January 18-20, 2023 | Plains/MW/GL/OV Winter Storm

[MesoscaleBanding]

Canadian Ens. mean track is similar to GEFS

 

[MesoscaleBanding]

EPS in the same ballpark as well...

 

[MesoscaleBanding]

Last 6 runs of the GFS 🤡

[Ohiobuckeye45]

Haha yeah this was a snoozer run. DEJA VU

[SNOWBOB11]

Not sure about that GFS run. It looked odd. Will probably bring back the storm on 00z or 06z.

[easton229]

A little OT, but I can’t recall the last time I saw this many cutters in a winter. 

[Pghsnow]

30 minutes ago, easton229 said:

A little OT, but I can’t recall the last time I saw this many cutters in a winter. 

Only 3 more months of cutters...😁

[StLweatherjunkie]

The 18-20 system seems likely to track too far north to benefit the OHV. Probably somewhere between tracks from Wichita to Chicago and Tulsa to Detroit. Perhaps a ~985 mb low near Fort Wayne, IN on Thursday afternoon?

Personally, I think the OHV has a better shot with the potential secondary system on the 21st that would track farther to the southeast 

[MidMichiganWx]

Funny how the GFS has been consistent with a storm and this run its just....gone basically lol

[StLweatherjunkie]

4 minutes ago, MidMichiganWx said:

Funny how the GFS has been consistent with a storm and this run its just....gone basically lol

Not sure who was defending off hour 6/18z runs earlier in the season (I know it wasn't you), but this is a perfect example of why they're a waste of computer resources.

[StormfanaticInd]

Until something definitively changes I'm going with the status quo. No snow for me. 😆

[Ingyball]

56 minutes ago, StLweatherjunkie said:

Not sure who was defending off hour 6/18z runs earlier in the season (I know it wasn't you), but this is a perfect example of why they're a waste of computer resources.

That would be me lol.

I still think they're a good source of data (they still get new observations just not as much as the 0z and 12z runs due to no upper air) so throwing them out completely would be limiting information. There will be a few quirky runs here or there but it's far less than 20 years ago and that's why we have forecasters. Tools don't always work. A forecaster needs to know when to use them or now.

It was actually a 06z run that started an awful dry trend with the pre-Christmas run here. The previous 0z run had been slightly drier but well within the range of previous runs. I thought this would be a quirky run but nope. That dry trend got worse and worse. This was likely due to data getting injected from reconnaissance. This is really where the 06/18z runs become valuable. Thankfully the GFS was wrong but we got pretty lucky with the robust snowsquall. 

[Ohiobuckeye45]

[Ingyball]

Canadian and GFS are back north tonight but at least the Canadian is wet so I'd be happy with that. The GFS can go home.

[SNOWBOB11]

Euro coming in with a weaker more moisture starved system. More similar to the GFS.

[Indygirl]

Looks like any snow fun for central Indiana has been canceled, right? 

I guess waiting for another day’s models is the only thing to do at this point 

[StLweatherjunkie]

8 hours ago, Ingyball said:

That would be me lol.

I still think they're a good source of data (they still get new observations just not as much as the 0z and 12z runs due to no upper air) so throwing them out completely would be limiting information. There will be a few quirky runs here or there but it's far less than 20 years ago and that's why we have forecasters. Tools don't always work. A forecaster needs to know when to use them or now.

It was actually a 06z run that started an awful dry trend with the pre-Christmas run here. The previous 0z run had been slightly drier but well within the range of previous runs. I thought this would be a quirky run but nope. That dry trend got worse and worse. This was likely due to data getting injected from reconnaissance. This is really where the 06/18z runs become valuable. Thankfully the GFS was wrong but we got pretty lucky with the robust snowsquall. 

Yes, all model guidance has improved over the past 20 years, primarily because of a vast increase in computational resources. That doesn't justify misallocation of those resources. 

