January 2011 Thoughts: Area of Warmth Could Be Somewhat Limited

[Harry]

Several quick points to hopefully clarify things:

1. For the data I posted, the coefficient of determination was: 0.333 and the coefficient correlation was 0.577. That sample used monthly data comparing the reconstructed NAO vs. the AO. Those figures, while not great, are far from indicative of a near meaningless relationship. Unfortunately, neither the AO nor NAO were regularly recorded regularly prior to 1950. The NAO has been reconstructed from a variety of sources. It is an educated guess as to what it might have been.

2. When the AO and NAO are restricted to December-February, the coefficient of determination is 0.551 and the coefficient of correlation is 0.742.

3. The historic data from the reconstuctions was only one part of a larger argument. Ensemble data and ENSO-Teleconnection Index data were also used.

Everything together, not just the piece that relies on reconstructed data, presents a reasonable case that January could well see a more limited area of warm anomalies than is typical during moderate/strong La Niña events. The reconstructed data only hints that there may be some past precedent for the kind of pattern I expect to prevail given general blockiness. Certainly, even if one casts aside the reconstructed data, one could still point to such La Niña events as 1916-17 to note that there were past events that deviated from the more recent experience with moderate/strong La Niñas.

In the end, once January comes to a close, we'll see how things verified.

To the bolded..

You may wanna see this Don..

http://www.cgd.ucar.edu/cas/jhurrell/indices.data.html#naostatseas

[eduggs]

Several quick points to hopefully clarify things:

1. For the data I posted, the coefficient of determination was: 0.333 and the coefficient correlation was 0.577. That sample used monthly data comparing the reconstructed NAO vs. the AO. Those figures, while not great, are far from indicative of a near meaningless relationship. Unfortunately, neither the AO nor NAO were regularly recorded regularly prior to 1950. The NAO has been reconstructed from a variety of sources. It is an educated guess as to what it might have been.

That helps, thanks!

2. When the AO and NAO are restricted to December-February, the coefficient of determination is 0.551 and the coefficient of correlation is 0.742.

3. The historic data from the reconstuctions was only one part of a larger argument. Ensemble data and ENSO-Teleconnection Index data were also used.

Everything together, not just the piece that relies on reconstructed data, presents a reasonable case that January could well see a more limited area of warm anomalies than is typical during moderate/strong La Niña events. The reconstructed data only hints that there may be some past precedent for the kind of pattern I expect to prevail given general blockiness. Certainly, even if one casts aside the reconstructed data, one could still point to such La Niña events as 1916-17 to note that there were past events that deviated from the more recent experience with moderate/strong La Niñas.

In the end, once January comes to a close, we'll see how things verified.

By isolating those January months in which AO or NAO was in the negative phase, I would expect - la nina or not - temperature composites to look cooler across the north and northeast. Certainly a conceptual model of the NA pattern would suggest the AO could dominate or at least mute the impact of La Nina in part of the country. It's excellent that you provided the data to support this idea - and to counter the prevailing sentiment regarding La Nina. But I still believe your sample size is too small to make rigorous statistical conclusions. Regardless, you've persuaded me... I agree with your forecast. But it hinges on a continued negative AO/NAO and persistent "blockiness," which is no sure thing.

[VAwxman]

Just a general question here. There have been at least 2 persistent blocking features that ive noticed popping up this season, one obviously being the NAO, but the Aleutians block has also been very persistent. Is the forcing behind this ridge from La Nina?

To an extent it is. You typically do have a ridge in the North Pacific with a la nina, but this one has been displaced farther north, like it was with the ninas of old (note that in the warm nina years like 98-99 and 99-00 it was farther south). I think the jet extensions we have seen coming off eastern Asia, tied into the snowcover over there, have helped push this ridge farther north.

[donsutherland1]

To the bolded..

You may wanna see this Don..

http://www.cgd.ucar....tml#naostatseas

Many thanks, Harry.

[donsutherland1]

But I still believe your sample size is too small to make rigorous statistical conclusions.

On that, I strongly agree. It is unfortunate that the sample sizes are so small. There most definitely is a lot of uncertainty. If that data, alone, were indicating cold, I would be hesitant to support that idea.

[Harry]

Many thanks, Harry.

NP. I have really enjoyed your discussions. I know from a few that site/data is VERY reliable ( as far as the NAO goes ) and figure it may help some. Have one for older AO stuff too but unsure of how reliable that is.

