Tornado "confirmed" by dual-pol radar signature

[Jeff Duda]

The Lake Charles NWSFO has issued an LSR confirming a tornado based (I'm assuming) solely off of dual-pol radar signatures (i.e., a CC hole).

2101 3 E KURTHWOOD VERNON LA 3134 9311 FT POLK DOPPLER RADAR USING DUAL POLARIZATION CONFIRMED TORNADIC DEBRIS SIGNATURE NEAR KURTHWOOD.

I'm sort of shocked they would just issue the LSR without apparently actually looking for ground truth. Anyone know anything more?

 

[Mark Ellinwood]

I don't think it's that much of a stretch to issue the LSR, especially if they're confident in it. The technology and experience with dual pol has gotten to a point where putting out a LSR for a debris signature could be done reliably.

If the survey reveals no tornado, I will gladly eat my own words and wait for the technology/experience to evolve further.

 

[Bob Hartig]

Today is the first I've heard of anything like it. Dick McGowan has a thread going about this on FB, and I'm surprised to see how much consensus there is that the Correlation Coefficient furnishes conclusive evidence of a tornado even without ground proof. The thinking is that the CC can discern debris, and a tornado is the only thing that can loft debris to the height of a radar scan. That makes sense, though I hate to admit it. I'm already not crazy about the phrase "Doppler-indicated tornado," and something in me just balks at the thought of "Doppler-confirmed tornado," or whatever it will be called.

 

[rdale]

Remember that the LSR is preliminary. I'd be shocked if it passes SPC qualifiers as-is.

 

[Rob H]

Remember that the LSR is preliminary. I'd be shocked if it passes SPC qualifiers as-is.

Not to be snarky, but how is this any different than "LAW ENFORCEMENT SPOTTED A BRIEF TORNADO" other than possibly being more accurate? It's still an LSR, and a human will likely verify in person.

I'm not really shocked by this, because the methodology has a high degree of confidence, but it's kind of weird seeing the first one. In one year, I'm sure it will be old hat to all of us.

 

[Jeff Snyder]

I'm a little hesitant about using any radar data to "confirm" an event or act as "ground truth" even with the knowledge that LSRs are preliminary and may or may not end up in Storm Data. ρhv (a.k.a. CC) is an extremely good discriminator for classifying non-meteorological from meteorological scatterers. The current tornado debris signature consists of a co-location of low rho_hv, high Zh, low Zdr, and a strong Doppler velocity couplet. However, there *are* other reasons for lofted debris other than a tornado (e.g., strong microburst, strong RFD, etc.). All the very low rho_hv tells us is that, in this case, there were non-meteorological scatterers in the radar resolution volume at the given time. If it's around a velocity couplet, one can infer that the non-meteorological scatterers consists of lofted tornado debris, but is that "ground truth"? I'd much rather not rely upon remote sensing data (save for, perhaps, high-resolution satellite data being used to identify tornado tracks in forested and/or remote areas) when it comes to "ground truth", however. As Rob noted, relayed reports from spotters, chasers, law enforcement, and the "general public" may or may not be any good yet such reports are used in LSRs at time, so maybe this isn't really much different.

 

[Bob Hartig]

I find your post reassuring, Jeff, as I know it comes from the research side, and I trust (am I being naive?) that the operational side will take its cues from you guys.

 

[Andy Wehrle]

Jeff, did the images in your original post come from Gibson Ridge? If so, which version? My Level III Version 1.78.3 doesn't appear to display the dual-pol data.

 

[Jeff Snyder]

I find your post reassuring, Jeff, as I know it comes from the research side, and I trust (am I being naive?) that the operational side will take its cues from you guys.

