2024-2025 La Nina

[GaWx]

37 minutes ago, bluewave said:

Your colder example is the exception to the rule which will eventually happen from time to time. But most of the time the models adjust warmer as we get closer we get to the forecast period. The colder pattern than originally forecast around the start of December was the result of the models adjusting to the +AAM rise. But at the same time they underestimated the forcing currently in MJO 4 going over +2.00. Stronger forcing in that part of the world accelerates the Pacific Jet and weakens the -EPO. So you had one event strengthening the -EPO and another weakening it. The height falls showing up near Alaska around the 7th with a faster jet than forecast weren’t there previously when the models were underestimating the forcing near the Maritime Continent. This process lead to the quick change in the week 2 forecast to much warmer than previous runs. This is why I was mention earlier back in this thread that the EPS was probably underestimating the Pacific Jet.

 In other words, the models have a cold bias in the E US. They adjust warmer more often than adjusting colder as we get closer. This has been the case for probably 8+ years. The CMC ensemble actually has had the coldest bias at 2 meters of the big 3. I used to subscribe to a company that analyzed biases in detail. That’s how I know this. The W US is a different story. The bias is much closer to neutral with a few cities sometimes actually showing a warm bias due to overdone W ridging/underdone W troughing.

[bluewave]

42 minutes ago, GaWx said:

 In other words, the models have a cold bias in the E US. They adjust warmer more often than adjusting colder as we get closer. This has been the case for probably 8+ years. The CMC ensemble actually has had the coldest bias at 2 meters of the big 3. I used to subscribe to a company that analyzed biases in detail. That’s how I know this. The W US is a different story. The bias is much closer to neutral with a few cities sometimes actually showing a warm bias due to overdone W ridging/underdone W troughing.

The ridge in the East has been one of the most under-forecast features beyond 7 days going back to December 2015 which leads to warmer temperatures than originally forecast. My guess is that this has something to do with the models underestimating long range convection near the Maritime Continent. Since forcing in that region leads to higher 500mb heights and temperatures in the East.

11-15 EPS forecast significantly underestimated Maritime Continent forcing

 

Verification much stronger forcing

 

 

 

 

[mitchnick]

I'm sorry, I  don't believe this model bias talk.  If there was a proven bias, we would be hearing about it from all the Mets as the reason for their blown forecasts. Anyway, the models are run so many times a day, what runs are you going to use and how often do you check? And is this bias on all the models all the time or some of the models some of the time? It's a moving object that never stops.

There's no doubt the models are flawed, but using 1 particular time on  run vs another 1 particular time on a run to prove a point is cherry picking where I come from. Especially if you can take a different time on the same run and show the model has gotten colder.

Frankly, I'm still trying to figure out the point of the exercise. I have my thoughts, of course.

[bluewave]

35 minutes ago, mitchnick said:

I'm sorry, I  don't believe this model bias talk.  If there was a proven bias, we would be hearing about it from all the Mets as the reason for their blown forecasts. Anyway, the models are run so many times a day, what runs are you going to use and how often do you check? And is this bias on all the models all the time or some of the models some of the time? It's a moving object that never stops.

There's no doubt the models are flawed, but using 1 particular time on  run vs another 1 particular time on a run to prove a point is cherry picking where I come from. Especially if you can take a different time on the same run and show the model has gotten colder.

Frankly, I'm still trying to figure out the point of the exercise. I have my thoughts, of course.

If you go back in these various threads including the ones in the NYC Metro forum you will see me positing model biases all the time. This is how I have been able to identity where the potential errors were to the forecasts going back years in these threads. I apply these corrections manually. If you can train the models with AI to adjust for these biases, then my guess is that the forecasts for weeks 2 and 3 would greatly improve. This could also lead to better seasonal forecasts beyond the 1 month range.

[donsutherland1]

2 hours ago, 40/70 Benchmark said:

I was a bit nervous for a spell, but feel better about my December call now.

Your call looks good. The latest ECMWF weeklies and CFSv2, which show persistent warmth for an extended period of time, raise questions about assumptions in some circles e.g., BAMWX, about this winter's seeing a predominantly cold pattern. The seasonal guidance suggested otherwise, even as month-to-month variability is likely.

[GaWx]

38 minutes ago, mitchnick said:

I'm sorry, I  don't believe this model bias talk.  If there was a proven bias, we would be hearing about it from all the Mets as the reason for their blown forecasts. Anyway, the models are run so many times a day, what runs are you going to use and how often do you check? And is this bias on all the models all the time or some of the models some of the time? It's a moving object that never stops.