[donsutherland1]

Pac NW tidbit.

This morning the temperature stood at 27°F (-2.8°C) in Seattle. Recent runs of the GFS have indicated that a bout of severe cold could overspread the Seattle area beginning in 7-10 days, possibly preceded by a long-duration snow event. The 1/2/11 6z run of the GFS brings historic-type cold to the Seattle area in the longer range. The meteogram follows:

While exact details will likely differ, the idea of an EPO-/PNA-/AO- pattern around 1/10 can be favorable for cold outbreaks. Considering the forecast EPO, it is likely that some of the world's coldest air could be available to be tapped. Hence, a severely cold periof for the Pacific Northwest is probably more likely than not. Whether the cold reaches the historic magnitude shown on the 6z GFS remains to be seen.

[donsutherland1]

The 1/3/2011 18z run of the GFS essentially takes away the Arctic blast that had repeatedly been modeled for just after mid-January. The 348-hour 850 mb temperature anomalies forecast on that run are:

At this time, I don't believe it is correct. With the ensembles suggesting an AO of -4.5 to -2.5 and PNA of 0 to +1.5 at that time, the composite temperature anomalies for January (1950-2010) for that setup were:

At the same time, the 18z GFS ensembles at 348 hours showed:

The 18z GFS ensembles are strongly consistent with the historic composite pattern that would be expected based on the teleconnection indices. Hence, at this time, I do not believe the 18z run of the operational GFS is likely to be correct in the long-range. Instead, I continue to expect a major Arctic outbreak that will push into the eastern U.S. just after mid-month.

[pcbjr]

The 1/3/2011 18z run of the GFS essentially takes away the Arctic blast that had repeatedly been modeled for just after mid-January. The 348-hour 850 mb temperature anomalies forecast on that run are:

At this time, I don't believe it is correct. With the ensembles suggesting an AO of -4.5 to -2.5 and PNA of 0 to +1.5 at that time, the composite temperature anomalies for January (1950-2010) for that setup were:

At the same time, the 18z GFS ensembles at 348 hours showed:

The 18z GFS ensembles are strongly consistent with the historic composite pattern that would be expected based on the teleconnection indices. Hence, at this time, I do not believe the 18z run of the operational GFS is likely to be correct in the long-range. Instead, I continue to expect a major Arctic outbreak that will push into the eastern U.S. just after mid-month.

Thank you, Don, for a voice of reason!

[Snow Wizard]

The 18z GFS ensembles are strongly consistent with the historic composite pattern that would be expected based on the teleconnection indices. Hence, at this time, I do not believe the 18z run of the operational GFS is likely to be correct in the long-range. Instead, I continue to expect a major Arctic outbreak that will push into the eastern U.S. just after mid-month.

Looks like the West is screwed again. When do you think our next chance is? Talk about frustrating.

[donsutherland1]

Looks like the West is screwed again. When do you think our next chance is? Talk about frustrating.

The Eastern Arctic outbreak would come after a period of notable cold in the Pacific Northwest, not instead of it.

[Snow Wizard]

The Eastern Arctic outbreak would come after a period of notable cold in the Pacific Northwest, not instead of it.

I hope so. The 12z and 18z GFS really took most of it away from us. They could have been bad runs though.

[donsutherland1]

I hope so. The 12z and 18z GFS really took most of it away from us. They could have been bad runs though.

And the 1/4/2011 0z GFS returned the extreme cold to the Pacific Northwest.

It also returned the Arctic outbreak to the East. In short, it is much more reasonable given the expected pattern evolution.

[Ottawa Blizzard]

Don, things finally seem to be looking up for my region. I have high hopes for the system that moves in the middle of next week.

[thenewbigmack]

The 1/3/2011 18z run of the GFS essentially takes away the Arctic blast that had repeatedly been modeled for just after mid-January. The 348-hour 850 mb temperature anomalies forecast on that run are:

At this time, I don't believe it is correct. With the ensembles suggesting an AO of -4.5 to -2.5 and PNA of 0 to +1.5 at that time, the composite temperature anomalies for January (1950-2010) for that setup were:

At the same time, the 18z GFS ensembles at 348 hours showed:

The 18z GFS ensembles are strongly consistent with the historic composite pattern that would be expected based on the teleconnection indices. Hence, at this time, I do not believe the 18z run of the operational GFS is likely to be correct in the long-range. Instead, I continue to expect a major Arctic outbreak that will push into the eastern U.S. just after mid-month.

