When do you start consulting models?

Dec 8, 2003
Kansas City, Missouri
For the amateur mets out there (that means the ones that aren't forced to do it on a daily basis because it's your JOB), I was wondering when you start seriously consulting the computer models in the season. For me, I refuse to look at them anytime of year beyond the spring (seems tedious unless you have the goal of CHASING something). And then I might take a look at a medium to long range forecast once a week ... which then moves into becoming a daily ritual by May. Then there's a little routine it seems like I have to go through every day. Wake up, look at the convective outlooks, look at the models, check out what people are saying about them (anyone who knows more than me will do, which is just about everyone ... but I do pay a little more attention to a few in here). The re-check the scenario each day before a possible event ... then it's time to go through the ritual of looking at RUC and soundings data. And then there's that persistent question of how much stock should someone really place in all the math ... and which models have been more consistent in verifying than the others. My judgment is always messed up somehow, but at least I do learn some new stuff every year. Just wondered if anyone else has a similar routine, or what they do as the season rolls around each year.

Depends on how active the pattern is. On non-active patterns, I will usually check every other day to see if there are any changes in the pattern that are pronounced. ..unless I am bored. ;) When it is obvious an active pattern is developing, I will check once a day usually on the 00z runs. When a storm is imminent, winter or summer type 48 hours or less, I will check every run made on each specific model as discussed below.

Model of choice short term is ETA and I will usually do an ETA/GFS analysis together, keeping in mind model biases and the like. I will use the UK model rarely. As models are simply just a GUIDE and not meant to replace what reality is and go from there OR replace forecasters intuition and experience, I try to mix the models' forecasts with intuition and gamble. If it is obvious the models are waaaay out there, I will throw em out and simply do a forecast based on what historically has happened and rely on the reality and not the models.

Since Im not a Pro Met, I have nothing to lose but my pride ;)
The only time I look at models in the off season is if there is a snow storm coming or something else of interest. I start checking models daily around March. I just glance them over to make sure there is nothing coming up. Once we start getting into the thick of season I will check the same resources several times a day. I usually will look at SPC, the ETA and GFS first thing in the morning before class. Once I get done with class during the day I will take a more in-depth look at the models, I will check SPC at every update, I will look at satellite loops and discussions at NOAA, I read the Vortex100 forecast discussion, and of course see what the good people of Stormtrack are thinking. When there is a potential severe weather event coming up, I am usually glued to the computer for several hours a night looking at the same thing over and over again. It is kind of ridiculous. I just get real anxious for the upcoming chase and forecasting seems to help relieve some of the tension.
Usually in the off-season, I only check the models after I have been reading up on winter weather stuff, etc to try and put some of my newfound knowledge to practice. Which in all likelyhood, is about every-other day because I spend about an hour or so a day reading something new about weather which is usually a paper by someone.

But starting about Feb 25th, I make a daily habit out of checking the models to see what the pattern looks like for the next 1/2 month or so as I've had some of my best chases pre March 20th (granted I've only been chasing two years :)). But when severe weather is expected, I usually check the models a 0z, 6z, 12z, and 18z and refer to all the other AFDs, HWOs, SPC Products, etc. in the area of interest, I don't have much data on the road, so I have to get as much of it as possible before leaving so I can make good decisions off that.
I start paying attention to the convective outlooks about this time of year. I'll be watching models by end of the month.

In season, my daily routine is a quick check just after 7am of the Day 1 and 2 outlooks and the five NWS discussions that affect our viewing area, along with the five HWOs. After I get the kids run around to school, etc, I look at other sources -- what people are saying here, what our meteorologists think, etc. From there, it depends on the Day 1 outlook -- I usually have a deployment plan worked out by mid-morning.

When our area is included in a convective risk, my two biggest uses of the models are at the two ends of the spectrum. I check them to see if I can make out the reasons for the Day3 outlook, and I monitor the RUC and the hourly mesoscale page throughout Day 1.
Being a relative newbie, I don't monitor any forecasts further out than the Day 3 Convective Outlook. It is difficult enough for me to "see" a pattern developing 3 days out, let alone 6 or 7 days out. Since I work from home over the internet during the day, my main interest is following the current day's RUC and the Mesocale Analysis readings. My personal challenge, just for fun, is to try to outguess/anticipate the SPC's current day products by tracking actual vs. forecast surface observations such as dewpoint, along with satellite observation, and soundings on the Storm Machine site. My biggest thrill is doing my own little 6-hour forecast for an area on the fringes, within, or even outside of the official day 1 outlook area, and then seeing a meso discussion issued for the area a few hours later!
I watch the GFS pretty much all year. It's usually just a quick perusal, though. If something catches my eye, I'll investigate further.

