Useful parameters for forecasting lightning

Thomas Mattek

Enthusiast
Apr 2, 2018
3
7
1
Missouri
Hello! I wrote the following as an answer to a question in the Discord and thought it may be a useful reference/discussion point.

The original question was as follows:




"Is there a parameter or set of parameters that can determine if storms will produce more CG lightning then others? Some storms produce a lot while some just dont. Most of the ones that hit here dont much. Just periodic surprises. But the back side of last nights mcs had several in a row somewhat spaced out"
-Chris(Western WI) at 12:11 AM, August 7th, 2019


I answered as follows, following on the source listed in this old thread on the subject:

This article, listed on NWS Raleigh's site, on a major lightning event from 2008 sums up some possible parameters, based on the idea that graupel and hail in the mid-levels are important:
Parameter 1)Availability of moisture within the -10 to -30 temperature depth range, usually anticipated by using Precipitable Water(PW)

Parameter 2) -10 to -30 layer CAPE. Higher instability is associated with more graupel and hail formation in this layer.

Parameter 3) Normalized CAPE or NCAPE, looking for above 0.1 or 0.2 to determine how "fat" the CAPE is. Parameter

4) K-Index, based on mid-level laspe rates and moisture.

Parameter 5) Finally, the NAM and SREF have probability indices of 100 or more lightning strikes, detailed in the following paper: 4.3 A physically-based parameter for lightning prediction and its calibration in ensemble forecasts (2005 - Annual2005_lightning)

With regards to parameters, it looks like only N-CAPE is available on current SPC mesoanalysis. If anyone knows of any models that include K-Index and -10 to -30 CAPE that would be appreciated!



There's also the following lightning checklist included:
 

Attachments

  • Like
Reactions: Jeff House

rdale

EF5
Mar 1, 2004
6,942
468
21
49
Lansing, MI
skywatch.org
The ECMWF has lightning forecasts but for short-term use I'd look at MRMS. It has a 0-30 minute lightning strike forecast, and AllisonHouse is testing a push alert system for that.

(NOTE: The NAM and SREF have been sunset by NCEP so no further development work is being done on those models before they are cut off.)
 

Dan Robinson

Staff member
Jan 14, 2011
2,501
2,168
21
St. Louis
stormhighway.com
In my experience, this has been an enigma that I've yet to figure out. The most intense CG barrages I've experienced have been these:

June 12, 2004 - Mulvane, Kansas
June 22, 2016 - Troy Grove, Illinois
August 7, 2018 - Alton, Illinois
July 21, 2017 - Clifton, Illinois

The thing most of those had in common was strong to extreme instability with at least 20 kts of 500mb flow. That being said, I have chased many other events with nearly identical parameters to those that didn't produce prolific lightning.
 
Regarding the ECMWF lightning parameter mentioned above, I've put it to the test in recent weeks over here in Europe, specifically the UK.

The main issue I've found with it is that it is conditional on where the model 'develops' thunderstorms (I know it's not a CAM, so I'm using the term 'develop' very loosely!). Thus, if the model has no convective precip, is has no lightning density - which, overall, is sensible - but it then doesn't give you an indication of the lightning frequency/density 'potential'. I guess one could look at the area where it models precip and then just take that as a broader indication of the density.

Prediction the frequency of lightning is something I do on a regular basis for clients - it's a subjective assessment, overall, with a somewhat subjective risk level indicator. A mixture of numerical products (CAPE), forecast upper ascent profile analysis, pattern recognition, etc, is used.
 

Thomas Mattek

Enthusiast
Apr 2, 2018
3
7
1
Missouri
That's an interesting point about the ECMWF, and I believe the SPC's method is contingent upon the same thing i.e. actually developing convective precip.
 
I think updraft strength in the -10 to -30 zone predicts TOTAL charge generation pretty well. The trickier issue is the type of lighting.

My personal hypothesis is that given the same updraft strength, higher precipitable water tends to favor CGs over IC lighting. In my experience high CAPE with relatively low precipitable water will produce tons of rapid IC lighting with only occasional CGs, while high CAPE COMBINED with higher precipitable water will produce far more frequent CG strikes, but a bit less IC lighting. The caveat is in the mesoscale environment you can sometimes get the same precipitable water effect when there's persistent updrafts in an area were midlevel moisture has accumulated due to extensive adjacent stratiform precipitation. I think for this reason CG activity often peaks later in the evening when updrafts aren't quite as strong but storms are congealing into a more mature MCS with extensive anvil and a growing adjacent stratiform precipitation zone. I've also noticed that storms that are downwind (usually more north/east in the line) will produce the most CGs. These storms aren't as strong, but they're usually surrounded by a much large anvil. Isolated storms that have just developed on the southern/western end of a complex (and thus haven't produced a large anvil yet) often produce a lot of IC lighting but only occasional CGs.

I also think the previous paragraph applies more to the common negative CG strikes. Unlike negative CG barrages which happen post-peak instability when storms are congealing and there's a lot of stratiform precip around, positive CG barrages happen most often during peak instability. Positive CG barrages are harder to predict, but I recall studies showing increased positive CG activity during phases of abnormal charge distribution in tornadic supercells. One hypothesis is that the charge structure changes when hydro-meteors that have already fallen at some distance are ingested back into the mesocyclone and lifted to the -10 to -30 zone a second time. In any case, it's likely a product of specific kind of storm structure, and thus can't be easily predicted based on general sounding parameters.
 
Last edited: