Winter Banter Thread

[Winterweatherlover]

This model just doesn't quit.

 

https://www.pivotalweather.com/model.php?m=gfs&p=prateptype_cat-imp&rh=2023022118&fh=252&r=us_ne&dpdt=&mc=

[donsutherland1]

20 minutes ago, etudiant said:

Is there any model which predicted the mild winter we are enjoying in NYC?

What surprises me is that none of them appear to have adjusted to the reality of a really low snow environment.

The Bastardi 'Hail Mary' March chill forecast is right in line with all the other models, just a bit more extreme in its focus on the model rather than what's actually happening. Qualitatively though they are all in the same boat, at least as far as I can tell. Maybe we need a new paradigm for modeling the seasonal forecasts.

 

I can't fully agree. The modeling did very well on temperatures but poorly on snowfall (which is beyond the skill of contemporary seasonal guidance). They were not in the "same boat" so to speak. Here's why:

The seasonal models did remarkably well in forecasting a warm winter across much of the Northern Hemisphere, including the "non-winter" in Europe (JB had been calling for a severely cold European winter and, expressing political sentiments to go along with it, energy shortages).

The ECMWF seasonal forecast:

The outcomes (December 1-February 19):

Seasonal guidance does poorly when it comes to snowfall, as it snowfall is a function of discrete events (precipitation and temperatures) that are beyond what is possible today. But the temperature idea was well done.

Subseasonal Forecasts:

But even if seasonal forecasts can be terrible when it comes to snowfall (I had expected somewhat above normal snowfall seasonally based on the limited experience with 3rd year La Niñas e.g., 2000-01), one could make skillful forecasts at the subseasonal range. Even as potential appeared from time to time in December, one needed to be cautious about assuming there would be a big snowstorm. As the events drew closer, it was clear that there would be no big snowstorms. The same held true throughout January and February to date.

The subseasonal guidance (weeklies out to 2-3 weeks) and extended range of the EPS/GEFS did very well in telegraphing the persistent warmth in January and February. For example, below are charts from the weeks 3-4 periods on the January 2 ECMWF weeklies. At that time, Joe Bastardi was calling for the "hounds of winter" to "howl" from January 10-30. He posted maps of past severe cold shots to amplify his forecast.

Every AmWx forecaster in the monthly forecast challenge for January saw a warmer than normal January in Boston, New York City, and Washington, DC. The same was true for February. The 11 forecasters who participated for both months had both January and February warmer than normal.

Calling for Extremes:

In general, one should not call for extremes without strong and sufficient evidence. One needs to respect uncertainty at extended ranges and context for such events. The March 1-April 15 call for NYC to have a mean temperature anomaly of 5° below normal and 20" of snowfall is an extreme forecast. The last such occurrence was in 1956. Such occurrences have occurred only during a handful of seasons. All of those seasons were during a notably colder climate regime. That the climate has changed rendering old analogs useless needs to be respected. Great forecasters are those who can adapt to change when their old frameworks and principles lose utility. Rigidity ultimately leads to forecasting failure when big change occurs.

 

[George001]

On 2/20/2023 at 12:19 PM, donsutherland1 said:

In general, as noted on a number of occasions in the past, one should make or accept extreme forecasts without sufficient evidence for such an outcome. Without sufficient evidence, contextual understanding, and respect for uncertainty in the extended range, such forecasts are almost always doomed to fail.

The March 1-April 15 forecast anomaly of 5° below normal for New York City provides the latest example. March has yet to begin, but some of the ensembles now extend into the opening days of March.

For an illustration of where things stand, the EPS forecasts a March 1-6 anomaly of 0.7° below normal. That the EPS is also potentially evolving toward a warmer pattern is set aside. In fact, by 360 hours, the 2/20 0z run shows a 500 mb pattern similar to that of March 9, 2006 that preceded a period of above normal temperatures. At that range, a lot can still change.

But going back to the forecast -0.7° anomaly, if that anomaly verifies, then the remainder of the period March 7-April 15 would need to average 5.6° below normal to achieve a 5° below normal average for March 1-April 15. Put simply, as March approaches, extent of cold needed to achieve the extreme forecast will need to become even more extreme.