The last time we had this conversation you posted a article about how off-hour runs aren't that bad. I didn't have the chance to reply, but I remember that file showing 06/18z runs having substantially fewer observations to work with. You are aware of this, but I don't understand why you are still in favor of running a model out to day 16 with those inferior initial conditions. THAT is the misallocation of resources I'm referring to, because errors in initial conditions amplify as they propagate through the model run. 

You mention trends associated with the pre-Christmas blizzard. Were those trends on day 16? Would an ensemble of short/medium range models using the resources currently allocated to long-range GFS runs do a better job of identifying those trends? In other words, would a larger ensemble of CAM guidance (perhaps run hourly like the RAP/HRRR) out to hour 60 be more useful than 6/18z GFS runs at identifying the trend you mentioned for the pre-Christmas blizzard?

I was surprised when ECMWF started doing 6/18z runs, but you'll notice that they don't go out to day 10 like the 00/12z runs do. Maybe they added these off-hour runs because they don't have short/medium range guidance to more quickly identify important trends? From my experience at the NWS, most forecasters don't have the gumption to filter through previous model runs and if the 6/18z runs are populated in the AWIPs database then that's what they're looking at. Assuming the forecaster knows when or how to use them is likely to validate what is commonly said about assumptions ...

At the end of the day, giving forecasters better tools should be NCEPs number #1 goal and we know that long-range 6/18z runs are inferior to 00/12z runs so why do we keep doing them?!?! I very much hope you agree that 06/18z GFS compute could be spent on better and more useful tools. 

Edited January 14, 2023 by StLweatherjunkie

[BuckeyeGal]

9 minutes ago, StLweatherjunkie said:

Yes, all model guidance has improved over the past 20 years, primarily because of a vast increase in computational resources. That doesn't justify misallocation of those resources. 

The last time we had this conversation you posted a article about how off-hour runs aren't that bad. I didn't have the chance to reply, but I remember that file showing 06/18z runs having substantially fewer observations to work with. You are aware of this, but I don't understand why you are still in favor of running a model out to day 16 with those inferior initial conditions. THAT is the misallocation of resources I'm referring to, because errors in initial conditions amplify as they propagate through the model run. 

You mention trends associated with the pre-Christmas blizzard. Were those trends on day 16? Would an ensemble of short/medium range models using the resources currently allocated to long-range GFS runs do a better job of identifying those trends? In other words, would a larger ensemble of CAM guidance (perhaps run hourly like the RAP/HRRR) out to hour 60 be more useful than 6/18z GFS runs at identifying the trend you mentioned for the pre-Christmas blizzard?

I was surprised when ECMWF started doing 6/18z runs, but you'll notice that they don't go out to day 10 like the 00/12z runs do. Maybe they added these off-hour runs because they don't have short/medium range guidance to more quickly identify important trends? From my experience at the NWS, most forecasters don't have the gumption to filter through previous model runs and if the 6/18z runs are populated in the AWIPs database then that's what they're looking at. Assuming the forecaster knows when or how to use them is likely to validate what is commonly said about assumptions ...

At the end of the day, giving forecasters better tools should be NCEPs number #1 goal and we know that long-range 6/18z runs are inferior to 00/12z runs so why do we keep doing them?!?! I very much hope you agree that 06/18z GFS compute could be spent on better and more useful tools. 

And.. breathe! @StLweatherjunkie @Ingyball

Kind reminder that this is a thread to discuss an upcoming weather system (that may or may not make everyone rip out their hair). 🙂 You both rock my socks, and I'd far rather have you use your giant brains to tell me what the heck this storm is going to do instead of (intelligently) arguing about 6z/18z runs. Take that to the Model Info forum and battle it out, por favor!

[BuckeyeGal]

27 minutes ago, Indygirl said:

Looks like any snow fun for central Indiana has been canceled, right? 

I guess waiting for another day’s models is the only thing to do at this point 

I don't think we should cancel anything at the moment. 🙂

[StLweatherjunkie]

6 hours ago, SNOWBOB11 said:

Euro coming in with a weaker more moisture starved system. More similar to the GFS.