There's no doubt the models are flawed, but using 1 particular time on  run vs another 1 particular time on a run to prove a point is cherry picking where I come from. Especially if you can take a different time on the same run and show the model has gotten colder.

Frankly, I'm still trying to figure out the point of the exercise. I have my thoughts, of course.

The cold E US bias is very well known because it is so obvious after analysis of numerous verifications and it has been going on for many years. I’m talking based on a compilation of all of the runs from every day for major cities throughout the US done scientifically by pro mets. It has nothing to do with cherry picking. If you still have any doubt, subscribe to Maxar. You and others would learn a lot from them. They have many pro mets on staff. Their forecasts always take these biases into account making them “bc” (bias corrected). They are not cold hypesters like JB nor cold Debby Downers. Maxar makes money off of their reputation for accuracy, not off of hype or down-playing. They forecast with emotion left out. Weenieism or emotional based bias is strongly frowned upon.

 When Chris and I talk about bias, we mean the average discrepancy between forecasted and verified actual.

[donsutherland1]

5 minutes ago, GaWx said:

The cold E US bias is very well known because it is so obvious after analysis of numerous verifications and it has been going on for many years. I’m talking based on a compilation of all of the runs from every day for major cities throughout the US done scientifically by pro mets. It has nothing to do with cherry picking. If you still have any doubt, subscribe to Maxar. You and others would learn a lot from them. They have many pro mets on staff. Their forecasts always take these biases into account making them “bc” (bias corrected). They are not cold hypesters like JB nor cold Debby Downers. Maxar makes money off of their reputation for accuracy, not off of hype.

 When Chris and I talk about bias, we mean the average discrepancy between forecasted and verified actual.

Model bias data can also be found here: https://frontierweather.dtn.com/verification.html

 

[GaWx]

11 minutes ago, donsutherland1 said:

Model bias data can also be found here: https://frontierweather.dtn.com/verification.html

 

Interesting! The DTN bias graphs for GEFS/EPS you linked me for days 6-15 averaged over the last 90 days are surprisingly showing a warm bias in the NE US. I wonder why.

 If I get a chance, I’m going to email my Maxar pro met contact. Maybe he can give me an update on the week 2 biases.

 

[bluewave]

31 minutes ago, GaWx said:

Interesting! The DTN bias graphs for GEFS/EPS you linked me for days 6-15 averaged over the last 90 days are surprisingly showing a warm bias in the NE US. I wonder why.

It was probably due to the ridge axis setting up over the Upper Midwest and Canada. So the models underestimated the temperatures under the strongest ridging. The ridge most other times has been located further east. So this has been where the typical cold bias has been. Unfortunately, DTN only posts the 90 day averages. So we don’t get to see the biases for specific patterns which produce a ridge in the East. Pattern dependent biases is the next level but they may not have the computer tech to pull off such specific data. I have noticed when the ridge sets up over the East this is when the cold biases occur. Shift it back to the upper Midwest and this is where the cold biases occur. But the northeast when averaged out over the last 10 years shows a consistent longer range bias beyond the what the shorter 90 day average shows.

Most recent 10 year 500 mb height anomaly stronger in East which is a reversal from previous 10 year period

 

 

 

 

 

[mitchnick]

44 minutes ago, donsutherland1 said:

Model bias data can also be found here: https://frontierweather.dtn.com/verification.html

 

 

49 minutes ago, GaWx said:

The cold E US bias is very well known because it is so obvious after analysis of numerous verifications and it has been going on for many years. I’m talking based on a compilation of all of the runs from every day for major cities throughout the US done scientifically by pro mets. It has nothing to do with cherry picking. If you still have any doubt, subscribe to Maxar. You and others would learn a lot from them. They have many pro mets on staff. Their forecasts always take these biases into account making them “bc” (bias corrected). They are not cold hypesters like JB nor cold Debby Downers. Maxar makes money off of their reputation for accuracy, not off of hype or down-playing. They forecast with emotion left out. Weenieism or emotional based bias is strongly frowned upon.

 When Chris and I talk about bias, we mean the average discrepancy between forecasted and verified actual.

What I was questioning goes back to my original post about grabbing a forecast frame and comparing it to another later forecast like Bluewave did while disregarding an opposite result on the same run for a different period. Go back to my original post.

I acknowledged the computers are flawed. My use of the word "bias" was in response to Bluewave's cherry pick...yes, cherry pick. 