In sum, unless overwhelming and credible evidence is presented, extreme forecasts over long timeframes are little more than speculation. There is no skill involved. There is no insight that informs them. The probability of their verifying is very low. If anything, they are noise seeking to grab the attention of social media, not legitimate attempts to inform the public.

The forecast in question will be verified at the end of the timeframe with periodic updates for purposes of illustrating the important points concerning extreme forecasts made in the absence of sufficient and credible evidence.

 

If it’s extreme cold yeah, it seems that now if the pattern is even the slightest bit unfavorable we get extreme warmth. It looks like the acceleration of climate change is a big reason why we are seeing so many extreme warm anomalies but very few extreme cold ones. Even out west has not been as cold as we were warm, and the trough has been parked over there the entire winter. I don’t understand how people can see this shit and then say “Welp there isn’t enough evidence that climate change is real and accelerating, it’s just cyclical”. Yes, weather is cyclical but this extreme warmth goes far beyond cyclical variation or natural change in the climate. This isn’t normal in any way whatsoever, I’m sorry but people who think this is just natural fluctuation or cyclical have their heads stuck up their asses, or are making big $$$ from oil companies. You bring up a good point about relying on old analogs and the issues that come with it due to our climate being warmer than it used to be.

[Brasiluvsnow]

Interesting drive today just north of Liberty ( NY that is ). I drove from Buffalo South and the temps were well above freezing when I began my drive , I dealt with some light rain and the surfaces were wet along the way but a few miles north of Liberty the temp suddenly dropped from 35 to 30  and all that was once wet turned white - even the highway. It was an interesting few miles until I got past Monticello/ Wurtsboro where the temp went back above freezing,,,,heck I even saw some trucks spreading salt a rare site this year

[Rtd208]

This is the "its 10 days away" winter.

[donsutherland1]

The below map was posted on Twitter to argue for an exceptionally cold March. I highlighted the section encompassing Boston, New York City, and Philadelphia. All three cities were shown on the map to have temperature anomalies of 10° or more below normal.

The official CFSv2 site has a very different forecast for March. Notably, it is anything but as extreme as shown on the above chart.

The difference between the charts should be a "red flag" for any serious analysis concerning the upcoming month. Historical data also suggests that the extreme nature of the forecast shown on the first chart would be extremely unlikely. Just one case (1885) from 1874 through 2022 saw March wind up 10° or more below the 1991-2020 normal value.

In sum, some measure of due diligence should be undertaken when it comes to extreme values shown on monthly guidance. One should not accept the numbers at face value. One needs much more evidence, especially given the timeframe involved. Therefore, the extreme forecast shown on the first chart is all but certain to be wrong. Moreover, the big gap between that forecast and that shown on the official CFSv2 site argues that the first map's algorithm may be flawed. The difference between 3/2-4/1 on the first map and 3/1-31 on the official map (one day) is not sufficient to explain the exceptional difference in forecast outcomes.

[GaWx]

1 hour ago, donsutherland1 said:

The below map was posted on Twitter to argue for an exceptionally cold March. I highlighted the section encompassing Boston, New York City, and Philadelphia. All three cities were shown on the map to have temperature anomalies of 10° or more below normal.

The official CFSv2 site has a very different forecast for March. Notably, it is anything but as extreme as shown on the above chart.

The difference between the charts should be a "red flag" for any serious analysis concerning the upcoming month. Historical data also suggests that the extreme nature of the forecast shown on the first chart would be extremely unlikely. Just one case (1885) from 1874 through 2022 saw March wind up 10° or more below the 1991-2020 normal value.

In sum, some measure of due diligence should be undertaken when it comes to extreme values shown on monthly guidance. One should not accept the numbers at face value. One needs much more evidence, especially given the timeframe involved. Therefore, the extreme forecast shown on the first chart is all but certain to be wrong. Moreover, the big gap between that forecast and that shown on the official CFSv2 site argues that the first map's algorithm may be flawed. The difference between 3/2-4/1 on the first map and 3/1-31 on the official map (one day) is not sufficient to explain the exceptional difference in forecast outcomes.

