Jump to content
  • Member Statistics

    17,502
    Total Members
    7,904
    Most Online
    Weathernoob335
    Newest Member
    Weathernoob335
    Joined

And we begin


Recommended Posts

My last post was mostly in jest but this one is serious. The last 3 years obviously didn't pan out and that's fine because no long lead signal is perfect. We're a long ways away from getting there. However, one thing that I would like to see is minimal gains in Sept. Don't remember which year it was but one of the recent ones has large gains in Sept. Have no idea if that means much but if we are going to have big gains this year I would like to see them happen during the prime window and then let the chips fall. 

Link to comment
Share on other sites

Cohen posted the graph below last November with the red line indicating the October SAI.  I manually added the blue and black squares to indicate the corresponding winter AO Index value, and whether the SAI performed well (blue) or poorly (black).  After a promising start, the SAI has performed poorly in each of the last 4 years.

78LYTY8.gif

Rather than drawing a link between snow cover advance and the AO, 3 years ago in this snow cover thread, met poster “millwx” discussed the link between snow cover extent (SCE) at week 42 (mid to late Oct) to population-weighted U.S. winter temperatures.  So, winter temperatures for the U.S., but with added weight given to the population centers in the NE quadrant of the country.  He noted that neither the SAI to AO correlation <NOR> the SCE in Week 42 to population-weighted U.S. temperature correlation were strong, but that he preferred using the latter for inclusion into seasonal predictions.  More info here:

https://www.americanwx.com/bb/topic/44473-and-we-begin/?do=findComment&comment=3066907

 

I put together the charts below to show this correlation using the top/bottom 20th percentile in each case…snow cover extent data goes back to 1966.

Here are the Top 10 Oct SCE cases.  6 of 10 had below normal temperatures.

qxKdiSt.gif

 

Here are the Bottom 10 Oct SCE cases. 5 of 10 had above normal temperatures.

cPLxxfa.gif

 

I was curious as to whether this correlation performed better/worse when including ENSO.

Here are the Top 4 Oct SCE / +ENSO cases (Positive Neutral or El Nino).  3 of 4 had below normal temperatures.

Qu8jxZw.gif

 

Here are the Bottom 4 Oct SCE / +ENSO cases.  3 of 4 had above normal temperatures.

R2J2qbw.gif

 

Here are the Top 6 Oct SCE / -ENSO cases (Negative Neutral or La Nina).  4 of 6 had below normal temperatures.  With -ENSO, one thing I noticed was that Oct SCE tended to be lower as the -ENSO strength increased (i.e. strong La Ninas tended to have lower Oct SCE compared to weak La Ninas).  I didn’t see this type of tendency with +ENSO years.

DvAjTP4.gif

 

Here are the Bottom 6 Oct SCE / -ENSO cases.  4 of 6 had above normal temperatures.

9oKmkwa.gif

 

Finally, here is a chart showing trends in fall Eurasian snow cover extent.

0EqCvsz.gif

 

Summary: After a promising start, the SAI has performed poorly over the past 4 winters.  When using top/bottom 20th percentile data, a correlation exists between Oct SCE at week 42 and population-weighted U.S. winter temperatures.  The correlation isn’t super strong, but one that can possibly be used to enhance winter season prediction ideas regardless of the ENSO phase.

Link to comment
Share on other sites

4 hours ago, griteater said:

Here are the Top 6 Oct SCE / -ENSO cases (Negative Neutral or La Nina).  4 of 6 had below normal temperatures.  With -ENSO, one thing I noticed was that Oct SCE tended to be lower as the -ENSO strength increased (i.e. strong La Ninas tended to have lower Oct SCE compared to weak La Ninas).  I didn’t see this type of tendency with +ENSO years.

DvAjTP4.gif

A strong La Nina tends to correlate with above normal temperatures in Siberia/Eurasia, which is where the vast majority of the snow cover area is located. This would make sense because -ENSO tends to induce a stronger Aleutian/Kamchatka ridge which forces cold air into North America and away from Eurasia. On the other hand, El Nino tends to create a strong Gulf of Alaska low (+EPO), which would have cold air behind it in Siberia, and warmer air in front of it in North America. Remember 2009-2010 was a very warm fall and winter globally, but there was a small area of extremely below normal temperatures in Siberia where the PV resided that cold season.

