• Stormtrack's forum runs on Xenforo forum software, which will be undergoing a major update the evening of Wednesday, Feb 28th. The site may be down for a period while that update takes place.

1/19-1/20/07 DISC: TX / OK / KS

NOTE: I split the following 4 posts (this one included) from the 1/19-1/20/07 FCST given the content of the posts. I greatly appreciate Rich's and Jonathan's posts thoughts about this intriguing situation.

Days like this make me glad I don't need to make forecasts for users... The rain-vs-sleet-vs-snow precip type forecast has been extremely difficult, and it's likely that the precip type is currently being dictated by only 1-1.5C temperature departure (on either side of freezing). This is further complicated when there is a deep isothermal layer (resulting in a high correlation of temp at different levels), for which a very small temp change can have a very large affect on sensible weather at the surface. Dan and I spent an hour or so yesterday looking at model output, and we both realized that the areas along and south of I44 were very much "on the fence" for snow (>2" anyway). As it turns out, this is the case, and then some (we had snow NW of I44, which isn't panning out much). You could get a feel of the extreme uncertaintly watching the 10pm newscasts yesterday... The local FOX affiliate had 8-14" in SW OK, for example (which I thought was a bad forecast to begin with, made even worse now), but all had snowfall 4-8" in OKC. It was interesting watching a live report from MSNBC from Lawton...

This is partly why I refuse to spend much time on winter weather forecasting unless I have to. Of the 3 winter storms that have impacted the OKC area this year (Nov 30-Dec 1, Jan 12-Jan 14, and the ongoing one), the "general consensus" precip type has been wrong twice. The forecasts were pretty good for the first storm, but lasn't weekend's storm ("mega-ice storm for OKC" -- sleet instead of frza, which has a huge sensible weather impact) and this storm (rain and mix instead of snow) haven't been so. I should rephrase that to imply that I'm talking about the situation, not the particular forecast (e.g. it's very difficult to forecast precip type when 1-1.5C make all the difference between a significant snowfall and a very cold rain, and it's difficult to account for, I suspect, cloud physics issues such as precip size distribution that may have led to the sleet-vs-frza type last weekend in the OKC area). From past experience, I will say that it certainly helps to have a strong artic intrusion immediately preceding (or during) the precipitation event (which is one significant difference between this storm and the previous 2 this year). Or, perhaps I'll just only forecast winter weather north of I80, where, this time of year, the precip type question doesn't seem to be as much of a question LOL.
Last edited by a moderator:

Your description of the forecast process for this winter storm is a great example of why the uncertainty can't be ignored in forecasting! The local forecasts for OKC did provide a range of potential snowfall (generally 4-8" as you mentioned), but such a forecast makes it seem that the snow is not in question. Here's my subjective break down of the ongoing event for OKC:

1. Measurable precip - very high probability (~95%).
2. Liquid precip amount - ~90% for > 0.25", ~60% > 0.5", ~30% > 1.0".
3. Precip type - ~10% all one precip type, and ~30% for the event to be dominated by either rain/freezing rain, sleet, or snow.

Local climatology suggests that snow is rare in OKC given the track of the ejecting shortwave trough (looks more like a SVR pattern in spring), and the lack of a pronounced cold intrusion across the Plains. Model forecasts were bouncing around quite a bit with precip type in the I-44 corridor for days, which should have reinforced the uncertainty in the forecast. Short range ensembles did nothing to clear up the picture in central OK. Actual observations through yesterday evening continued to cast doubt on an all snow event in central OK.

The most scientifically sound way to make a forecast for this event would have been to multiply the subjective probabilities for each occurrence to arrive at the unconditional probability of snow, sleet, or freezing rain:

Probability of roughly 6" snow (using 10:1 ratio, which is probably to high for this event) - ~.6 x ~.3 = ~.18

So, my back of the envelope calculations yield a roughly 1 in 5 chance of heavy snow accumulations in OKC (not accounting for ranges of snow:liquid ratio and ground temps), and an equal chance for a half inch of cold rain!

Rich T.
Geez... the 00z OUN sounding as about as borderline as is possible... The elevated, isothermal warm layer is about 2.8km thick, and it's almost exactly at 0C / 32F. What's interesting, however, is that we didn't have a little more sleet... Looking at the depth of the sub-freezing air beneath the warm layer, it looks deeper than we saw during last weekend's sleet event. Now, the subfreezing air (from 850mb to the sfc) isn't as cold, so I suspect that's the cause for the predominantly cold rain (vs. sleet). Of course, this just goes to show the difficulty of the forecast, since one can't expect the models to be able to forecast a relatively thick layer of the atmosphere with an error of less than 1C (the difference between rain and snow it looks, in this case).

EDIT: I should mention that many areas in central OK have changed over to light snow (including outside my apt in Edmond).
Yes, I changed over in NW Norman just about as the balloon was ascending.

This event was driven primarily by where UVV was being generated in the column. On the 18Z OUN sounding, the strongest inferred warm advection/UVV appeared to be in the subtle warm layer and not in the layer that contained sufficient ice nuclei. As the afternoon progressed, the radar showed pcpn becoming more showery and the 00Z OUN sounding indeed showed about a hundred j/kg mucape above H7, allowing parcels to be lofted into the minus 20C layer, increasing the probability of ice nucleation. That ice was then able to travel all the way to the ground given that the entire column was <=0C.

