Model Performance and Enjoyment of Chasing

JamesCaruso

Staff member
Joined
Jul 5, 2009
Messages
1,833
Location
Newtown, Pennsylvania
In a recent post in an unrelated thread, I "threatened" to start a new thread about this, so here goes...

Obviously model performance continues to improve year after year, and the HRRR model in particular sometimes predicts individual supercells with startling accuracy. Of course, its success rate at the storm scale is spotty, and we are not yet at the point where we target an area just because the HRRR shows a storm blowing up there.

But WHAT IF the HRRR improves to the point where we could be reasonably certain of a supercell forming at a certain time and place? Suppose it could pinpoint the time and place of tornadogenesis, and the tornado's path length and duration? (I have often imagined an admittedly far-fetched scenario where the authorities would set up a "parade route" for the tornado, with bleachers for viewing, maybe a festival and opening band before the main event )

If we ever get to this level of predictability, how would this impact your enjoyment of chasing? Would you still want to chase?

As for me, much of the challenge and the thrill would be gone... I would still want to see the storm, the same way I enjoy watching the surf be kicked up when a hurricane is offshore, even though that is predicted and publicized well in advance. But the loss of the forecasting uncertainty, the lack of challenge, and the elimination of the possibility of failure would take much of the fun out of storm chasing for me. It just would not be nearly as satisfying if the mystery were solved and the ending of the story is known. (On a side note, realizing that the challenge and possibility of failure are so much a part of what makes storm chasing enjoyable, has helped me learn to deal with chasing failures and frustrations... Or at least I try to remind myself about this!)

Of course the natural second question is, do you think we will ever get to this hypothetical level of predictability? If so, to what degree - for example, able to at least predict the storm but not the tornado? In what timeframe do you think we will get there? (Hopefully not while I am still able to chase!)

Even if you don't think we can ever get there because of chaos theory or whatever, I still hope you will share your thoughts on how you would feel about chasing IF perfect prediction could ever be achieved, or how you might feel with various levels of predictability - for example, maybe you would still go see a predicted supercell if tornado formation was still an uncertain variable, but if that could also be predicted maybe you lose interest entirely?
 
I'm pretty sure there was a discussion a bit like this on here before but it's an interesting on nonetheless.

I think the answer to 2) is no - I don't think we'll get to the stage of simulating exactly where storms will develop and move to - at least not any time soon. However, I think with short-range ensembles we'll get better a highlighting preferred areas, which, in chasing terms, is somewhat tantamount to the same thing. This is because you would be able to get into a better starting position, on average, than currently.
Of course, the main reason I think the answer is no is the fact that storms form for reasons which are not always readily apparent in the network of observations - for example, better heating over a certain type of land-use, or perhaps a very subtle convergence zone. Add into that the fact that the reason some storms form tornadoes and others don't may well be down to very local 'zones' of enhanced SREH, etc, also un-resolvable in observations, and pretty near impossible to simulate in a model, then even if we can simulate the development of the storm, simulating the formation of a tornado (or not) is extremely challenging.
But, in answer to 1), if we *could* do this then I would certainly still get out there and watch.
 
It won't ever happen, but if models could predict exact areas of tornadoes, length, and duration, it would ruin it for me. For me, half of the fun is the forecast. I love to play chess with Mother Nature and try to figure out what might happen next. I learn something new when I bust, and I enjoy it tremendously when I choose a target and it verifies. Again, knowing that a tornado would form at 7:27 in Ellis county Oklahoma on Bob's road would kill it for me.
 
It wouldn't change anything, because I pick targets based on model parameter data, not magical radar crystal balls. If you look at the HRRR every run on a given chase day for 4-6 hours, it's wildly back and forth on the actual placement of storms. Most chasers worth their salt can glean a confident target based on model data (raw parameters, not radar models). I don't even use the supercell parameter until after I've already looked at each individual level's raw wind data.