[bluewave]

1 minute ago, mitchnick said:

 

What I was questioning goes back to my original post about grabbing a forecast frame and comparing it to another later forecast like Bluewave did while disregarding an opposite result on the same run for a different period. Go back to my original post.

I acknowledged the computers are flawed. My use of the word "bias" was in response to Bluewave's cherry pick...yes, cherry pick. 

You missed the broader context of my analysis which is why you incorrectly used the term cherry pick.

[GaWx]

10 minutes ago, bluewave said:

It was probably due to the ridge axis setting up over the Upper Midwest and Canada. So the models underestimated the temperatures under the strongest ridging. The ridge most other times has been located further east. So this has been where the typical cold bias has been. Unfortunately, DTN only posts the 90 day averages. So we don’t get to see the biases for specific patterns which produce a ridge in the East. Pattern dependent biases is the next level but they may not have the computer tech to pull off such specific data. I have noticed when the ridge sets up over the East this is when the cold biases occur. Shift it back to the upper Midwest and this is where the cold biases occur. But the northeast when averaged out over the last 9 years shows a consistent longer range bias beyond the what the shorter 90 day average shows.

 

 

 

Thanks, Chris. Honestly, I’m still surprised to see there being so much EPS warm bias over the E US in the 6-15 over the last 90 days. It is even showing warm bias over the Great Lakes, where the mean ridge axis was:

[bluewave]

17 minutes ago, GaWx said:

Thanks, Chris. Honestly, I’m still surprised to see there being so much EPS warm bias over the E US in the 6-15 over the last 90 days. It is even showing warm bias over the Great Lakes, where the mean ridge axis was:

Yeah, it could be related to the warmest departures setting up just to the west of the main ridge axis near Minnesota and the Dakotas where the cold bias was. It’s also possible the drought enhanced the radiational cooling over the Northeast with so much high pressure and light winds. So the Euro could have missed that. Most other times over the years we had a cold bias on those charts when the ridge was in the East with higher dew points.

 

[mitchnick]

49 minutes ago, bluewave said:

You missed the broader context of my analysis which is why you incorrectly used the term cherry pick.

 I know exactly what it's all about, hence why you cherry picked.

[vegan_edible]

23 minutes ago, mitchnick said:

 I know exactly what it's all about, hence why you cherry picked.

no disrespect man but @bluewavehas a wealth of knowledge to share and i would not say his responses are cherry picked

[FPizz]

21 minutes ago, vegan_edible said:

no disrespect man but @bluewavehas a wealth of knowledge to share and i would not say his responses are cherry picked

You can have a wealth of knowledge while also cherry picking and having a bias.  

[bluewave]

31 minutes ago, FPizz said:

You can have a wealth of knowledge while also cherry picking and having a bias.  

Certain types of patterns have become more common over the last decade. So it’s not cherry picking pointing out the obvious. The bias being discussed involves the models. 

[GaWx]

 I just emailed my Maxar contact to see if he’ll give me an update on the latest ensemble biases based on internal analyses. I hope he responds. He (Brad) really is fantastic with this kind of analysis.

[Stormchaserchuck1]

9 hours ago, GaWx said:

The NG market is now down a whopping 15% vs a week ago. I learned quite awhile back that this market is smart (as @Stormchaserchuck1knows well) and is thus a great barometer of expectations of temperatures in the E US in the decently predictable medium range. Being down 15% is saying a lot as regards how things looked in the medium range one week ago, when near term cold looked like it might last through the first half of Dec.

Wow! I'm impressed that it even seems to lead model changes/trends. I was noticing movements early in the day, then 18z and 0z would adjust. Now the long range is very warm, with a strong SE ridge for mid and possibly late December. 

[Stormchaserchuck1]

Also, a Natural Gas drop from 3.4 to 3.0 is implying a +NAO for January. 

I think the Crude Oil/Gasoline-Natural Gas spread is currently #2 lowest, behind only last year. 

[PhiEaglesfan712]

6 hours ago, Stormchaserchuck1 said:

Also, a Natural Gas drop from 3.4 to 3.0 is implying a +NAO for January. 

I think the Crude Oil/Gasoline-Natural Gas spread is currently #2 lowest, behind only last year. 

Torch January was the easiest thing to predict for this winter.

[bluewave]

It’s no surprise the models have adjusted warmer going forward. The WPAC warm pool along the equator was the warmest for November at just over a +29.05°C. Forcing in these areas leads to a more pronounced Southeast Ridge. 
 

