2025-06-17 REPORTS: KS/OK

[gdlewen]

TL;DR. No tornado (yet again). Hopelessly-flawed chase that ended with an interesting encounter with a HP supercell.

Initially, June 17 was not to be a chase day, despite a MDT risk in central KS and very bullish NAM 12Z parameters in S KS (e.g. vicinity of Sedan.) However, most, if not all, CAMS snubbed SE KS, favoring convection further west. It was hard to resist, though, so I made a late decision to drive north.

Along the way an isolated cell formed SW of Wichita. There was also convection associated with a slow-moving cold front in W KS. Anticipating a northeasterly component to storm motion, the plan was to position somewhere north of Winfield, KS. (Far better would have been to approach Winfield from the south.)

While en route to Winfield, the storm SW of Wichita went supercellular and produced a tornado near Harper, KS at 2141Z. We stopped a few miles N of Winfield about 5:40PM to assess the state of things. (At this point we should have dropped S to Winfield and then W to Wellington.)

By 6:00PM, the "Harper Supercell" was over Wellington but was looking less and less like a classic supercell, so I decided there was no sense of urgency. We dropped slowly south while keeping the storm in sight, but never really got into a good position.

What I did not recognize was that this storm was in the process of transitioning to a HP mode....

​

​

KICT 0.5˚ Reflectivity Scan for 2330Z on 6/17/2025. Our location is marked by the black dot. We are just north of the inflow region of the developing HP Supercell. I totally failed to appreciate this until 2330Z.​

By 6:30 PM it was clear that :

1. we were facing a maturing HP supercell, and
2. it was almost certain to cut off our access to US-166 in Arkansas City, which was the best route to stay ahead of it as we headed home.

There were open routes to our E and N, but the optimal route was S to Arkansas City and then E.

Up until 6:30 this storm seemed not so very imposing. Lightning was no longer frequent outside the rain-obscured center of the mesocyclone, which I'll refer to as the "visual mesocyclone" from here on. The front flank precipitation core was descending slowly from the north, driving us slowly southward, but that was the direction we wanted to go anyways.

But it actually was imposing.

Video capture of HP Supercell taken 2330Z on 6/17/2025. Looking WSW down 142nd Road from its intersection with US-77, North of Winfield, KS. This image was selected because lightning has illuminated the mesocyclone.​

One of my goals in chasing is to correlate what I see from the ground with what the data is showing, but in this case I totally missed it.

In an effort to understand happened--how I failed to connect things--I used a video image capture where lightning illuminated the rain-obscured part of the mesocyclone. This permits a rudimentary photogrammetric analysis to estimate the size and location of whatever visible feature can be associated with the circulation (not the actual mesocyclone, of course). In this case, it is the region of the rain-filled mesocyclone that is uniformly illuminated by lightning that we are trying to measure, and we hope that is a reasonable proxy for the actual mesocyclone.

Caveat: this is a minimum estimate of any size because:

1. front flank precipitation partially obscures the northernmost edge, so this analysis is only as good as our estimate of that northern tangent ray, and
2. the entire base of the cell was rotating so we are restricting the numerical analysis to the portion of the actual mesocyclone for which I feel we can make sensible measurements (the proxy)

Estimate of Size and Position of the "Visual Mesocyclone", Based on Measurements of the Video Image. Our location is marked by the black dot.​

The minimum radius is about 6km—since this is an estimate based on a visible feature of the circulation, the mesocyclone itself is undoubtedly larger.

This analysis all seems plausible, but if a StormTrack member can see an error, I would appreciate hearing about it. (The only thing worse than being wrong is never getting the chance to be right.)

One more check on the plausibility of the analysis: a cross-section of the radar through the inflow region:

Cross-Section Geometry: the circle is the photogrammetric estimate of the size and position of the visible feature used as a proxy for the HP mesocyclone; the cross section lies along the dotted line; a solid line joins the camera position and the center of the "visual mesocyclone".	Actual Cross-Section through the HP mesocyclone. The box indicates our estimate of the size and position of the rain-filled portion of the mesocyclone.

The visual record and data appear to be consistent, barring any serious errors. (I'm comfortable with it.)

ANYWAY: Despite being "out of position", we still have good visibility into the inflow region of the developing HP supercell. I would still prefer to be looking NW into it...

