Making your own forecast is important.

The morning of 6/22/10 a coworker showed me the SPC outlook and suggested he may chase a boundary near the ND/SD border. After looking at the obs and model data for a couple hours it was seeming likely this area was not going to see much in the way of "chase-worthy" storms. The subseqent 1630 and 2000z outlooks changed little despite data suggesting some areas could be trimmed back. I suspect this continuity was maintained after the SPC backed off in the ND area just a few days previous on 6/17 where there ended up being a significant tornado outbreak. The northern half of the Moderate Risk area on 6/22 was essentially a bust while the southern half was gold for the patient Iowa chaser. The 12z GFS model from that day offered valuable clues as to where and when a successful chase may occur. An "SPC chaser" may not have this valuable insight. I should note SPC forecasts are not designed for chasers but are a useful situational awareness tool. A chaser with forecast skill can fine tune an SPC forecast to increase probability of success. That said, I recommend making your own forecast first then compare notes with the SPC forecast.

I put together a power point titled "SPC Storm Chasers" outlining the case on 6/22. This case is not ideal to illustrate my point as a patient chaser who went into the center of the 10% tornado threat would have witnessed a couple nice cells and perhaps a tornado in fading light... but, a chaser who was simply seeking a target of opportunity close to home in a large portion of the Moderate Risk area or a big chunk of the 5% tor threat would have come home empty handed.

The following link will direct you to about a 3MB power point presentation.
http://www.plainschase.com/misc/spcchasers.ppt
 
This is an important lesson for any newbie storm chaser. I recall as a fledgling meteorology student on one of my first chases trying to convince my partner that we needed to head to western Illinois for a chase (this is back in 1997). He preferred east iowa. His target was more likely to generate generalized severe weather and corresponded well with SPC, mine was more tornado friendly and aligned with SPC, but not quite as much. We ended up going to Iowa and the storms initiated early, lined out and produced severe weather and no tornadoes.

The western Illinois storms initiated later, stayed discrete and produced a couple, brief ropes. It was a poor day to chase but it taught us some lessons about forecasting on our own and not relying on the SPC to be tailored for storm chasing.
 
Last edited by a moderator:
Good points for sure. I like using the outlooks for two things: making sure I didn't overlook something, and verification that my forecasts aren't completely out of whack. If you're not doing the forecasting yourself and don't understand the weather that day, you won't be able to adjust well - and unless you let luck play a large part in your chasing, you will need to adjust!

While the lesson of not following the SPC risk zones/outlooks is a good one for beginning chasers, I'm not sure I'd recommend ever looking at the GFS for a chase day. It's a long range, low resolution model and the SPC SREF or RUC-derived products would do a much better job.
 
I year ya!

Yeah man I hear ya! SPC forecasters are talented meteorologists but they are not 10x better then the rest like some think. I think that presentation of yours would be very valuable to some of SPC worshiping friends. Very good message my friend.
 
I would add that the SPC has the whole nation to cover, so is looking at the Big Picture. Don't overlook the local NWS forecasters. Their primary concern is much smaller (their CWA).

I've noticed that there are some who seem to do a really good job explaining their thought processes in the Forecast Discussions. Letting them know that you appreciate their words of wisdom might encourage more of that. Just google "forecast discussion" along with the two letter abbreviation for the State (or States) under consideration to find the link.

FDs help me paint a picture of the day so I better know what to look for in the models and Day 1 mesoanalysis.
 
When I first started chasing locally... I chased merely off of SPC outlooks.... I think most chasers that do use the outlooks do so because they may not have very much experience or knowledge... and they know that the SPC are the official guys and issue all the "official" outlooks.... watches... etc
 
The other thing is that the SPC forecasters are making their forecasts for the emergency management community, other WFOs, and the general public. Their forecasts are not targeted towards chasers. It's a risk forecast, not a what you will see forecast.

I'll check out this powerpoint. I might use it for some of my training I'm putting together for other OU Student forecasters.
 
Good powerpoint presentation!

I have been tracking storms for 3 years now, when I first started I would follow the SPC's guidelines like it was the end all, be all forecast. As I have studied and prepared more, I look at the models and then pick my forecast location accordingly. Sometimes I do end up being right with the SPC, other times we are on opposite ends of the spectrum. That is how forecasting is though, everyone has a somewhat different approach and viewpoint to it.
 
While the lesson of not following the SPC risk zones/outlooks is a good one for beginning chasers, I'm not sure I'd recommend ever looking at the GFS for a chase day. It's a long range, low resolution model and the SPC SREF or RUC-derived products would do a much better job.

I'll even add to what Robert said. One of the first things I do on the day of a chase is look to see how various models have been handling what's currently ongoing that morning, and/or what occurred the evening before and overnight. This is where a healthy does of situational awareness is necessary.

