2020-21 Global Model QPF/Snowfall Evaluation For PHL. 2/28-3/1 Event. GFS Says You Will Miss Me When I Am Gone & Comes In 1st; UKMET Last.

[Rainshadow6.6 10,360]

This is one that you can read into it what you want and no one can entirely be wrong.

First the premise of a 1,2 inverted trof events failed.  The second wave went south.  The earlier model runs I believe they all had the two snow outcome.  In fact the European I know had more snow with today and less with Thursday. So it was not until the last two sounding runs that the models gave up on today.  So some of these are two event solutions (not one) that Tom & I could not parse because they bled into each other on the ensembles.

So that being said, being too wet and too snowy is not a surprise outcome.  But if you look at the last too runs, you have a picture of what the models were forecasting for just Thursday.  In that regard on the snow side, the UKMET coming in first is a deserved finish.  It hit the PHL outcome on the head.  As good as the EPS has been is as inconsistent (or just one of the model guys) as the ECMWF has been this season.  A deserved last place for a too snowy an outcome.

On the qpf side the GGEM came in first with the driest solutions.  Just like the GEFS, if you are wetter than the GGEM, you are in trouble.  Since the GFS decided to go with the wetter than wet route with the event(s), it came in last.

Also SLR is a functionality of omega in the DGZ and not how cold we are.  Both ABE (in the cold air) and ACY (in the heavier snow band) had lower ratios than the local PHL area.  This was not an ASOS effect.

 

[Rainshadow6.6 10,360]

On 2/11/2021 at 6:51 PM, tombo82685 said:

euro

2/11 0z: .5 snow .6

2/11 12z: .25

2/12 0z: .12 

2/12 12z: .23 

eps

2/11 0z: .4  snow 1.1

2/11 12z: .24 snow .2

2/12 0z: .12 snow .2

2/12 12z: .29 snow .2

We will include the 12z run today 2/13.

BTW we will start Monday night/Tuesday event with the 12z run today.  I know the models are dancing around with a lighter precursor Monday day (tough for ensembles to parse out, just start including any pcpn after 12z Monday). Funny the GFS no longer has 66 consecutive hours of freezing rain.

[tombo82685 14,181]

euro

2/13 12z: .78 snow .1

2/14 0z: .90

2/14 12z: .92

2/15 0z:  .87

2/15 12z: 1.14

eps

2/13 12z: .74  snow .3

2/14 0z: .96

2/14 12z: .77

2/15 0z:  .83

2/15 12z: .83

[Rainshadow6.6 10,360]

Another event, another rule of thumb that works.  Think about it with the two events this week.  The GEFS was the driest ensemble/model and ended in first. This was not an ASOS tipping bucket glazing with the freezing rain.  The two CoCoRaHs sites in PHL had 0.04". The last place Euro is pictured, but the general pattern with all of the models/ensembles were the same.  This was an instance where pcpn did not lurch to the north or northwest at the last second and as the track of the low (another rule of thumb) was too east of Hatteras.  The "snow" side.  The GFS was the default winner as its poor low level thermal resolutions worked to its advantage in this event.  BTW The Euro was correct with the freezing rain predominating in SE NJ vs sleet depicted by the NAM. 

 

[Rainshadow6.6 10,360]

On 2/13/2021 at 7:23 PM, tombo82685 said:

euro

2/13 12z: .78 snow .1

eps

2/13 12z: .74  snow .3

12z run today (the 15th) will be the last.

[Rainshadow6.6 10,360]

On 2/13/2021 at 7:23 PM, tombo82685 said:

euro

2/13 12z: .78 snow .1

2/14 0z: .90

2/14 12z: .92

eps

2/13 12z: .74  snow .3

2/14 0z: .96

2/14 12z: .77

Yeah 12z run today (15th) last with this one and first with the Thursday event. You guys are really getting superstitious with that one.