---

This may all be fun and games now, but at the time I was dismayed by what can only be described as the chaser version of "chess blindness". I figured we could only beat the storm to Arkansas City if we left immediately and made zero stops along the way. (This was a shame, because if we had moved south sooner we would have had a better view.)

It was close, but we got to US-166 and skedaddled east and got out of there just as the hail and heavy rain arrived. We got hit by a few large hailstones but only suffered minor dents, and then we were in the clear. What followed was a game of leapfrog along US-166 with multiple stops where we could get a good view of the storm—the further east you go the more hilly the terrain gets, but there plenty of good spots to stop along the way.

Near Dexter, KS, we got a chance to take a panorama with sufficient visual landmarks to analyze the visual mesocyclone:

Panorama of Approaching HP Supercell taken with Nikon Z6, and stitched with Hugin. The panorama has a HFOV of 154˚. Looking W from US-166 near Dexter, KS.

Once again, the forward flank rain curtain is partially obscuring the northern edge of the visual mesocyclone, which makes estimation of that edge uncertain. US-166 breaks north here, making an S-curve before heading east again, so we are actually facing W in this panorama.

For the estimate of the mesocyclone dimensions we used the circular arcus-like cloud feature which spanned the visible circumference, therefore (again) limiting the analysis to a visible feature of the actual circulation.

KICT 0.5˚ Reflectivity Scan for 0015Z on 6/18/2025. Our location is marked by the black dot, and the size and position of the visual mesocyclone are indicated by the black circle and the tangent line segments indicate the limited of the visible feature used the analysis.	KICT 0.5˚ Storm-Relative Velocity Scan for 0015Z on 6/18/2025. Our location is marked by the black dot, and the size and position of the visual mesocyclone are indicated by the black circle and the tangent line segments indicate the limited of the visible feature used the analysis.

We are very close to the leading edge of the visual mesocyclone at this point: about 1km, which is consistent with the 0000Z Dexter METAR cloud base report and the angle subtended by the ground-cloud base height in the picture. The estimated radius is ~ 4.5km and the meso is about to cross US-166, after which the HP supercell will be S of us (again).

This is too close, and we did not stay long at this position.

We made a few more stops, but the general configuration of the storm did not change much. When we got to Cedar Vale, it became clear that the storm could beat us to US-75 in Bartlesville, OK if we weren't careful, so gave up the chase. We did beat the storm to Bartlesville by about 5 minutes, which is good because US-75 was flooded by heavy rains as the storm passed through.

SUMMARY: It was kind of fun in the end, and I learned a few things on this trip. First: not to be complacent (crisis is the cure for complacence).

But also I have been trying to keep costs down by using only Basic RadarScope on an iPhone. This causes an odd "Perceptual Claustrophobia" that makes it hard for me to assess what's going on with a storm, especially when I try to show both reflectivity and storm-relative velocity panes at the same time. Others may be successful with only an iPhone but it's hard for me, so I think it's time to move up to GRLevel 2 or 3 on an iPad/tablet/PC.

---

Credits: Py-Art was used for radar analysis, MetPy for chase prep (not posted for the sake of brevity), Hugin for panorama preparation and Gimp for image analysis.

 

[Randy Zipser]

(The only thing worse than being wrong is never getting the chance to be right.)

Great tag line. You are so, so right and wise beyond your years!

I was wondering why a photogrammetric analysis of a thunderstorm-scale mesocyclone would be vital to the advancement of knowledge about rotation on the tornado scale. Your thoughts?

 

[Randy Zipser]

This permits a rudimentary photogrammetric analysis to estimate the size and location of whatever visible feature can be associated with the circulation (not the actual mesocyclone, of course)

The reason I asked the question above has to do with terminology. Terrestrial photogrammetry is a very distinct procedure to process physical motion across an image. An essential precursor to this procedure is to determine correct geographical scaling, which is then input as the initialization step of the photogrammetry data-process sequence. It seems that what you had done in your analysis was to use the rain-obscured part of the mesocyclone as an estimate of the overall mesocyclone circulation size, which, in turn could be scaled using other geographical land markers (e.g., roads, rivers, county lines, etc.) superimposed over the radar scan image. But, that is not photogrammetry, rather, it is image scaling, a part of photogrammetry (which has a different purpose).

Please do not misunderstand, I'm not trying to nit-pick your excellent analysis. I'm trying to determine what your objective is in this presentation. If I understand it correctly, you are doing a post-ground-truth confirmation between what you observed (your photo images) and what concurrent radar data scans show (relative to your location) for Doppler velocity. I'm assuming that there would be no further processing of your photographic imagery beyond that point in your analysis, such as determination of tangential velocities of cloud tags or striations in and above the circular shelf cloud base. Is my understanding correct?