 Don,

  Part of the discrepancy may be due to the first map being based just on today's 6Z run, alone. Although I don't trust the WxBell algorithms used to produce that first map (I believe their extremes are often overdone), I'm guessing it was a cold run. In contrast, it looks like the maps at the bottom are based on an average of runs from 6Z 2/19 through 0Z 2/22. Note the "initial conditions".

 Am I looking at this correctly?

 Regardless, I'm no JB fan due to his mainly cold biases, which I think are partially due to having energy clients. Reading between the lines, he often seems to want natural gas prices to rise.

[donsutherland1]

50 minutes ago, GaWx said:

 Don,

  Part of the discrepancy may be due to the first map being based just on today's 6Z run, alone. Although I don't trust the WxBell algorithms used to produce that first map (I believe their extremes are often overdone), I'm guessing it was a cold run. In contrast, it looks like the maps at the bottom are based on an average of runs from 6Z 2/19 through 0Z 2/22. Note the "initial conditions".

 Am I looking at this correctly?

 Regardless, I'm no JB fan due to mainly cold biases, which I think are partially due to having energy clients. Reading between the lines, he often seems to want natural gas prices to rise.

Good find. I didn't realize that the WeatherBell map was based on a single cycle. That's bad practice.

1. The sample size for the guidance is minimized over long timeframes (monthly period, in this case) increasing the risk of error. Under WeatherBell practice, one is essentially providing a monthly forecast from an operational model. That's worse than choosing the 15-day GFS forecast over a comparable EPS or GEFS forecast. This practice greatly increases the risk of error.

2. Even small changes between runs can lead to big differences in outcomes between cycles, generating a lot of noise from the run-to-run differences. Whipsawing between extremes doesn't build confidence in the forecasts.

NCEP engages in far better practice from its using multiple runs. Sample size is increased and the forecasts are smoothed (essentially transforming the CFSv2 into a larger ensemble). NCEP uses the initial conditions from the most recent 3 days (12 runs) ending at 0z on the day the map is posted. Tropical Tidbits does, as well, but posts the maps 4 times per day.

On account of the two issues cited above, it's difficult to argue that the WeatherBell maps are useful. The run-to-run discontinuity masks what could be a skillful signal, especially when one draws within 1-2 weeks of a forthcoming month. The lack of continuity is particularly damaging when one is trying to forecast at long timeframes where uncertainty is already very high. The WeatherBell practice also generates extremes that are unrealistic. There's no value in a map that shows a scenario that is all but certain to be far off what actually occurs.

It will probably be useful for illustrative purposes to post the actual March outcome against the WeatherBell and NCEP maps.

Finally, there has been a periodic debate on Social Media when users have posted even 10-day operational model maps depicting extreme outcomes. Most of the criticism of that practice--ranging from low probability of occurrence to skewing public perceptions--has come from private and public sector meteorologists. The posted WeatherBell map is an example of the kind practice that the meteorologists have found unhelpful or even harmful.

[Wannabehippie]

I cannot believe that yesterday Mercer county NJ got an EF2 tornado, in f-ing February. Confirmed by NWS. 

[HailMan06]

1 hour ago, Wannabehippie said:

I cannot believe that yesterday Mercer county NJ got an EF2 tornado, in f-ing February. Confirmed by NWS. 

Central/South Jersey the new tornado alley?

[donsutherland1]

Which is right? Note the difference isn't in the use of Fahrenheit vs. Celsius. The difference is in whether the anomalies are warm or cold in parts of the Southeast. The upcoming weeklies for the March 6-13 period will probably offer a hint of where the European guidance is leaning.

Tropical Tidbits:

WeatherBell:

 

 

[SRRTA22]

17 hours ago, Wannabehippie said:

I cannot believe that yesterday Mercer county NJ got an EF2 tornado, in f-ing February. Confirmed by NWS. 

I can't believe that an f2 tornado occured with dews in the 40s, and temps in the 50s. I seriously didn't think that was possible. 

