2025-2026 ENSO

[LibertyBell]

2 hours ago, mitchnick said:

Not the case for BWI at all those years unfortunately. 93/94 was the worst. Horrendous. 

I remember there was a sharp cut off near Philly, I would no longer have any hair left if I had to go through that south of there.

 

[mitchnick]

49 minutes ago, LibertyBell said:

I remember there was a sharp cut off near Philly, I would no longer have any hair left if I had to through that south of there.

 

I'm away from the coast and 50 miles north of my former BWI location, and they did much better in these parts in 93/94. Unfortunately, the change in location has not replaced the hair loss! 

[LibertyBell]

16 minutes ago, mitchnick said:

I'm away from the coast and 50 miles north of my former BWI location, and they did much better in these parts in 93/94. Unfortunately, the change in location has not replaced the hair loss! 

You are not alone in this affliction my friend =\

[bluewave]

22 hours ago, mitchnick said:

I agree 100% models are too snowy. I've been an unapologetic snow lover all my life and started following snow threats religiously since the 72/73 winter. Since then when modeling was first being used as a tool, they have ALWAYS been too snowy beyond 24 hours. I can remember complaining to a friend in high school (before 1976) that the model forecasts were too snowy. My point is, modeling has always been too snowy outside of 24 hours for a variety of reasons and to use recent years' snowy +24 hours modeling errors as proof of anything is a mistake imho. 

It’s all about the way model biases have been evolving over the years. We went many years near the I-95 corridor when the long range models weren’t snowy enough beyond a few days. Which is the complete opposite of the model error pattern since 2018-2019 with the record breaking Pacific Jet. 

The January 1978 KU  snowstorm was originally forecast to be a rainstorm the day before and we got a surprise 15” snowstorm on Long Island. Then in February 1983 the original forecast a few days out was for the heaviest snows to stay down near Philly. But instead the heaviest snows shifted north into NYC Metro as the models got under 60 hours.

The January 1987 KU event was forecast to be snow quickly changing to rain but instead we got a surprise 6-12”. The December 1988 surprise Norlun wasn't even forecast at all and we got a narrow 6-12” band.

The January 1996 blizzard was originally forecast to get suppressed to the south of NYC from 120 hrs out. But came north in the models only 24 hrs out. The late January 2000 snowstorm was forecast to be OTS and delivered record snows in the Carolinas and over 6” around NYC. 

The record January 2016 snowstorm was only forecast to be 6-12” the day before around NYC and turned out to be 20-30”. 

So model errors in good snowfall seasons tended to be underdone beyond 24 to 48 hours out. 

What the models have been doing since 18-19 is showing heavy snows beyond 72 to 120 hrs only to correct further north with the heavy snowfall axis. So the error has been for the storm tracks to verify warmer and further north than originally forecast. 

This model error pattern worked for us with colder storm tracks to our Southeast which dominated before 18-19. Since the long range forecasts could still afford to shift north changing suppression to a KU. These days the storm track is already too far north so any shift north under 60 hrs gives us to more rain. 

So cutters, huggers, and suppressed Southern Stream storm tracks have lead to the record low snowfall from Philly to Boston last 7 years. The big snowstorm has always been beyond 120 hrs out. So this has become the dominant model forecast storm track bias.

The one good thing about repeating model biases is that I use them in real time to correct the long range model forecast biases leading to better forecasts than the models.

 

[Terpeast]

1 hour ago, bluewave said:

It’s all about the way model biases have been evolving over the years. We went many years near the I-95 corridor when the long range models weren’t snowy enough beyond a few days. Which is the complete opposite of the model error pattern since 2018-2019 with the record breaking Pacific Jet. 

The January 1978 KU  snowstorm was originally forecast to be a rainstorm the day before and we got a surprise 15” snowstorm on Long Island. Then in February 1983 the original forecast a few days out was for the heaviest snows to stay down near Philly. But instead the heaviest snows shifted north into NYC Metro as the models got under 60 hours.

The January 1987 KU event was forecast to be snow quickly changing to rain but instead we got a surprise 6-12”. The December 1988 surprise Norlun wasn't even forecast at all and we got a narrow 6-12” band.

The January 1996 blizzard was originally forecast to get suppressed to the south of NYC from 120 hrs out. But came north in the models only 24 hrs out. The late January 2000 snowstorm was forecast to be OTS and delivered record snows in the Carolinas and over 6” around NYC. 

The record January 2016 snowstorm was only forecast to be 6-12” the day before around NYC and turned out to be 20-30”. 

So model errors in good snowfall seasons tended to be underdone beyond 24 to 48 hours out. 

What the models have been doing since 18-19 is showing heavy snows beyond 72 to 120 hrs only to correct further north with the heavy snowfall axis. So the error has been for the storm tracks to verify warmer and further north than originally forecast. 