[tombo82685 14,181]

euro

2/15 12z: .83 snow 6.7

2/16 0z: .83   snow 6.2

2/16 12z: .75  snow 7.3

2/17 0z:  .87 snow 8.7

2/17 12z: .74 snow 6.9

2/18 0z: .82  snow 6.5

eps

2/15 12z: 1.1 snow 5.3

2/16 0z: .85 snow 5

2/16 12z: .95 snow  6.5

2/17 0z: .87 snow 6.7

2/17 12z: .84 snow 7.2

2/18 0z: .82 snow 6.1

[Rainshadow6.6 10,360]

On 2/13/2021 at 7:23 PM, tombo82685 said:

euro

2/13 12z: .78 snow .1

2/14 0z: .90

2/14 12z: .92

2/15 0z:  .87

2/15 12z: 1.14

eps

2/13 12z: .74  snow .3

2/14 0z: .96

2/14 12z: .77

2/15 0z:  .83

2/15 12z: .83

The GEFS was drier than everyone with this event.

[Rainshadow6.6 10,360]

19 hours ago, tombo82685 said:

euro

2/15 12z: .83 snow 6.7

eps

2/15 12z: 1.1 snow 5.3

We might have to stop at the 00z run on the 18th with this one with the way the snow is expected to fly in.

[Rainshadow6.6 10,360]

Pinch me.  Now I have seen everything.  What if you have a precipitation event and the perpetually too wet GEFS is the driest?  Well this is one case where the actual precipitation amount was wetter than the GEFS.  Although, the GEFS had the second most accurate average.

Which brings us to the other dichotomy since the SSW flexed its muscles.  There is the OP EC which is bronze.  Something always goes wrong with bronze.  Then there is copper (The EPS), which is the only model I use.  It causes you pain (because it does not snow like the NAM), but it saves you pain, because the truth is close.

Enough of worming Moonstruck into the discussion, but as much as the Euro has acted like a snowbird wanting to go to Florida with every forecast track (How about that ice!!!!  Nevermind.), the EPS has been fairly steady.  Give it another first place finish.  The UKMET is proving this cold season just because you score well at 500mb doesn't always translate to the surface and its atrocious wet bias with this system earned it last place.

This was another not exactly clear cut scenario as a low pressure system slid to the east of PHL on Monday day.

 

 

 

[Rainshadow6.6 10,360]

On 2/15/2021 at 8:57 PM, tombo82685 said:

euro

2/15 12z: .83 snow 6.7

2/16 0z: .83   snow 6.2

2/16 12z: .75  snow 7.3

2/17 0z:  .87 snow 8.7

2/17 12z: .74 snow 6.9

2/18 0z: .82  snow 6.5

eps

2/15 12z: 1.1 snow 5.3

2/16 0z: .85 snow 5

2/16 12z: .95 snow  6.5

2/17 0z: .87 snow 6.7

2/17 12z: .84 snow 7.2

2/18 0z: .82 snow 6.1

I have to wait for today to be over before doing this event.  But events wait for no one and can we start Monday's with the 12z run today?  (19th).

[snowlurker 1,044]

Speaking of model performance

 

[Rainshadow6.6 10,360]

2 hours ago, snowlurker said:

Speaking of model performance

 

Lol.

[tombo82685 14,181]

euro

2/19 12z: .44  snow  .5

2/20 0z: .42  snow .8

2/20 12z: .42 snow 1.3

2/21 0z: .42 snow  1.1

2/21 12z: .52  snow 1

2/22 0z: .45  snow 1.2

2/22 12z: .53 snow 2.9

eps

2/19 12z: .38  snow 1.1 

2/20 0z: .4  snow 1

2/20 12z: .4 snow 1.4

2/21 0z: .43 snow 2.1

2/21 12z: .48 snow 1.5

2/22 0z: .43  snow 1.8

2/22 12z: .51  snow 1,6

[Rainshadow6.6 10,360]

Maybe there will be a snow event that we won't have to say, but there was banding nearby. Assessment is in the eye of the beholder.  Is it misplaced banding?  Is it forecasting banding to not be at PHL even though it was not put in the right place? Is it snowfall algorithms and not the model itself? 

This is another tough one.  As far as what fell from the sky in terms of water equivalent, the OP ECMWF by being the driest was the most accurate.  35 out of 36 model/ensemble forecasts for this event were too wet.  What about the banding area? They would have been close there, but still too wet on the other sides which was about 90% of the area. The still too wet for its own good GEFS came in last place.  The NAM might have given it a run.