 

[gdlewen]

Thanks for the question, Randy. Without discussion, "storm observing" (to use Roger Edwards' term, I think) is an open-loop process where trial and error is how one advances. Therefore any discussion is a plus.

As for your question, "why a photogrammetric analysis of a thunderstorm-scale mesocyclone would be vital to the advancement of knowledge about rotation on the tornado scale"? I'm more of a weather enthusiast, so I can only give a general answer based on my knowledge.

I think it's important to have a detailed understanding of storm-scale rotation because the tornado is embedded in it, and interacts with it. There are visual details which the camera and one's eyes can capture that radar and instruments cannot. Photogrammetry permits one to take the visual scale, the personal experience, and the instrumental record and put them all together.

For the professionals, I think the same reasons probably hold--to get a better understanding of the totality of what's going on in a tornadic supercell. I remember reading a 2015 paper by Wakimoto et.al. on the 2013 El Reno Tornado. It was fascinating how the Mobile X-band radar and photogrammetry were combined to provide details on the circulation on a scale of meters.

Now--why do I torture myself on storms which "do not produce"?

First: To learn. It's all about gaining knowledge. In addition to the general acquisition of knowledge, specifically: I believe there is a lot to learn from the "null result". There's a paper by Weckworth on drylines that failed to initiate deep convection during one of the IHOP field exeriments. I'm glad that paper was published. Analysis of "why something did not happen" seems to be as important as analysis of "why something did happen". Maybe they are two sides of the same coin.

Because only about 20% of supercells produce tornadoes, there are a lot of "null results". How do you understand the difference if you don't understand the subtrahend (the non-tornadic supercells)?

Second: To improve as a storm observer (chaser). This is just personal to my situation: without detailed post-analysis, it will take longer for me to improve. Others may not need this to advance.

Kinda rambled on there. Did I answer your question?

 

[gdlewen]

If I understand it correctly, you are doing a post-ground-truth confirmation between what you observed (your photo images) and what concurrent radar data scans show (relative to your location) for Doppler velocity. I'm assuming that there would be no further processing of your photographic imagery beyond that point in your analysis, such as determination of tangential velocities of cloud tags or striations in and above the circular shelf cloud base. Is my understanding correct?

Yikes. Yes, I did misunderstand your question. No, I do not intend to measure tangential velocities from the images. At this point I could: with a decent estimate of the distance to the circulation, the video I acquired could be leveraged to estimate the tangential velocities.

It's complicated, though. A simple reduction of "positional changes of cloud elements" is confounded by

1. the combination of translation, and
2. rapid development of scud on the leading edge of the meso due to the very strong inflow winds (e.g. local change.)

These effects cause the lowest part of the circular cloud base to appear to rotate clockwise (counter-rotation). So you see both clockwise and counterclockwise rotation at the same time.

It's a surreal effect! But it means I have to subtract those components from the apparent motion to estimate a "true" tangential velocity.

Ha ha--I can be very much task-driven. I just may have been given a reason to try it....

 

[Randy Zipser]

Kinda rambled on there. Did I answer your question?

Yes, you did answer the question very well, gdlwwen! And, you are exceptionally right about learning to extract meaningful knowledge from the "null cases." One of the foremost questions that the original Tornado Intercept Project was attempting to answer over 50 years ago was to determine why the vast majority of thunderstorms do not produce tornadoes, the exact "flip-side of the coin" from what most, if not all, the NSSL et al. research efforts (and funding) went into. Or, using another analogy, instead of looking at the haystack, we were looking for the needle. Ordinary storms (null case) versus severe storms (special case).

About your other point: the "surreal" differential motions in the cloud structure(s) higher up in the parent storm (while a tornado is on the ground) is something that we never even considered much in those early days of tornado research, since we were concentrating on determination of windspeeds and air motion near the ground, which at that time, was a big unknown. Only later, in the late 1990s and early 2000s, did computer software technology develop sufficiently to allow a three-dimensional view of the entire cloud structure from Doppler radar scanning. However, I still think that terrestrial photogrammetry techniques could be useful using visual imagery of a tornadic storm taken from a distance, say 2-3 miles away. I have not read of any studies using that approach, at least when I last looked into that area of research.