The PV tends to favor North America in La Ninas and Siberia in El Ninos, at least from what I've seen. 95-96 and 83-84 were two La Nina years that had significant PV intrusions into the United States..early February '96 had -60F temperatures in northern Minnesota, and Christmas 1983 is a very well known arctic outbreak. All of the stronger El Ninos have had very limited PV intrusions...82-83 and 97-98 had none at all, 15-16 had one very brief PV outbreak around Valentine's Day when NYC hit -1F.

Link to comment
Share on other sites

  • 2 weeks later...
  • 2 weeks later...

Earlier today, "Crankywxguy" on Twitter posted:

Surprised we haven't heard the usual nonsense that snow in October in Eurasia = crushing I95 winter. Could it be that it does not work? Yup.

Even as persistent warmth might mask the reality of winter's approach, it is worth looking into the point raised above. 

Even as Eurasian snow advance may have some correlation with blocking, which would favor increased snowfall in the eastern U.S. during the winter all other factors notwithstanding, such a correlation very likely is weak. It is insufficient to forecast the predominant state of the Arctic Oscillation (AO) with much confidence. Therefore, even today, the AO cannot be forecast with reliability at seasonal timescales. 

Had one relied on Eurasian snow cover, recent winters should have featured a lot of blocking. In fact, the AO+ dominated. Last winter was just one such case.

Winter 2016-17 was characterized by a super-dominant AO+ phase. During the December 2016-February 2017 period, 80% of days had positive AO values and more than half of the days that winter had values of +1.000 or above.

December-February Average: +1.008
Lowest Value: -2.228, February 14, 2017
Highest Value: +4.742, December 21, 2016

Link to comment
Share on other sites

5 minutes ago, donsutherland1 said:

Earlier today, "Crankywxguy" on Twitter posted:

<i>Surprised we haven't heard the usual nonsense that snow in October in Eurasia = crushing I95 winter. Could it be that it does not work? Yup.</i>

Even as persistent warmth might mask the reality of winter's approach, it is worth looking into the point raised above. 

Even as Eurasian snow advance may have some correlation with blocking, which would favor increased snowfall in the eastern U.S. during the winter all other factors notwithstanding, such a correlation very likely is weak. It is insufficient to forecast the predominant state of the Arctic Oscillation (AO) with much confidence. Therefore, even today, the AO cannot be forecast with reliability at seasonal timescales. 

Had one relied on Eurasian snow cover, recent winters should have featured a lot of blocking. In fact, the AO+ dominated. Last winter was just one such case.

Winter 2016-17 was characterized by a super-dominant AO+ phase. During the December 2016-February 2017 period, 80% of days had positive AO values and more than half of the days that winter had values of +1.000 or above.

December-February Average: +1.008
Lowest Value: -2.228, February 14, 2017
Highest Value: +4.742, December 21, 2016

Yeah, correlations are pretty meaningless without rigorous statistical tests to determine the likelihood that such an outcome is distinguishable from random chance. Crudely, if you look at griteater's SAI performance evaluation for the last 19 years, 11/19 times positive SAI has occurred with negative AO. That seems like an outcome totally consistent with random chance, and therefore implies minimal dependence of the winter AO on SAI.

Link to comment
Share on other sites

23 hours ago, TriPol said:

Checking in. How are we doing?

Here's a copy/paste of a post I just made in the MA sub:

 

Personally, I'm not convinced that the SAI is a meaningful as it was first sold to us a few years back. I have theory as to why it isn't as reliable but I'll get into that at the end of this post. 

So far the Oct SAI has been showing AN snowcover in Siberia and the month has been progressing in such a way that would support higher odds of a -AO based on Cohen's theory. Looking at the data, the closest match to this day in recent years is 2014. If you guys remember the SAI disco in 2014, we were all banking on a -AO. Oops. 