[donsutherland1]

6 hours ago, Stormchaserchuck1 said:

Also, a Natural Gas drop from 3.4 to 3.0 is implying a +NAO for January. 

I think the Crude Oil/Gasoline-Natural Gas spread is currently #2 lowest, behind only last year. 

That's probably a reasonable expectation. The preponderance of the guidance had been showing the following teleconnections states for the winter:

AO/NAO: Predominantly positive with some variability
PDO: Strongly negative to negative
EPO: Generally neutral to positive

An early-season EPO-/AO-/NAO-/PNA+ does not necessarily mean that the seasonal guidance is off. If anything, it strengthens the idea of a more variable winter than the last two.

Until there is stronger evidence, one should not assume that the early-season pattern will likely predominate. There will be more variability, so there could be some additional episodes of such states. However, to get to the kind of averages depicted on the three-month seasonal guidance, one would need to see sustained periods of AO+/NAO+, PNA-, and EPO+.
 

 

[40/70 Benchmark]

7 hours ago, Stormchaserchuck1 said:

Also, a Natural Gas drop from 3.4 to 3.0 is implying a +NAO for January. 

I think the Crude Oil/Gasoline-Natural Gas spread is currently #2 lowest, behind only last year. 

Agree with that, but doesn't necessairly mean an awful month.

[40/70 Benchmark]

12 hours ago, FPizz said:

You can have a wealth of knowledge while also cherry picking and having a bias.  

Isn't the goal to cherry pick the correct solutions, as that as what he has consistently done for the past 3 years or so....would you like him to pick some terds to level off the verification score?

[40/70 Benchmark]

I can tell you that I am pretty confident I am going to beat the guidance this month by a comfortable margin, and the primary reason why is taking what Chris has been saying into consideration, as opposed to dismissing it out of frustration.

[40/70 Benchmark]

1 hour ago, donsutherland1 said:

That's probably a reasonable expectation. The preponderance of the guidance had been showing the following teleconnections states for the winter:

AO/NAO: Predominantly positive with some variability
PDO: Strongly negative to negative
EPO: Generally neutral to positive

An early-season EPO-/AO-/NAO-/PNA+ does not necessarily mean that the seasonal guidance is off. If anything, it strengthens the idea of a more variable winter than the last two.

Until there is stronger evidence, one should not assume that the early-season pattern will likely predominate. There will be more variability, so there could be some additional episodes of such states. However, to get to the kind of averages depicted on the three-month seasonal guidance, one would need to see sustained periods of AO+/NAO+, PNA-, and EPO+.
 

 

Bingo-

[jm1220]

15 hours ago, GaWx said:

The cold E US bias is very well known because it is so obvious after analysis of numerous verifications and it has been going on for many years. I’m talking based on a compilation of all of the runs from every day for major cities throughout the US done scientifically by pro mets. It has nothing to do with cherry picking. If you still have any doubt, subscribe to Maxar. You and others would learn a lot from them. They have many pro mets on staff. Their forecasts always take these biases into account making them “bc” (bias corrected). They are not cold hypesters like JB nor cold Debby Downers. Maxar makes money off of their reputation for accuracy, not off of hype or down-playing. They forecast with emotion left out. Weenieism or emotional based bias is strongly frowned upon.

 When Chris and I talk about bias, we mean the average discrepancy between forecasted and verified actual.

I don’t like what the data and verifications are either for my backyard but it is what it is. The weather doesn’t care. I learn a lot from reading these analyses backed by said data I don’t like. Hopefully one of these days soon the W PAC cools down so our outcomes can change. 

[jm1220]

24 minutes ago, 40/70 Benchmark said:

Isn't the goal to cherry pick the correct solutions, as that as what he has consistently done for the past 3 years or so....would you like him to pick some terds to level off the verification score?

Exactly. Bluewave might not be posting what a lot of us want to read but he’s also been accurate. I’ll take accurate and don’t want to see it vs JB-esque weenie garbage that never verifies. 

[40/70 Benchmark]

18 minutes ago, jm1220 said:

Exactly. Bluewave might not be posting what a lot of us want to read but he’s also been accurate. I’ll take accurate and don’t want to see it vs JB-esque weenie garbage that never verifies. 

Yea, one thing I have learned over the past few years is to remain humble and always open to alternative thoughts...don't be afraid to admit error, incoporate feedback and alternate methodologies. A closed-mind and an expanding skill set are mutually exclusive.