convection forecasting.

With regard to #1, severe convective forecasting is complicated and at times counterintuitive to the synoptician in that more synoptic scale lift does not necessarily lead to more severe weather. Certainly, one needs a lifting mechanism in order to have parcels reach their LFC, above which they can work off of convective instability. However, synoptic scale lift is one of many possible mechanisms to achieve this (surface heating, moisture convergence, local topographic effects, etc.). Additionally, subsidence can be a problem if it produces a warm layer in the mid levels (i.e. a capping inversion) that is too strong for lifting mechanisms to overcome. Sometimes, though, this subsidence-induced cap can be a good thing, as it clears out everything, resulting in surface heating and building up CAPE.

Another point I want to make is that once the parcel reaches its LFC and the subsequent updraft takes off, whether or not there's a huge amount of synoptic scale lift is largely irrelevant. Generally speaking, whatever synoptic scale subsidence there is isn't going to offset the updraft that's going up at an order of magnitude faster. It might, however, prevent too many parcels from reaching their LFC, keeping the overall thunderstorm coverage more scattered and discrete. All this to say that the "ideal" amount of synoptic lift for severe weather outbreaks differs (in both directions, more and less synoptic lift) on a case by case basis.

Regarding #2, I agree that there is utility in using global model vertical velocity. It can help identify whether the model has initiated convection, and by association whether the model breaks whatever capping inversion there is. However, one has to be careful in interpreting this, especially in the context of trying to evaluate the severity of an upcoming severe weather outbreak. Many experienced severe weather forecasters know that it's not necessarily a good thing (for severe weather) to see the model output large vertical velocities, and that it's not necessarily a bad thing to see it output very small vertical velocities. I suppose if one where to forecast for overall thunderstorm area coverage or QPF, yes you would want lots of VV. But forecasting the magnitude of a severe wx outbreak and assessing tornado/hail/wind risk is a lot different in that you can have a few scattered cells doing all sorts of damage, but these cells would have a small signature in the model's VV and QPF fields.

Again, QVectorman, I don't want to insinuate that you're not aware of the above issues, but at least from your posts it wasn't clear to me that you were. Also, this may prove helpful for less experienced readers of this board who may wish to improve their own understanding of the complexities behind severe weather forecasting.

[wxsmwhrms]

Just realized there was a new thread on this, posted this in the other 3/2 thread.

Thanks for posting this article. I skimmed the article and got some interesting stuff out of it. I am going to be honest, my questioning of some of your posts a few days back were more that I was not seeing the same things you were in the data (as others mentioned as well). You seemed to be indicating a lack of upward motion collocated with the strong instability in the southern part of the outbreak, and I was not seeing that at all. It seemed clear that strong jet support would be spreading over MS/TN/AL during the afternoon and evening yesterday given the model forecasts of vertical motion, jet placement, and QPF, and that in fact occurred. It was actually a really cool case where you could see the jet ascent moving into the Arklatex on water vapor during the early afternoon, and convection lit up along and ahead of the front as that happened. The models forecast that very well for several days.

Obviously, every time there is a big weather event, there will be people presenting pros and cons leading up to it as to why it will or won't happen. But one thing I think needs to be stressed is that EVERY severe weather event has potential failure modes if one looks hard enough - but that doesn't mean that the event will fail. To me, when you have as dynamic a setup as the models were forecasting with this event combined with the potent instability yielded by the fact that the lead system earlier in the week made it easy for this system to tap into, you can go too far into the weeds trying to find reasons why an outbreak won't happen. Furthermore, once you get within 36 hours of the event, I don't know why you would be looking at synoptic model forecasts of vertical motion to try to find convection as this 7 year old paper describes. We now have high resolution models such as the 4km SPC/NSSL WRF models that explicitly forecast convection. The SPC WRF did an awesome job with yesterday's event as far as convective evolution, and I have been finding it to be a great tool. Not perfect, of course (didn't do real well with the early week event), but IMO a better tool than trying to use NAM vertical motion as a proxy.

To me, you look at the synoptic models to try to get an idea of the mass fields, the synoptic vertical motion, etc., and then use high resolution modeling to try to anticipate initiation and mode as you get closer to the event. You will obviously still have severe convective forecast busts because we are not perfect and models aren't perfect, but I think we are seeing huge busts (like high risk events where nothing much happens) becoming more and more a thing of the past.

Just my two cents...

[QVectorman]

I think it's a good thing to play devil's advocate once in a while and ask yourself "what can go wrong", and you certainly played that role these last couple days. However, I'm not sure this was the right setup to bang the subsidence drum.

Certainly strong subsidence can be an issue. I once had a complete bust storm chase in OK back in 07. There was a tornado watch up, strong low-level jet, extremely high CAPE, and even a surface boundary (might have been a dryline but can't remember for sure), yet we got nothing. The problem was that a ridge was building in, and we were under negative vorticity advection, and hence, quite a bit a subsidence.

However, on a day like yesterday where you had both isentropic lift and positive vorticity advection, it doesn't even make sense to look for subsidence. You might be able to find it in small pockets here and there, but it's definitely not going to be an issue on the large scale. IMHO.

Fair enough I agree And usually, as they say, what can go wrong will go wrong. That has happen a couple times the past few years more so than not. But we can say without a doubt Friday nothing went wrong, it played out pretty much perfectly every where the ingredients were present.

I believe any given svr wx day searching for +VV in the atm. can be an additional tool in finding "sweet spots". Especially when you area trying to narrow down to a single location to chase and all other parameters are fairly equal across different areas. +VV's are usually the last thing in my list I look for and usually use it as the tie breaker to pick locations based on best +VVs.

And I'm with you on the OK chase. On the Plains I have seen/had several busted storm chases and after careful post analysis came to the conclusion that the busts coincided with -VV being progged in the region on the models (results of...NVA and/or CAA etc) and after 2 or 3 busts mine and SPC is when I started checking out -VV areas along with +VV areas.