This model error pattern worked for us with colder storm tracks to our Southeast which dominated before 18-19. Since the long range forecasts could still afford to shift north changing suppression to a KU. These days the storm track is already too far north so any shift north under 60 hrs gives us to more rain. 

So cutters, huggers, and suppressed Southern Stream storm tracks have lead to the record low snowfall from Philly to Boston last 7 years. The big snowstorm has always been beyond 120 hrs out. So this has become the dominant model forecast storm track bias.

The one good thing about repeating model biases is that I use them in real time to correct the long range model forecast biases leading to better forecasts than the models.

 

If you’re talking about the biggest KU storms, I tend to agree. But there have been instances in recent winters where sub-KU storms have actually trended south in the last 24-72 hours.

In the 2023-24 winter, my subforum (at least the central part) got 4-6” flush hits twice in the same week that trended to us the last minute after being modeled to go through PA/NY. 

This past winter, we’ve had a few happen the same way. Including one in late Feb that missed us and hit SE VA with a foot of snow, leaving us high and dry. 

So I guess it can cut both ways.

[snowman19]

@bluewave Getting back to your point of repeating global weather patterns, if the summer/fall plays out this way, besides the continuation of the +AMO, we may see another *possible* central Atlantic “Nino” develop along with a “Modoki” La Niña, which have both become much more common over the last 10+ years:

[michsnowfreak]

20 hours ago, LibertyBell said:

I wanted to share something, not sure how accurate this is, but I found this while looking up big heat

 

https://www.reddit.com/r/newyorkcity/comments/1epkptr/central_park_ny_highest_maximum_temperature_f/

That 106° reading was on 07/09/1936 during the incredible 1936 heatwave that effected most of the US from the Midwest to the east coast. It was the highest temperature ever recorded in NYC. In those days the official weather bureau temperature was measured at the old Battery Maritime Building near Battery Park where it was always a little cooler due to it's proximity to the water. So the actual temperature in midtown may have been more like 112° if not higher.

My mother remembered that heat wave and said that people left their tenement windows open during the day even if they were at work. A very different city then.

This is the deadliest heatwave on record in the U.S. Detroit saw temps over 100° for 7 straight days (peaking at 104 on day 1 and 7). The city allowed people to sleep in parks since their homes/apartments were baking. The MI state record was set in Mio with a temp of 112°.

[LibertyBell]

1 hour ago, michsnowfreak said:

This is the deadliest heatwave on record in the U.S. Detroit saw temps over 100° for 7 straight days (peaking at 104 on day 1 and 7). The city allowed people to sleep in parks since their homes/apartments were baking. The MI state record was set in Mio with a temp of 112°.

omg I've never experienced anything like that, in 1993 we had 3 straight days of 100+ which peaked at 102 on the third day and that was my best heatwave ever.

[LibertyBell]

1 hour ago, Terpeast said:

If you’re talking about the biggest KU storms, I tend to agree. But there have been instances in recent winters where sub-KU storms have actually trended south in the last 24-72 hours.

In the 2023-24 winter, my subforum (at least the central part) got 4-6” flush hits twice in the same week that trended to us the last minute after being modeled to go through PA/NY. 

This past winter, we’ve had a few happen the same way. Including one in late Feb that missed us and hit SE VA with a foot of snow, leaving us high and dry. 

So I guess it can cut both ways.

That actually reminds me of February 1989

[LibertyBell]

2 hours ago, bluewave said:

It’s all about the way model biases have been evolving over the years. We went many years near the I-95 corridor when the long range models weren’t snowy enough beyond a few days. Which is the complete opposite of the model error pattern since 2018-2019 with the record breaking Pacific Jet. 

The January 1978 KU  snowstorm was originally forecast to be a rainstorm the day before and we got a surprise 15” snowstorm on Long Island. Then in February 1983 the original forecast a few days out was for the heaviest snows to stay down near Philly. But instead the heaviest snows shifted north into NYC Metro as the models got under 60 hours.

The January 1987 KU event was forecast to be snow quickly changing to rain but instead we got a surprise 6-12”. The December 1988 surprise Norlun wasn't even forecast at all and we got a narrow 6-12” band.

The January 1996 blizzard was originally forecast to get suppressed to the south of NYC from 120 hrs out. But came north in the models only 24 hrs out. The late January 2000 snowstorm was forecast to be OTS and delivered record snows in the Carolinas and over 6” around NYC. 

The record January 2016 snowstorm was only forecast to be 6-12” the day before around NYC and turned out to be 20-30”. 

So model errors in good snowfall seasons tended to be underdone beyond 24 to 48 hours out. 