Which brings us to snow.  So somehow the driest ECMWF was the snowiest (and most inaccurate) model/ensemble for this event at PHL.  I did see runs where its axis of heaviest snow cut thru PHL.  Is this a ptype issue? A snowy bias? An algorithm bias? The (most accurate) GFS was less snowy in spite of being wetter and the NAM for all its foibles would have been the most accurate model for this event at PHL.  It hit the warm layer aloft hard and was right (for there)?

We may be back at this point (until the next "upgrade") where some sort of EE rule can be used.  Use the EC/EPS qpf but use the NAM microphysics to adjust the snowfall?

BTW you are going to see more of this next season.  The V16 GFS would have come in last by being too snowy.

 

 

 

 

[Rainshadow6.6 10,360]

On 2/20/2021 at 9:36 AM, tombo82685 said:

euro

2/19 12z: .44  snow  .5

2/20 0z: .42  snow .8

2/20 12z: .42 snow 1.6

2/21 0z: .42 snow  1.6

eps

2/19 12z: .38  snow 1.1 

2/20 0z: .4  snow 1

2/20 12z: .4 snow 1.4

2/21 0z: .43 snow 2.1

We can include the 12z run tomorrow, the 22nd.

[Rainshadow6.6 10,360]

@susqushawn, first I was dead wrong about PHL getting 0.25" - 0.33" from this event.  Other than the Canadian models, the other models were good, if not a bit too dry.  A sin of omission is still a sin. There are two instances I have noticed where pcpn overperforms the models in synoptic scale set-ups.  One is with deep neg tilt trofs. The other when a low pressure system physically moves thru the area.  The models either omitted or were too slow to show this true secondary and I just accepted that solution. 

Onto the event itself and modeling performance.  We had another use the OP ECMWF for the qpf, but not for the snow.  (If you notice the next version of GFS numbers with this, it may be a similar mantra.  It's QPF was good for PHL, it's snowfall would have landed it in next to last place).  The CAN GGEM which was inconsistent and extreme from the get go landed in last place with both qpf & snowfall.  The GEFS which was pretty conservative with snowfall for PHL (poor resolution may have advantages in marginal snowy situations). The NAM which typically does better when lows stay on land than once they hit or form on the Atlantic didn't do that bad either.  It was slow like all the other models to recognize the secondary, right for the wrong reason?

 

[Rainshadow6.6 10,360]

On 2/20/2021 at 9:36 AM, tombo82685 said:

euro

2/19 12z: .44  snow  .5

2/20 0z: .42  snow .8

2/20 12z: .42 snow 1.3

2/21 0z: .42 snow  1.1

2/21 12z: .52  snow 1

2/22 0z: .45  snow 1.2

2/22 12z: .53 snow 2.9

eps

2/19 12z: .38  snow 1.1 

2/20 0z: .4  snow 1

2/20 12z: .4 snow 1.4

2/21 0z: .43 snow 2.1

2/21 12z: .48 snow 1.5

2/22 0z: .43  snow 1.8

2/22 12z: .51  snow 1,6

We can start the on again/ off again / on again Saturday event with the 00z run this morning (24th).

[tombo82685 14,181]

euro

2/24 0z: .33

2/24 12z: .54

2/25 0z: .4

2/25 12z: .37

2/26 0z: .43

2/26 12z: .52 snow .1

2/27 0z: .33 snow .6

 

eps

2/24 0z: .41

2/24 12z: .39 snow .1

2/25 0z: .38

2/25 12z: .32

2/26 0z:  .32 

2/26 12z: .45 snow .2

2/27 0z: .33 snow .6

 

Sunday

euro

2/25 12z: .68

2/26 0z: .92

2/26 12z: .9

2/27 0z:  .8

2/27 12z: .87

2/28 0z: 1.19

eps

2/25 12z: .57

2/26 0z: .76

2/26 12z: .75

2/27 0z: .78

2/27 12z: .93

2/28 0z: 1.03

[Rainshadow6.6 10,360]

15 hours ago, tombo82685 said:

euro

2/24 0z: .33

2/24 12z: .54

eps

2/24 0z: .41

2/24 12z: .39 snow .1

I know you love double events simultaneously.  We can start Sunday-Monday with the 12z run today (25th).