Seems like we're in the medium range doldrums now that the system in question is in the middle of the Pacific Ocean. I circled this storm in green on the 00z Thursday analysis:

Since the system was sampled as it tracked across the east coast of Asia on Tuesday/Wednesday it's plausible that older runs were better. EPS means made a substantial jump (from north IN to south IL) with the 12z Thursday run when it moved offshore so take trends since 00z Thursday with a grain of salt. 

Assuming the above analysis isn't completely false then I'd expect a ~996mb low within the green oval at 12z Thursday morning. Posting here so that I can come back on Thursday and double-check.

[StLweatherjunkie]

3 minutes ago, StLweatherjunkie said:

Seems like we're in the medium range doldrums now that the system in question is in the middle of the Pacific Ocean. I circled this storm in green on the 00z Thursday analysis:

Since the system was sampled as it tracked across the east coast of Asia on Tuesday/Wednesday it's plausible that older runs were better. EPS means made a substantial jump (from north IN to south IL) with the 12z Thursday run when it moved offshore so take trends since 00z Thursday with a grain of salt. 

Assuming the above analysis isn't completely false then I'd expect a ~996mb low within the green oval at 12z Thursday morning. Posting here so that I can come back on Thursday and double-check.

Assuming no major changes, this system should be partially sampled by the US upper air network for 12z Monday runs and should be fully sampled by the 12z Tuesday runs. 

[Hiramite]

00Z low position ensembles for hours 126, 138 and 150.

Where all three exit our forum area is pretty darn similar.

[SNOWBOB11]

1 hour ago, StLweatherjunkie said:

Seems like we're in the medium range doldrums now that the system in question is in the middle of the Pacific Ocean. I circled this storm in green on the 00z Thursday analysis:

Since the system was sampled as it tracked across the east coast of Asia on Tuesday/Wednesday it's plausible that older runs were better. EPS means made a substantial jump (from north IN to south IL) with the 12z Thursday run when it moved offshore so take trends since 00z Thursday with a grain of salt. 

Assuming the above analysis isn't completely false then I'd expect a ~996mb low within the green oval at 12z Thursday morning. Posting here so that I can come back on Thursday and double-check.

Good analysis.

[StormfanaticInd]

2 hours ago, StLweatherjunkie said:

Yes, all model guidance has improved over the past 20 years, primarily because of a vast increase in computational resources. That doesn't justify misallocation of those resources. 

The last time we had this conversation you posted a article about how off-hour runs aren't that bad. I didn't have the chance to reply, but I remember that file showing 06/18z runs having substantially fewer observations to work with. You are aware of this, but I don't understand why you are still in favor of running a model out to day 16 with those inferior initial conditions. THAT is the misallocation of resources I'm referring to, because errors in initial conditions amplify as they propagate through the model run. 

You mention trends associated with the pre-Christmas blizzard. Were those trends on day 16? Would an ensemble of short/medium range models using the resources currently allocated to long-range GFS runs do a better job of identifying those trends? In other words, would a larger ensemble of CAM guidance (perhaps run hourly like the RAP/HRRR) out to hour 60 be more useful than 6/18z GFS runs at identifying the trend you mentioned for the pre-Christmas blizzard?

I was surprised when ECMWF started doing 6/18z runs, but you'll notice that they don't go out to day 10 like the 00/12z runs do. Maybe they added these off-hour runs because they don't have short/medium range guidance to more quickly identify important trends? From my experience at the NWS, most forecasters don't have the gumption to filter through previous model runs and if the 6/18z runs are populated in the AWIPs database then that's what they're looking at. Assuming the forecaster knows when or how to use them is likely to validate what is commonly said about assumptions ...

At the end of the day, giving forecasters better tools should be NCEPs number #1 goal and we know that long-range 6/18z runs are inferior to 00/12z runs so why do we keep doing them?!?! I very much hope you agree that 06/18z GFS compute could be spent on better and more useful tools. 

The models have gotten much worse since these "upgrades" started

[StormfanaticInd]

Edited January 14, 2023 by StormfanaticInd