Even during chase season, I don't look at model data any more than necessary. I don't see much value in trying to forecast an event 384 hours out. I won't start paying close attention to things until a day or two out - by then, I'll have more model data (and better solutions, I should hope) to work with, and forecast discussions to provide a second opinion.

Then, on the day of the event, I throw all my analysis out the window and drive to the center of the moderate risk. :p
For me, it's all for grins and giggles for forecast hours greater than 150 hours (or so). Between 72-150 hrs, I start to look at a model solution a bit more seriously. At this time, I look for consistency from run-to-run and start looking for trends. By 72 hrs from the forecast event, I begin to take forecast positions of synoptic features at face value (albeit cautiously). By 24 hrs out, I try to identify what mechanisms may cause mesoscale features to develop during the next day (outflow boundaries, etc). The day of, I rely on observations and really don't look at the models too much.

Then, on the day of the event, I throw all my analysis out the window and drive to the center of the moderate risk.

lol - of course I know you're kidding, but had to add that I've attempted this method before (everyone has - fess up) ... unfortunately it only nets a catch once in a great while ... unless you get a big day like 5/29, with supes developing ahead of the oncoming system in sporadic locations, I always want to be as close to ground zero for initiation as possible, which usually means a couple hour drive to the west somewhere for me. Don't know why I'm going off in this direction in the thread, but had to laugh at that because I've done it more than once!

Cool thoughts, Gabe - interesting to me how much emphasis different people place on different things ... one of the BEST things about chasing and forecasting, IMO, is that it is has as many individual approaches as there are individuals doing it!
I look at models constantly because - right now - they're the best shot we have at predicting long-rage patterns or pattern change. I don't refuse to look at models just because of what time of year it is.......that's being closed-minded. The Earth doesn't care what time of year it is - and neither should we when considering severe weather. We saw a tornado last November near OKC.

I look at models any time of year because I chase any time of year.

I typically use models several days out. Rarely do I put much stock in anything beyond a few days, but I will start paying very close attention to anything that shows up no more than 5 days out.

As the time nears, I start comparing the models, how they've changed, and what patterns they're taking on. I also use SPC's analysis to see what it is they're looking at. Since they do all sorts of convective weather, I tend to do my own hand analysis over their graphically circled area to see where the tornado possibilities are. As (hopefully) jokingly mentioned about driving to the center of a MOD risk, I know that they could have a MOD risk for squalls, and without actually going through and deciminating the area to see what conditions are coming together where, I could wind up chasing a MOD risk squall and miss a SLGHT risk supercell.

In the day or two before, pending my travel plans, I'll start focusing my attention on the ETA, as well as keeping an eye on the GFS. In the day of, the RUC becomes a very useful tool.

This will be my second season where I'm doing my own real forecasting. Aside form the models, chatting among other chasers, and sometimes a gut instinct will also play a significant roll. Especially as I continue to grow in my forecasting. All the help, the better!
Anywhere from a few days to over a week before an event, if I'm really bored. For real, I don't start looking until the night before. I don't sit and study these things for days upon days looking for trends, I just take it with a grain, check the real-time obs, and go.
Anywhere from a few days to over a week before an event, if I'm really bored. For real, I don't start looking until the night before. I don't sit and study these things for days upon days looking for trends, I just take it with a grain, check the real-time obs, and go.

That's probably one of the best pieces of advice. You don't really want to focus too much on model output a few days before the event, as it's really just an educated guess. Heck, the models can be clueless up until the event itself. I usually glance at the GFS every now and then, but don't really buy into a particular solution until the day before the event. Then, the day of the event... Your best bet is doing your own analysis via satellite, nexrad, profilers, sfc obs, etc.. The RUC will also help guide you to where the best instability and shear should setup via the 12 HR forecast. If I recall correctly, the RUC actually did a *VERY* good job at predicting the unexpected Roanoke tornado about 3-6 HRS in advance, by showing rather significant directional shear atop 6000J/KG of instability, and high QPF output EXACTLY on top of Roanoke! The ETA (now NAM) proved to be rather useless for that event, as there was almost no directional shear shown, even on the 12Z run - So as Shane said, just take it all with a grain of salt (or whatever he was referring to, :lol:)


*GFS forecast to find a "region" several days in advance
*NAM analysis the day before
*Hand analysis the morning of the chase of significant features (as well as view profilers, satellite, etc.)
*Review RUC model output
*Adjust target if necessary
I usually don't look at models in the really 'Off' season. I will watch the SPC 3 day Outlook to see if they think something is on the horizon. If they show slight or possibly thunderstorm I may take a gander at the models.

During the season I look at them constantly. However I only use longer to medium range models / days to get an idea of a system approaching. I've found you can't trust them more that that. I will start trusting them and using them to fine tune within 24 hours of an even usually and possibly 36 occasionally. They just change too much to bet on them further out IMO.