[Rainshadow6.6 10,360]

This is a brief summary of events in which it snowed in Philadelphia.  There were other events where the models/ensembles forecast some snow (nothing major), and none fell at PHL.  I will deal with that at the end of the season.

Yes I know we can speculate what if all of those bands affected PHL (12/19 they got one) vs going around them.  In reality in the bigger picture of the entire Mount Holly CWA, those bands affected a minor percentage of the entire forecast area.  So PHL for this season was a closer proxy of the bigger picture than the luckier less. Yes PHL down at sea level and along a river is naturally going to have lower snowfall.

That being said, you avoid banding and the ensembles (consensus) will do better.  Individual models if they are misplacing the banding or not getting the ptype changes closer will suffer.  The pattern for this season has been the GGEM is not very accurate because its qpf are also not as accurate as other models.

Eh I should have removed the "also" from the most accurate qpf statement.

 

 

[Rainshadow6.6 10,360]

On 2/24/2021 at 6:16 PM, tombo82685 said:

euro

2/24 0z: .33

2/24 12z: .54

2/25 0z: .4

2/25 12z: .37

2/26 0z: .43

2/27 12z: .52 snow .1

eps

2/24 0z: .41

2/24 12z: .39 snow .1

2/25 0z: .38

2/25 12z: .32

2/26 0z:  .32 

2/27 12z: .45 snow .2

 

Sunday

euro

2/25 12z: .68

2/26 0z: .92

2/27 12z: .9

eps

2/25 12z: .57

2/26 0z: .76

2/27 12z: .75

 

We can go thru the 00z run this evening (00z/27th) with the first event. Thanks!

[Rainshadow6.6 10,360]

@susqushawn, another example of a low forecasting thru the area and modeling qpf being more too dry than too wet.

The EPS had quite the consistent and accurate qpf for PHL with this one and came in last.  Ironically its snow forecast came in last. The too dry GFS came in last and being thermally challenged in a marginal snow situation does not hurt the GFS as far as PHL goes. It had no snow predicted on any runs.

A couple of modeling trends that are holding this cold season.  The ECMWF does not add value to snowfall forecasts with marginal snow scenarios.  Yes we can always say it is snow falling vs accumulating, but it has had more around the edges snowy bias than the other three globals. If the too wet twins (GGEM, UKMET), chances are likely it is not going to happen. Even with this event where models bias is to be too dry.

 

 

[Rainshadow6.6 10,360]

On 2/24/2021 at 6:16 PM, tombo82685 said:

 

 

Sunday

euro

2/25 12z: .68

2/26 0z: .92

2/26 12z: .9

2/27 0z:  .8

eps

2/25 12z: .57

2/26 0z: .76

2/26 12z: .75

2/27 0z: .78

 

For the Sunday starting event given it almost starts directly after 7 am on the GFS let's make tonight's 00z/28th the last one for this one.  I love my mud wet.  The wetter the better.

[Rainshadow6.6 10,360]

On 2/24/2021 at 6:16 PM, tombo82685 said:

euro

2/24 0z: .33

2/24 12z: .54

2/25 0z: .4

2/25 12z: .37

2/26 0z: .43

2/26 12z: .52 snow .1

2/27 0z: .33 snow .6

 

eps

2/24 0z: .41

2/24 12z: .39 snow .1

2/25 0z: .38

2/25 12z: .32

2/26 0z:  .32 

2/26 12z: .45 snow .2

2/27 0z: .33 snow .6

 

Sunday

euro

2/25 12z: .68

2/26 0z: .92

2/26 12z: .9

2/27 0z:  .8

2/27 12z: .87

2/28 0z: 1.19

eps

2/25 12z: .57

2/26 0z: .76

2/26 12z: .75

2/27 0z: .78

2/27 12z: .93

2/28 0z: 1.03

0.96" already with wave I, wetter will be better with this one.