Bob -- I don't know what the NWS policy is regarding this (I suspect that there isn't a policy at this time). I guess it comes down to what constitutes "ground truth" and what an LSR about a tornado is supposed to signify. I think using the TDS as evidence of a tornado is great, and I've seen it mentioned in warning text and severe weather statements. I'm just not sure about using it as confirmation of a tornado. Keep in mind that the low rho_hv and reduced ZDR can persist after a tornado has ended as the lofted debris settles back to the ground. In big tornadoes large distances from the radar, the relatively high beam height may mean that non-meteorological scatterers (e.g., ground debris) likely will be temporally offset from tornadogenesis and dissipation. It's still an extremely useful signature (as is the Zdr column for locating convective updrafts, etc.), regardless. [My experience with dual-pol data largely comes from high-resolution, mobile X-band radars, so it's a little different from comparatively coarse-resolution 88D data]

 

[Jeff Snyder]

Jeff, did the images in your original post come from Gibson Ridge? If so, which version? My Level III Version 1.78.3 doesn't appear to display the dual-pol data.

Andy -- I know you were referring to Jeff D., but yes, the newer versions of GR3 display dual-pol data. If I recall correctly, these products were added in version 2.0 of GR3. It's well, well worth the upgrade.

 

[Bob Hartig]

Keep in mind that the low rho_hv and reduced ZDR can persist after a tornado has ended as the lofted debris settles back to the ground. In big tornadoes large distances from the radar, the relatively high beam height may mean that non-meteorological scatterers (e.g., ground debris) likely will be temporally offset from tornadogenesis and dissipation.

That thought had occurred to me--the first part, anyway, but your comment about distant tornadoes simply adds to it and underscores the potential time disconnect between a tornado's life span versus when debris appears on the radar and how long it persists. Another thought: How does the radar handle debris that gets carried and deposited several miles downshear? Or does such debris get dispersed over distance to the point where it's not a factor?

 

[Skip Talbot]

How does dual pol do at differentiating the size of the debris? If the updraft is pulling in some leaf, twig, or grass debris it seems to me there could be some false positives. On the other hand, weak tornadoes detected on the radar in a forested area that are pulling in light material, may not be apparent at all to damage survey crews.

 

[Mike Smith]

Without going into the pros and cons of using radar to "confirm" tornadoes; remember the half-beamwidth rule. If less than half of the beam is filled then no echo is displayed. So it takes a considerable amount of debris to fill more than half of the diameter of the beam.

 

[Jeff Snyder]

How does dual pol do at differentiating the size of the debris? If the updraft is pulling in some leaf, twig, or grass debris it seems to me there could be some false positives. On the other hand, weak tornadoes detected on the radar in a forested area that are pulling in light material, may not be apparent at all to damage survey crews.

I'm not sure. In my experience, CC (rho_hv) is *extremely* sensitive to size/shape variability, particularly when there is a large variance in dielectric factor, and debris often has such variability. I suppose that smooth, nearly-uniform small debris may not lead to as much of a decrease in rho_hv, but we know that's not what tornado debris typically "is". The Canton Lake tornado from 5/24/11 was sampled by some mobile, polarimetric radars (NOXP and DOWs), and, if I recall those data correctly, the "debris" signature became much less apparent when the tornado moved over the lake compared to what it was on land. In that case, though, unless one considers lake water to be "debris", the transition from land to water resulted in a significant decrease in debris loading and content.

There is a brand new paper paper (it's not in print yet) that addresses some of these issues, however:

Bodine, D., M. Kumjian, R. Palmer, P. Heinselman, and A. Ryzhkov, 2013: Tornado damage estimation using polarimetric radar. Wea. Forecasting, in press.

I haven't had time to read this paper yet, though. If you have access to Early Online Release papers for WAF, however, you should check it out.

 

[Shawn Schuman]

The Canton Lake tornado from 5/24/11 was sampled by some mobile, polarimetric radars (NOXP and DOWs), and, if I recall those data correctly, the "debris" signature became much less apparent when the tornado moved over the lake compared to what it was on land. In that case, though, unless one considers lake water to be "debris", the transition from land to water resulted in a significant decrease in debris loading and content.

I know this is a bit off-topic, but if anyone is interested in reading this paper a manuscript was posted at AMS last year. Interesting read.

https://ams.confex.com/ams/26SLS/webprogram/Paper211754.html