What the models have been doing since 18-19 is showing heavy snows beyond 72 to 120 hrs only to correct further north with the heavy snowfall axis. So the error has been for the storm tracks to verify warmer and further north than originally forecast. 

This model error pattern worked for us with colder storm tracks to our Southeast which dominated before 18-19. Since the long range forecasts could still afford to shift north changing suppression to a KU. These days the storm track is already too far north so any shift north under 60 hrs gives us to more rain. 

So cutters, huggers, and suppressed Southern Stream storm tracks have lead to the record low snowfall from Philly to Boston last 7 years. The big snowstorm has always been beyond 120 hrs out. So this has become the dominant model forecast storm track bias.

The one good thing about repeating model biases is that I use them in real time to correct the long range model forecast biases leading to better forecasts than the models.

 

AI and quantum computing should be able to help with this too.

[bluewave]

2 hours ago, Terpeast said:

If you’re talking about the biggest KU storms, I tend to agree. But there have been instances in recent winters where sub-KU storms have actually trended south in the last 24-72 hours.

In the 2023-24 winter, my subforum (at least the central part) got 4-6” flush hits twice in the same week that trended to us the last minute after being modeled to go through PA/NY. 

This past winter, we’ve had a few happen the same way. Including one in late Feb that missed us and hit SE VA with a foot of snow, leaving us high and dry. 

So I guess it can cut both ways.

These models just can’t see the Southeast Ridge and strength of the Pacific Jet past 120 hrs. So the snowfall beyond 120 hrs has been shifting further north over time.

February was a prime example of this model bias. The 30”+ totals which were forecast around NYC wound up much further north. Places like Toronto got the record 30” snowfall with only 7.1” in NYC

 

[LibertyBell]

1 hour ago, michsnowfreak said:

This is the deadliest heatwave on record in the U.S. Detroit saw temps over 100° for 7 straight days (peaking at 104 on day 1 and 7). The city allowed people to sleep in parks since their homes/apartments were baking. The MI state record was set in Mio with a temp of 112°.

1953 - The worst tornado of record for the state of Michigan killed 116 persons. Flint MI was hardest hit. The tornado, half a mile in width, destroyed 200 homes on Coldwater Road killing entire families. (The Weather Channel)

 

Interesting you had a tornado on this date in 1953, we had a historic one in 1953 in the northeast too, in Worcester, MA.

1953 was a historically hot summer here with our longest streak of 90 degree days ever (with two historic hot stretches with 4 100+ stretches split between the two superheatwaves!)

[TheClimateChanger]

On 6/7/2025 at 12:43 PM, LibertyBell said:

I wanted to share something, not sure how accurate this is, but I found this while looking up big heat

 

https://www.reddit.com/r/newyorkcity/comments/1epkptr/central_park_ny_highest_maximum_temperature_f/

That 106° reading was on 07/09/1936 during the incredible 1936 heatwave that effected most of the US from the Midwest to the east coast. It was the highest temperature ever recorded in NYC. In those days the official weather bureau temperature was measured at the old Battery Maritime Building near Battery Park where it was always a little cooler due to it's proximity to the water. So the actual temperature in midtown may have been more like 112° if not higher.

My mother remembered that heat wave and said that people left their tenement windows open during the day even if they were at work. A very different city then.

Not accurate. The high was 102F at the Weather Bureau office at the Battery on that date. The monthly mean was a comfortable 74.9F. The 106F was observed at the Park, which, as suggested, would have been secondary to the Weather Bureau office.

[LibertyBell]

1 hour ago, TheClimateChanger said:

Not accurate. The high was 102F at the Weather Bureau office at the Battery on that date. The monthly mean was a comfortable 74.9F. The 106F was observed at the Park, which, as suggested, would have been secondary to the Weather Bureau office.

I think July 1966 was hotter, it had the highest temperature of record at both LGA (107) and JFK (104) and EWR (105) which was tied twice in 1993 and later broken in 2011 (108).

NYC somehow only recorded a maximum of 103 in 1966, which was 4 degrees lower than LGA and a degree lower than JFK.

[Terpeast]

4 hours ago, bluewave said:

These models just can’t see the Southeast Ridge and strength of the Pacific Jet past 120 hrs. So the snowfall beyond 120 hrs has been shifting further north over time.

February was a prime example of this model bias. The 30”+ totals which were forecast around NYC wound up much further north. Places like Toronto got the record 30” snowfall with only 7.1” in NYC

 

One of the things we’re working on with AI models is using hindcast climatology where the 10-15+ day cold bias is much less or no longer there. This might help solve the problem of models being too cold in their medium range forecasts. It’ll also be interesting to see if that also solves the bias of weaker upper level wind flow predictions vs reality.