You are correct in the assessment of the sfc-based cold layer being deeper today than last week. The coldest temperature in the layer was about minus 4.8 deg C last week vs. minus 4.0 C at 18Z today. That isn't too far off as you mentioned.

There has been much discussion about the processes that contributed to last weeks dominant sleet pcpn-type vs. freezing rain. That case was a fairly significant outlier. Among the drop size distribution and resident time in the minus 4.8 C layer theories, a couple others have surfaced. One is how the ice nuclei interact with supercooled water drops (Pruppacher and Klett 1997, Pitter and Pruppacher 1973). Their research suggests that if ice nuclei (IN) comes in contact with a supercooled water drop (as opposed to the IN suspended within the drop), a warmer activation can be achieved. Furthermore, Roberts and Hallet 1968 found that if ice nuclei have been previously involved in ice crystal formation, the "pre-activation" can skew things to a warmer nucleability. Thus, despite clay IN generally activating less than minus 10 deg C, if it has been used as an IN before, it can activate at temperatures as low as minus 4 deg C.

Whatever the case, it goes to show that there is high uncertainty with winter precipitation type (probabilistic winter equations as RT alluded to), especially in these outlier cases. The parallel to convective weather is anticipating every tornado in a low cape, high shear scenario. There is so much we have to learn.

In retrospect, this type of forecast challenge is often found in other areas of the country. Take, for example, the precipitation type challenge that east coast forecasters face in the coastal metropolis where the population is in the tens of millions. Winter forecasting is fun! :)

Tough forecast

First off, I've been very impressed by the quality of posts on this event, and from some of the people I'm now seeing contribute to ST. Maybe all these rules and regulations is paying off after all...

This was one very tricky forecast as whats been alluded by many on here, the WAA was just a little too much for this system to give us the snow that was originally expected, not to mention the track the system took, If I'm not mistaken, the WRF model seemed to handle the track of the storm pretty well. Being it seems we're getting more folks on here from the NWS and NSSL on here ( a real good thing).. the following is a log of the conditions I experienced which hopefully may aid in the winter precip. research project that is going on. I estimated on the wind speeds and snowfall rates..so, whoever is affliated with the project looking at this will pass this on:

I left my house in Piedmont just after 2pm, a light to moderate rain was falling that would briefly mix with snow, temp. was 33F, ESE wind at 10mph.. I got to Okarche at 2:25 where some sleet started mixing in, temp. 32, E to ESE wind at 10-15mph. I took the back way west from Okarche then at 9mi. west of Okarche went north to Hwy 33 where precip. rapidly transitioned to snow.. conditions worsened substantially going westbound on 33 to Watonga where snow occasionally came down moderately close to 1"/hr. . Road conditions quickly became slush covered. I arrived in Watonga at about 3:05.. temp. was 30F with an east wind at 10mph... snow was falling about a half inch/hour, Within 5 minutes it turned to sleet, then pure rain, I thought that was strange it did that so fast... I continued west on Hwy 270/3 heading toward Seiling at 3:25... literally 3 miles outside of town heading west, precip. turned back to snow... falling close to an inch/hour ontil just before Seiling... some fog was present between Watonga and Seiling... visibilities down to one half mile at times.

Got to Seiling at 4:10pm as the snow was abating and actually by 4:30, the sun briefly poked through as the 1000-1500 ft ceiling broke. After that I went east on some barely traveled country road that took me to Canton Lake where I got some nice pictures and video. Maximum snow depth was around 7" in Seiling and Canton Lake.

Went back through Watonga around 6:30pm as snow picked up again... about 0.5-0.7in/hour rate.. temp. was 31F... S to SSE wind averaging about 15mph. Stopped in Kingfisher then as I went through Okarche around 8:30.. came into a burst of snow that was about 1"/hr.. temp. was 32F even.

Overall, a great day to experience winter... I may be heading up to Michigan's U.P the following week... after that, I'll be breaking out the chase video's as I think spring will be on my mind.
I just watched an unnamed broadcast meteorologist claim that the "warm nose" appeared overnight in central OK and wrecked the snow forecast. Well, we were discussing that potential problem in this forum last night based on the 00z OUN sounding:


The layer around 750 mb was above freezing and pretty moist already, with warmer temperatures upstream. The warm nose remained at 12z in a saturated profile, with apparent wet bulb cooling of the dry layer just above the ground:


The saturated warm nose remained in the special 18z sounding from OUN:


It appears that part of the 850-700 mb layer was just a little too warm (and moist) to allow sufficient evaporational cooling to reach freezing. Also, precip rates may not have been great enough to promote cooling of that layer through melting. Thus, we were stuck with a mix of freezing rain and sleet for much of the event, followed by mostly rain as surface temperatures warmed just above freezing.

Whether anyone likes it or not, such a situation begs for probabilities to express the substantial uncertainty in precip type. Small (barely detectable) differences in temperature structure can make all of the difference in these winter wx forecasts, and you can't expect numerical models to solve that problem. About the best we could do is say moderate-heavy snow was probable in far wrn and nrn OK, *possible* in central OK, and unlikely much south of the I-44 corridor. Unfortunately, much of this uncertainty was glossed over in the effort to hit one out of the park. Folks often forget that guessing pitches and swinging for the fence is a good recipe for a strike out.

Rich T.