Current graph:

kZl5U15.jpg

 

Current Chart:

l1n5uG0.jpg

 

Current Anomaly:

LJG2yO5.jpg

 

The last year that had a larger anomaly in Siberia is 2014:

FmXVEti.jpg

 

How is it looking over the next 10 days or so? Pretty good in general. Snow extent south of 60N looks like it will advance quite well over the next 10 days. This is just the 12z gfs panel but ens and previous op runs all support a similar outcome:

 

gfs_asnow_asia_41.png

 

 

IIRC, the importance of the anomaly is based on snow cover growth south of 60N. 2014 was way ahead of this year but this year is going well south of 60N and progs look favorable.  So far Oct 2017 is doing well for increased odds of a -AO based on Cohen's theory. Unfortunately, we've been fooled once or twice or thrice recently so does any of this really matter? I think there is some merit (probably more than we think after getting fooled) but I've formed a bit of a theory as to why it seems to not work so well.

2014 was a SLAM DUNK based on the SAI theory but the end result was pretty lame. We did get a -AO in Jan but the winter on the whole was definitely not "blocky" in the AO/NAO domain. The bigger story in 13-14 and 14-15 was the persistent -EPO. Cohen spun this well because the -EPO did produce cold in the areas that support his SAI theory but nobody can ever convince me that this was not a case of "right for the wrong reasons". A quote from Cohen's blog at the beginning of Oct is what got my gears turning about why the SAI/-AO correlation has been off the last few years:

"Arctic sea ice reached its annual minimum on September 13th. Despite that Arctic sea ice is now growing, large swaths of the North Pacific side of the Arctic basin are ice-free (Figure 11). Recent research has shown that regional anomalies are important and the sea ice region most highly correlated with the winter AO is the Barents-Kara seas region where low Arctic sea ice favors a negative winter AO. However it is early and the magnitude of winter sea ice anomalies in the Barents Kara Seas and other seas in the Arctic are not currently known.  Given that sea ice is running below normal, this currently favors more extensive Siberian snow cover in the coming weeks, followed by a strengthened Siberian high and a weakened polar vortex/negative AO this upcoming winter."

 

The thing is, the low ice area in the Barents and Kara Seas are pretty much guaranteed almost EVERY year now. Unless arctic sea ice makes some sort of unprecedented comeback, low ice volume in this region during the Fall will continue indefinitely. If this anomaly is causing increased Siberian snowcover (seems like it might) then we should have a -AO almost every winter (which we aren't). See where I'm coming from? The current sea ice over the last 10 years has gone through rapid and substantial changes but Cohen's research stretches back many years before that. So my theory is that the SAI theory pre-2007 probably works much better than post 2007. Especially post 2012. I'm probably over simplifying this but it makes sense in my head. If the SAI is grounded on Siberian snowcover in Oct over large time scales but there has been significant changes in weather patterns due to reduced sea ice volume that can't recover to 1980-2010 climo then the entire premise is probably permanently flawed. That's my theory and I'm sticking to it. lol

Link to comment
Share on other sites

2 hours ago, 40/70 Benchmark said:

No one cares about SAI anymore....comparatively speaking lol

What does that have to do with this thread?   SAI is kind of nonsensical because dates and parameters are arbitrary and the sample size is small.  But a snapshot on one day in late October vs other similar years MAY be helpful.  Especially when you compare snow on our side of the NHEM.

Link to comment
Share on other sites

4 hours ago, weathafella said:

What does that have to do with this thread?   SAI is kind of nonsensical because dates and parameters are arbitrary and the sample size is small.  But a snapshot on one day in late October vs other similar years MAY be helpful.  Especially when you compare snow on our side of the NHEM.

Not an indictment of the thread at all. Just loosely related.

 Carry on.

Link to comment
Share on other sites

2 hours ago, 40/70 Benchmark said:

Seriously, I feel its another piece to the puzzle...has its value, but the significance was initially overstated. 

Cohen's last post in his AER blog said something very similar to my thoughts about low arctic sea ice degrading the usefulness of the SAI post 2007. I wasn't too off the wall when I brought it up. 

It does seem like this year is hinting towards increased odds at early winter blocking. Still a long ways to go but there are some signs showing up at least. We'll know much more in the next 6 weeks. 

Link to comment
Share on other sites

10 hours ago, Bob Chill said:

Cohen's last post in his AER blog said something very similar to my thoughts about low arctic sea ice degrading the usefulness of the SAI post 2007. I wasn't too off the wall when I brought it up. 

It does seem like this year is hinting towards increased odds at early winter blocking. Still a long ways to go but there are some signs showing up at least. We'll know much more in the next 6 weeks. 

Agree.

I'm going for a bookend winter, I think.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...