While I normally don't spend much time looking at either the GFS or the NAM on the actual day of the chase, I will peruse both models, especially if one or the other has been performing very well. However, it's helpful to review RUC data, observed soundings, mesoanalysis parameters and other high resolution models, including those showing composite reflectivity, to guage what "might" happen on the actual day of the chase. With that being said, forecast models will often handle mesoscale, microscale and storm-scale details very poorly (e.g., evolution of a mesolow); therefore, it's important to closely monitor real-time observational data and mesoanalysis data to see how an event is actually unfolding.

I think this is a point in time where further elaboration on numerical model data is necessary. Models are essentially the use of equations to predict the state of what is essentially a fluid of air parcels at a future point in time; this is hardly an exact science. These equations used by the numerical models to simulate what the atmosphere might look like at a future point in time are not precise, which will sometimes lead to flawed prognostications. Additionally, we do not have such an intricate network of observational data for the models to ingest over parts of the ocean, mountainous regions and in deserts. If the initial state of the atmosphere is not being properly sampled in certain areas, there are likely to be errors in how atmospheric conditions will evolve at a later point in time.

Where are boundaries located and where are they forecast to be later? What could happen to mitigate prognostications from coming to fruition (e.g., ongoing MCS)?

Regarding SPC outlooks, they're only issued every few hours on Day 1, so it's not uncommon to have convection occurring outside of a slight risk area before the new outlook is updated to reflect current trends. The SPC issues those outlooks on the basis of their best guess at how the atmosphere might behave, and no forecaster can ever be expected to be perfect. Also, it's not uncommon to sometimes see warnings being issued before a watch has had time to be coordinated over a given area, which again highlights the importance of situational awareness.

If you screw up, take a look in the mirror and learn something from it; don't direct your despondency towards Norman, OK.
 
Last edited by a moderator:
One model I love to look at is the HRRR. So far, I've found it to be correct much more often than not (i.e. yesterday's convection). While it completely missed the convective mess that was rolling through Michigan, it did accurately portray the thermodynamics post-MCS... and the location and timing of convective initiation later in the day.

In regards to the GFS vs NAM, I like to look at both to get an idea of the uncertainty involved. I also look at the Canadian GEM. If there's big discrepancies, you know there's a greater chance for things to go wrong.

To be honest, I have been using the RUC less often, replacing it with the HRRR. I find the RUC is often WAY too strong with instability (monster CAPE usually caused by erroneously high LL moisture), and usually too strong with dynamics past 6 hours (let alone 18 hours).
 
Robert, thanks for that HRRR link. I've seen several mentions of it here on Stormtrack and was wondering what it was. Always good to have another tool in the forecasting toolbox.
 
One model I love to look at is the HRRR. So far, I've found it to be correct much more often than not (i.e. yesterday's convection). While it completely missed the convective mess that was rolling through Michigan, it did accurately portray the thermodynamics post-MCS... and the location and timing of convective initiation later in the day.

I feel like the HRRR is hit-or-miss sometimes, especially in weakly forced environments. There are also days when it clearly misses convection that is ongoing at the beginning of the run, which makes the rest of the run suspect, in my opinion. If it doesn't grasp what's happening now, how can it forecast what's going to happen? When I use it, I like to go through and watch 3 or 4 consecutive runs of the model to see if it remains consistent or if it's all over the place.

Don't forget about the 4km WRF! I haven't had a chance to compare it for verification this week, but last week it handled convection very well (at least in the high plains region I was paying attention to). If you use this model, keep in mind that the output for the late afternoon/evening isn't available until 07-08Z, and until the graphic is updated, it will still contain the data from the previous day's run.

It's important to remember though that even if you use these high-res models with their simulated reflectivity outputs, you should still only use them as a guideline rather than as a definitive forecast.
 
Thanks for the compliments. The summer season is the slow season at my office as the bread and butter of DoT forecasting is wintry precipitation. This gives me time to look a little more in depth at convective stuff and put together these cases. I opted to produce another one for July 17 of this year where the SPC and I both essentially had the same tornado forecast based off of 00z data. I worked the midnight shift, had a target in mind, slept for perhaps four hours, took five minutes to look at the satellite/radar/mesoanalysis, then took the 100 minute drive south. Had I taken more time to look at more variables and later model guidance when I got up I may have opted simply get more rest in lieu of a chase.

This was a case where stronger than advertised forcing killed the cap too quickly in the target area. Meanwhile, a stronger and more convergent LLJ helped produce an unexpected beauty of an elevated supercell in central Minnesota.

PowerPoint here:
http://www.plainschase.com/misc/spcchasers2.ppt
 
Back
Top