Models like the HRRR are the chasing equivalent to those "magic pill diets" that are designed for people with money but who are too lazy to exercise and eat right; they want the prize without the effort. Personally, I'd rather succeed or fail based on my own interpretation of real model output, of raw parameter model output....than just clicking the HRRR every hour and driving to where the storm is.

Maybe the only thing about the scenario you describe here is, it would get highly annoying to see everyone showing up late to the place I'd already been at for an hour, and they didn't even put the legwork in.
 
Given the steady improvement of computer technology, it's really only a matter of time before predictions at the storm-scale become regularly fairly accurate. A recent BAMS article demonstrated (or at least, argued in a compelling fashion) that the technology to run a global convective-scale model operationally is already coming around and will probably be available in around 10 years or so. Model grid spacing (for experimental or semi-operational regional models) will certainly be on the sub-kilometer scale within 10-20 years. The bigger issue is whether there will be sufficient gains in observational systems to provide accurate enough atmospheric state estimates to provide good enough forcing datasets for the models of the future. The amount of observations that have been added to the operational data assimilation systems used in today's operational models is incredible compared to that from a few decades ago...orders of magnitude increases there. However, major gaps - temporal, spatial, and category-wise (i.e., some atmopsheric fields like temperature are measured more comprehensively than moisture or wind) - remain to be filled by up-and-coming technology, as well as by technology not yet discovered/invented. Honestly, I'm not sure the observational data will improve to the point where extremely accurate predictions, at least from deterministic models, of individual storm initiation and tracking, as well as internal-storm evolution, will become feasible in our lifetimes. As Paul alluded to, ensembles are the future - I highly suspect you will see more and more ensemble products being made available as time passes. Deterministic forecast graphics will always be around, but you will probably see somewhat less of them as probabilistic forecasting becomes a better way to improve forecast accuracy (and represent uncertainty).

As far as the hypothetical situation posed by the OP and the subsequent question: I don't know for sure. Part of the fun of chasing (for me) is the thrill of the prediction. I take pride in making an accurate forecast of the future atmospheric state, at any spatial and temporal scale, and I get upset when I make particularly bad predictions. So in that regard, the improvement in technology would remove any emotional investment I would have in the activity. The other part of the fun is, of course, seeing the crazy sh-t that mother nature can produce - the rarest of the rare of events. I really love seeing the wonder of nature in action. Therefore, the improvement in technology would probably lead to an overall greater enjoyment from the activity since it would increase my success rate, thus saving me effort, time, and money from fewer busted chases. Of course, it may also lead to what the OP mentioned in "parade-style" chases where people crowd individual roadways minutes or hours in advance knowing there is a very high likelihood of a tornado occurring at that spot at some future time. I hate crowds when chasing. So if it got to that point, I would probably stop chasing to avoid them.

Besides, by that point, we will probably have virtual reality good enough to almost fully replicate being near an actual supercell and tornado, so I could probably pay $250 in future money to spend 10 minutes in a VR machine that would simulate a storm chase.
 
Besides, by that point, we will probably have virtual reality good enough to almost fully replicate being near an actual supercell and tornado, so I could probably pay $250 in future money to spend 10 minutes in a VR machine that would simulate a storm chase.

Hell, if you're gonna settle for pretending, just close your eyes and use your imagination. Easier and free.
 
Thanks Mike, and thanks all for sharing your thoughts about this. Sorry for duplicating an old thread, guess I missed that one.

In the other thread that Mike linked to, there was a post by Bob Schafer in which he said he would still chase in the predictive scenario; I am para-phrasing, but basically because forecasting is just a means to an end and if it was about the challenge then why not get rid of all the technology and go old-school? This is an interesting point; it shows that we all define for ourselves the balance between how much "help" we want vs how much of a "challenge" we want. Most of us use all available technologies to maximize our chances of success, yet at the same time don't want it to get TOO easy...
 
Thanks Mike, and thanks all for sharing your thoughts about this. Sorry for duplicating an old thread, guess I missed that one.