[Nibor]

3 minutes ago, donsutherland1 said:

Which is right? Note the difference isn't in the use of Fahrenheit vs. Celsius. The difference is in whether the anomalies are warm or cold in parts of the Southeast. The upcoming weeklies for the March 6-13 period will probably offer a hint of where the European guidance is leaning.

Tropical Tidbits:

WeatherBell:

 

 

The difference here is criminal. 

[IrishRob17]

25 minutes ago, donsutherland1 said:

Which is right? Note the difference isn't in the use of Fahrenheit vs. Celsius. The difference is in whether the anomalies are warm or cold in parts of the Southeast. The upcoming weeklies for the March 6-13 period will probably offer a hint of where the European guidance is leaning.

Tropical Tidbits:

WeatherBell:

 

 

Interesting. Lets throw another into the mix:

[TWCCraig]

The difference is probably resulting from using different climate averages. Tropical tidbits looks to be using CFSR 1981-2010, Weatherbell is using ECMWF ERA-5 1991-2020. Weathermodels.com appears to be using some 20-year average. The differences could also be resulting from whether it's a instantaneous snapshot of temperature anomalies, or the average of the anomalies over the past 6 to 12 hours

[Rjay]

Weathermodels.com and weatherbell are both run by clowns. 

[donsutherland1]

22 minutes ago, IrishRob17 said:

Interesting. Lets throw another into the mix:

This map is in line with the Tropical Tidbits one when one considers the differences between Fahrenheit and Celsius.

[Stormlover74]

 

 

[forkyfork]

4 hours ago, Rjay said:

Weathermodels.com and weatherbell are both run by clowns. 

they are straight up messing with the algorithms to show colder outcomes. ridiculous 

[GaWx]

6 hours ago, donsutherland1 said:

Which is right? Note the difference isn't in the use of Fahrenheit vs. Celsius. The difference is in whether the anomalies are warm or cold in parts of the Southeast. The upcoming weeklies for the March 6-13 period will probably offer a hint of where the European guidance is leaning.

Tropical Tidbits:

WeatherBell:

 

 

Don,

 I earlier ITT said I was concerned about WxBell anomalies often having too large extremes and thus not trusting their anomaly maps. However, I decided to try to answer your question based on an objective analysis:

 1. I decided to focus on Athens, GA, because you're focusing on the SE, the 360 hour forecast map I posted below shows them at ~55 F at 7 PM EST on 3/9/23, there's a large discrepancy there in the anomalies between WxBell and TT, and there's easy to find history for Athens.

2. WxBell (using 1991-2020 as a base) has them 4 F BN, implying a 59 F norm for 7 PM on March 9th for the period 1991-2020.

3. Tidbits (using 1981-2010 as a base) has them ~1.75 C AN or ~3 F AN, implying a 52 F norm for 7 PM on March 9th for the period 1981-2010.

4. So, I next needed to find Athens' average 7 PM March 9th temperature for both 1991-2020 and 1981-2010. 

5. I found a great site with hourlies for any date going back to 1981. This link is for 3/9/1981. Just click on 7 PM on the top chart and then change the year to get the same for 1982-2020:

https://www.wunderground.com/history/daily/us/ga/athens/KAHN/date/1981-3-9

6. Based on the data from this site, the 1991-2020 avg for 7PM on March 9th at Athens is 57 F. WxBell implied it was up at 59. Thus, WxBell's Athens normal for 7PM that day is 2 F too warm, which means that WxBell's 4 F BN should have instead been 2 F BN.

7. Based on the data from this site, the 1981-2010 avg for 7PM on March 9th at Athens is 56 F. TT implied it was only 52. Thus, TT's Athens normal for 7 PM that day is 4 F too cold, which means that TT's 3 F AN should have instead been 1 F BN or ~0.5 C BN.

8. Conclusion: The 7 F discrepancy between the WxBell and TT maps regarding the 7 PM March 9th normal at Athens (as representation for the SE) was due to a combination of them using different base periods (1 F), WxBell using a normal that is 2 F too warm, and TT using a normal that is 4 F too cold. So, although all three are factors, TT (4 F) was surprisingly the biggest factor rather than WxBell (2 F) or the different base periods (1 F). Without doing this, I would have guessed that WxBell was the biggest factor based on my prior impression about them.

