SB CAPE vs ML CAPE

[Andrew Clope]

I know there likely isn't an easy answer for this question. But I was wondering what you guys use as a guideline on which one to use when forecasting and nowcasting?

What are the pros/cons to each in the analysis process?

I would imagine that ML CAPE infers elevated convection rather than sfc based? But I could be completely wrong on that.

Also, sfc based always seems substantially higher. so I could see why folks would want to use that one instead.

 

[nickgrillo]

I'll make it short. Using a ML parcel gives you a better idea of things, specifically PBL moisture depth. For instance, if CAPE is higher using a surface-based parcel, it's usually due to shallow moisture (often the result of convective mixing in the afternoon).

"I would imagine that ML CAPE infers elevated convection rather than sfc based? But I could be completely wrong on that."

No, that's what CINH is used for. For instance, if CINH is very large using a mixed-layer or surface-based parcel ascent, that means any storms which do develop will be drawing in air above the layer of CINH (the inversion or the "cap"). This does not neccessarily apply to ongoing supercells, which have the ability to alter their surrounding environment. They can continue to draw in PBL air despite strengthening CINH.

"Also, sfc based always seems substantially higher. so I could see why folks would want to use that one instead."

Yes. And in some cases, that is evidence of shallow moisture, which is bad for storms (if you're looking for tornadoes, anyway).

 

[Greg Blumberg]

ML means "mean-layer or mixed-layer (I've seen them be interchangeable)" and SB means "surface-based". These are referencing to the type of parcel that is being used to calculate CAPE/CIN/LI/EL/LCL/LFC/etc. SB means that you are choosing the temperature and dew point value at the surface as the initial characteristics of your theoretical chunk of air (also called parcel) that is being lifted throughout the depth of the troposphere. So, SBCAPE is the CAPE that you would get if that parcel is what is being drawn upwards into an updraft. Mean-layer, often refers to a parcel whose initial characteristics are determined by taking an average of the temperature and dew point values within the lowest 100 mb of the atmosphere. MLCAPE will almost always be less than SBCAPE, and it is often more realistic because the averaging is roughly approximating the fact that a parcel's characteristics will change as it mixes with the air around it while rising through the boundary layer. The surface-based parcel doesn't include this effect. However, MLCAPE also requires that you have the thermodynamic characteristics of the lowest 100 mb well-represented. So in general, SBCAPE is more resistant to errors in moisture and temperature than MLCAPE, but MLCAPE can be more realistic. Both are useful, but it depends upon the situation at hand and also knowing how they are calculated when you apply them.

The MU "most unstable" parcel is often used for elevated convection. It will tell you which parcel (from what level in a profile) you need to lift to get the maximum instability. Sometimes this is a layer that is above the surface, which is often the source region for elevated convection.

 

[Kelton Halbert]

Greg pretty much nailed it.

As far as when/how to use it, it can vary from situation to situation.

The ML parcel will give you a much more realistic CAPE profile when your surface observation for an observed sounding profile seems wrong, or if it seems skewed to higher values by a shallow moist layer. Also, as Greg said, it better handles the turbulent mixing that takes place in the boundary layer. It is often assumed that a parcel is not allowed to mix with it's environment when it rises, which is often not true, so the mixed layer parcel is more physical.

You could consider it part of an ensemble of parcels. Rather than just looking at one, you consider other processes that might be taking place within the profile. No one parcel is perfect, and perhaps lower MLCAPE vs. SFCCAPE could help reveal issues with the forecast for severe convection by revealing thermodynamic characteristics that are unfavorable for strong updrafts. Or likewise, stronger MUCAPE vs. SFCCAPE could reveal the potential for elevated convection rather than surface based convection.

 

[Jeff Duda]

However, MLCAPE also requires that you have the thermodynamic characteristics of the lowest 100 mb well-represented. So in general, SBCAPE is more resistant to errors in moisture and temperature than MLCAPE, but MLCAPE can be more realistic. Both are useful, but it depends upon the situation at hand and also knowing how they are calculated when you apply them.