In the other thread that Mike linked to, there was a post by Bob Schafer in which he said he would still chase in the predictive scenario; I am para-phrasing, but basically because forecasting is just a means to an end and if it was about the challenge then why not get rid of all the technology and go old-school? This is an interesting point; it shows that we all define for ourselves the balance between how much "help" we want vs how much of a "challenge" we want. Most of us use all available technologies to maximize our chances of success, yet at the same time don't want it to get TOO easy...

Don't be sorry for duplicating an old thread, it's pretty much impossible to read through everything on this forum and figure what has and has not been discussed before, and this is a very interesting topic. I just wanted to add that link in here because it was similar.

As I was thinking about answering this question again, I could not figure out just how much enjoyment I actually get out of the forecast. I mean, I do use every means necessary to try and get myself into position to see tornadoes. From all of the models, to the SPC, to discussions on here and what other chasers are saying, I will try and use everything at my disposal to see a tornado. I think 90% of why I chase is to actually be there to see the storm do its thing. I think the other 10% is not necessarily the forecasting ability and saying "Wow, I got my target exactly right and the atmosphere did exactly as I thought it would", but rather it is the positioning, navigating, radar interpretation, and everything else that goes into seeing the tornado once storms develop. Plenty of chasers end up getting to the right general area of a tornadic supercell, but it seems that only a small percentage end up getting the really great video, especially in HP storms. I like trying to continually improve upon the actual hunt, navigation, and positioning once you are in range of the storm.

As far as setting up bleachers and selling tickets to tornado events, I really don't think models will ever become accurate enough to pin down the exact location and track of a tornado very far ahead of time. There is just too much that can change in the span of a few minutes with these storms. But, if somehow they do become that accurate, I would be there for as many tornadoes as possible. I enjoy seeing the storm and the process of a developing tornado too much to let crowds, or the thrill of the hunt no longer existing, get in the way of seeing the storm. It does sound like a bad sci-fi movie though. 50 years from now models are pin-point accurate with tornado events and everything is going great. The government is making money selling tickets to see supercells and tornadoes all across the country. Towns are able to prepare and be safe from tornadoes days in advance. Then one spring the storms become self-aware and start veering off course, attacking the viewers and destroying towns no longer prepared for tornadoes because the models said nothing would happen there. It'll be called TORNADOGENESIS: Deadly Dryline Days......... ...... lol
 
As I was thinking about answering this question again, I could not figure out just how much enjoyment I actually get out of the forecast. I mean, I do use every means necessary to try and get myself into position to see tornadoes. From all of the models, to the SPC, to discussions on here and what other chasers are saying, I will try and use everything at my disposal to see a tornado. I think 90% of why I chase is to actually be there to see the storm do its thing. I think the other 10% is not necessarily the forecasting ability and saying "Wow, I got my target exactly right and the atmosphere did exactly as I thought it would", but rather it is the positioning, navigating, radar interpretation, and everything else that goes into seeing the tornado once storms develop. Plenty of chasers end up getting to the right general area of a tornadic supercell, but it seems that only a small percentage end up getting the really great video, especially in HP storms. I like trying to continually improve upon the actual hunt, navigation, and positioning once you are in range of the storm.

Good points Mike. I do truly enjoy the positioning/navigational aspects as well. Forecasting the time and place of a tornadic supercell is one thing, but to model out the entire evolution and path such that navigation is no longer even an issue would be logarithmically less probable and would have to somehow include the ability to impart a modeled visual of rain-wrapping and other variables affecting visibility! So hard to imagine that part of chasing ever going away. (Speaking of which, that's the part I feel I am worst at. Even after nearly two decades of chasing, the cumulative number of times I have been able to practice that in two week (sometimes shorter) chase vacations is not a huge number... And unlike forecasting, the skills are difficult to hone and impossible to practice in the off-season!)

I do like the forecasting too; for me I would probably say it's 80%/10%/10% - but at the same time it's hard to really delineate that way because they are so intertwined. Of course seeing the storm is the ultimate reason we chase to begin with, but the satisfaction level does go up for me if its a good forecast that got me there.