----------------------------

This 0Z 2/23/23 run's 360 hour EPS mean map has Athens at ~55 F:

[GaWx]

6 hours ago, IrishRob17 said:

Interesting. Lets throw another into the mix:

 Piggybacking off of my reply to Don's question that I focused on Athens, GA, for my answer:

 This Weathermodels map has Athens at ~1/2 C BN or ~1 F BN as of 7 PM on March 9th. The TT EPS mean map had Athens at ~55 F then. It says on your map that Weathermodels uses 20 years for climate. Now consider this for Athens using this great site:

https://www.wunderground.com/history/daily/us/ga/athens/KAHN/date/1981-3-9

2001-2020 avg for 7PM: 59 F

2003-2022 avg for 7PM: 59 F

 So, 55 F - 59 F = -4 F. 
 
 Thus, for Athens as a representative for the SE, the anomaly on their map of ~-1 F should have been more like -4 F if we assume they're using either 2001-2020 or 2003-2022 for climo.

[donsutherland1]

16 minutes ago, GaWx said:

 Piggybacking off of my reply to Don's question that I focused on Athens, GA, for my answer:

 This Weathermodels map has Athens at ~1/2 C BN or ~1 F BN as of 7 PM on March 9th. The TT EPS mean map had Athens at ~55 F then. It says on your map that Weathermodels uses 20 years for climate. Now consider this for Athens using this great site:

https://www.wunderground.com/history/daily/us/ga/athens/KAHN/date/1981-3-9

2001-2020 avg for 7PM: 59 F

2003-2022 avg for 7PM: 59 F

 So, 55 F - 59 F = -4 F. 
 
 Thus, for Athens as a representative for the SE, the anomaly on their map of ~-1 F should have been more like -4 F if we assume they're using either 2001-2020 or 2003-2022 for climo.

Thank you, GaWx. Tropical Tidbits needs to update its baseline, in this case. The 1981-2010 baseline is no longer being used. So, at present, the issues that exist with the maps that I'm now aware of are:

1. TT uses an incorrect baseline on its maps where anomalies are shown (its snow algorithm continues to treat sleet and freezing rain as snow in the 10:1 totals)

2. WxBell uses the CFSv2 as an operational model rather than averaging over multiple runs

[LibertyBell]

12 hours ago, TWCCraig said:

The difference is probably resulting from using different climate averages. Tropical tidbits looks to be using CFSR 1981-2010, Weatherbell is using ECMWF ERA-5 1991-2020. Weathermodels.com appears to be using some 20-year average. The differences could also be resulting from whether it's a instantaneous snapshot of temperature anomalies, or the average of the anomalies over the past 6 to 12 hours

why use climate averages at all? this to me is an example of pseudoscience.  who decided that 30 year averages should be used?  that's a subjective decision-- I'd much rather they show the actual temperatures not some proclaimed climate "average"

 

[Snowlover11]

winter of suck continues! Thanks Tony.

[North and West]

When we see coastal California getting snow and NWS San Diego issuing blizzard warnings

https://www.instagram.com/reel/CnF5ZKfPHww/


.

[LibertyBell]

5 minutes ago, North and West said:

When we see coastal California getting snow and NWS San Diego issuing blizzard warnings

https://www.instagram.com/reel/CnF5ZKfPHww/


.

Those mountains are pretty high, there are probably places there that average more snow than we do.

Elevation usually trumps latitude.

[North and West]

Those mountains are pretty high, there are probably places there that average more snow than we do.
Elevation usually trumps latitude.

Don’t you come in here with your logically drawn out thought process and reasonable explanations.


.

[LibertyBell]

3 minutes ago, North and West said:

Don’t you come in here with your logically drawn out thought process and reasonable explanations.


.

It still sucks though,   California gets the best of both worlds.

 

[North and West]

It still sucks though,   California gets the best of both worlds.
 

Don’t forget the raging infernos!




.

[LibertyBell]

Just now, North and West said:

Don’t forget the raging infernos!




.

Yeah but those have temporarily stopped-- one really good thing in this pattern.

They were going to majorly start rationing water if it wasn't for this winter.