I would actually argue the exact opposite is true. Any errors in the mixed layer are likely to be smoothed out when taking an average. I have seen plenty of observed soundings where the SBCAPE is very unrealistic because of bad or unrepresentative surface obs (e.g., superadiabatic surface layer). Other than that I agree with your post.

Keep in mind that the ML parcel isn't a real parcel, at least not when first lifted. MLCAPE is meant to approximate the effects of entrainment on the resulting CAPE of a parcel as it rises through the PBL. Usually an arbitrary depth (100 mb) is assumed, I guess maybe for consistency, but to be consistent with the theory behind using it, the source parcel characteristics should be averaged over the depth of the PBL, which can vary wildly from one location to another within an event, or from one event to another at a single location. I'm not sure why that's not done.

Last thing: MLCAPE is generally best applied during the day when turbulent mixing in the PBL is strong. It can be useful after sunset, too, but SBCAPE becomes more realistic then once the boundary layer starts to decouple, especially if you don't have storms already present.

 

[Greg Blumberg]

I would actually argue the exact opposite is true. Any errors in the mixed layer are likely to be smoothed out when taking an average. I have seen plenty of observed soundings where the SBCAPE is very unrealistic because of bad or unrepresentative surface obs (e.g., superadiabatic surface layer). Other than that I agree with your post.

That would be true as long as all errors are normally distributed.

I like the concept of the ML parcel not being a real parcel. I also agree that the parcel characteristics ought to be averaged over the depth of the PBL. I'm going to have to think more about the use of MLCAPE/SBCAPE around the time of the nocturnal transition...

 

[Tim Supinie]

I think it's time to mention that the surface point on a sounding is not from the balloon package itself, it's from a surface station in the vicinity of the launch point. If that station is wrong (as it was on the 00Z 25 April 2015 FWD sounding):



then it's the MLCAPE that's more accurate.

Sometimes the surface point is right and the sonde above it is wrong. You see this a lot in the moisture profiles because the moisture sensors on the radiosondes (and in general, really) can be kind of finicky. One example of this case might be the 00Z 1 August 2012 OUN sounding:



In this case, it might be the SB quantity that's more accurate (even though it really doesn't matter here). This case is a little weird, though. In summer, you can actually get these skin layers of moisture caused by all the plants on the ground pumping out water. However, a) that's a pretty huge moisture drop right above the surface for that to be the case and b) this was in the middle of a raging drought, so the plants probably weren't transpiring that much. This one might actually be a case where both are wrong, but in opposite directions (sonde is too dry, surface is too moist).

I've only mentioned which is more accurate from a data accuracy standpoint. Which is more accurate and which is more useful given the situation, as Jeff mentioned, might not be the same.

 

[Paul Knightley]

The problem with looking at model MLCAPE values is that you're generally at the mercy of what depth of layer is used (often 100 hPa but not always - happy to have extra input here!). In reality a storm never lifts a 'parcel' - it's just what we use to simplify the process to make it easier to model and understand. The storm is a continuous process. However, sounding and parcel analysis is a useful conceptual model. The depth of the parcel we 'lift' should really be varied depending on the situation (as Jeff says above, if you expect deep mixing of the PBL you need to lift a 'deeper' parcel'). You can modify soundings yourself and lift parcels - for example, using RAOB software.

 

[Steve Miller TX]

Personally for me, I always use MLCAPE in forecasting for severe thunderstorms. After all, there is alot more than just a surface parcel being ingested by an updraft and I think the MLCAPE is more representative of that. I also consider SBCAPE to be a good parameter for tornado forecasting as I consider the SFC and near-SFC environment (a few hundred feet above SFC) to be more critical once a supercell matures. I have no scientific studies available off-hand to support this, just my observations over the years.