I think the bottom line is that with so many variables both controllable and uncontrollable, it is the CHALLENGE that makes it all so satisfying. The possibility of failure makes success that much sweeter. If that CHALLENGE goes away, the experience becomes much less meaningful for me. It would be like taking a magic pill to achieve some athletic goal, without having to put in the work...



Sent from my iPhone using Tapatalk
 
Completely ignoring the scientific aspect of this hypothetical and treating it as just that, a hypothetical...

I'm in agreement with several others who have estimated roughly half of their enjoyment comes from the uncertainty and challenge in chasing as we know it today. So removing that entirely would have a significant adverse impact, in theory. Of course, I'd also be a lot more successful, which should help to compensate.

The real reason I think this would bring an end to my enjoyment of chasing is the resultant crowds, which I can literally envision rivaling a Taylor Swift concert under the right circumstances. The busiest chase days this past year really started to chip away at my enjoyment, and if that problem were to become a given on every single good tornadic supercell, it would be absolutely crushing. Actually, I imagine that if we're honest with ourselves, this is a big reason why we take such enjoyment in nailing a challenging forecast: because we end up being the heroes of the day with amazing footage to share, at the expense of all the suckers who chose poorly. We've probably all had greatly-enjoyable days where literally everyone else was on the same storm (e.g., Canadian on May 27), but if you've been chasing awhile, your favorite day is bound to be one where that wasn't the case.

Coming back to the real world, though...

I'm very appreciative to be chasing in the era we are. I think the balance between the challenge/thrill vs. our realistic success ratio is just about perfect where we are. I envy the light crowds that veterans enjoyed in the 70s-90s, but if I'm totally honest, I don't think that my personality would have accepted the type of low success rates inherent to that era. The crowds these days are really annoying at times, but ultimately, it's worth it to sit here knowing I'll likely witness at least one quality tornado next spring if I'm able to get out and try regularly. That wasn't the case 20 or 30 years ago.
 
Imagine a scenario 20 years in the future where Tornadoes are predicted with great skill and accuracy, yet small scale motions within storms such as the El Reno monster are still unpredictable. That transition period where thousands of chasers are able to converge on a tornadic cell well in advance and not understanding all the potential hazards would be a crazy time indeed.
 
Another interesting thought on this subject is the point when Humans become a Class I civilization (Controlling weather) Michio Kaku talks about this subject in a couple of his books.
 
I'm very appreciative to be chasing in the era we are. I think the balance between the challenge/thrill vs. our realistic success ratio is just about perfect where we are. I envy the light crowds that veterans enjoyed in the 70s-90s, but if I'm totally honest, I don't think that my personality would have accepted the type of low success rates inherent to that era. The crowds these days are really annoying at times, but ultimately, it's worth it to sit here knowing I'll likely witness at least one quality tornado next spring if I'm able to get out and try regularly. That wasn't the case 20 or 30 years ago.

Brett, I really like the sentiment you captured here and the way you articulated it. I think you are spot on and I agree wholeheartedly. The balance is truly perfect. Relates to my post above, in which I referenced Bob Schafer's post in another thread: we could go completely old-school if we wanted to be purists with maximum challenge, but most of us don't do that, we take the help technology affords us. But neither do we want to see a completely predictable scenario either. You're right, the balance today is just about perfect! Let's enjoy it while we can!
 
Imagine a scenario 20 years in the future where Tornadoes are predicted with great skill and accuracy, yet small scale motions within storms such as the El Reno monster are still unpredictable. That transition period where thousands of chasers are able to converge on a tornadic cell well in advance and not understanding all the potential hazards would be a crazy time indeed.

My fear in that scenario would be that the authorities would fence everybody out and we wouldn't get anywhere near the tornado: an "Exclusion Zone," as Stephen Henry theorizes in his post in the related thread Mike Marz references above.



Sent from my iPad using Tapatalk HD
 
Back
Top