Hey Don... I've been following your posts for quite a while now, just haven't posted. I live in the PNW, but that doesn't matter for this question. I saw your composite map above, based on the teleconnection indicies, and I'm wondering how you went about plotting that. I'm aware there is a site, which i have use, to plot certain years, months, etc... but are you actually able to plot indicies like above? Or did you go about it some other way? thanks.. and keep up the great work! I appreciate the input you give for the West as well, not many on here care to share their thoughts for us

[donsutherland1]

Hey Don... I've been following your posts for quite a while now, just haven't posted. I live in the PNW, but that doesn't matter for this question. I saw your composite map above, based on the teleconnection indicies, and I'm wondering how you went about plotting that. I'm aware there is a site, which i have use, to plot certain years, months, etc... but are you actually able to plot indicies like above? Or did you go about it some other way? thanks.. and keep up the great work! I appreciate the input you give for the West as well, not many on here care to share their thoughts for us

Thanks for the kind words. I took all the dates for January (1950-2010) where the indices were in a given range (as had been forecast on the GFS ensembles) and then used CDC's daily re-analysis site (http://www.esrl.noaa...sites/nssl/day/) to plot the maps. I had to use my own data file (.txt format), as there were 77 dates. The data was submitted to CDC's ftp page (ftp://ftp.cdc.noaa.g...incoming/Dates/) . Click on "Incoming." Then, the folder to which to paste the data is "dates." The txt file's format is:

# of dates

each date (yyyymmdd format)

e.g.

3

20080105

20090124

20090125

[thenewbigmack]

Thanks for the kind words. I took all the dates for January (1950-2010) where the indices were in a given range (as had been forecast on the GFS ensembles) and then used CDC's daily re-analysis site (http://www.esrl.noaa...sites/nssl/day/) to plot the maps. I had to use my own data file (.txt format), as there were 77 dates. The data was submitted to CDC's ftp page (ftp://ftp.cdc.noaa.g...incoming/Dates/) . The folder in which to paste the data is "dates." The txt file's format is:

# of dates

each date (yyyymmdd format)

e.g.

3

20080105

20090124

20090125

Thank you for the quick reply! This all makes sense, however, I'm wondering if you actually went through and had to research all those dates for January, or if you have some way of having those dates listed for you given the forecasted indicies. I'm looking to do something similar as to what you did, but for different times throughout the winter, so the inidcies will obviously vary. Thanks again!

[donsutherland1]

Thank you for the quick reply! This all makes sense, however, I'm wondering if you actually went through and had to research all those dates for January, or if you have some way of having those dates listed for you given the forecasted indicies. I'm looking to do something similar as to what you did, but for different times throughout the winter, so the inidcies will obviously vary. Thanks again!

I took the daily index values: ftp://ftp.cpc.ncep.noaa.gov/cwlinks and copied them into Word and then saved it as a text file. Afterward, I imported the text file onto a spreadsheet, sorted by the data, then constructed the text file to be uploaded to the "Dates" folder e.g., converted 1/1/2010 to 20100101. When creating the text file to be uploaded to the "Dates" folder, be sure not to have any extra spaces and also be sure that the number of dates actually matches the dates listed e.g., if there are three dates, then the number has to be 3. Either issue will prevent the file from running.

[thenewbigmack]

I took the daily index values: ftp://ftp.cpc.ncep.noaa.gov/cwlinks and copied them into Word and then saved it as a text file. Afterward, I imported the text file onto a spreadsheet, sorted by the data, then constructed the text file to be uploaded to the "Dates" folder e.g., converted 1/1/2010 to 20100101. When creating the text file to be uploaded to the "Dates" folder, be sure not to have any extra spaces and also be sure that the number of dates actually matches the dates listed e.g., if there are three dates, then the number has to be 3. Either issue will prevent the file from running.

Ahhh makes sense. Sorry for all the questions, this is definitely something that could help me progress as far as my learning goes. Something that still doesn't click is how you go about eliminating all the dates that don't fit the specifications. Once you have dates eliminated for each indicie, then do you go through and find the repeating dates for all the selected indicies some how? Thank you for the help once again, sorry to be a pain

[donsutherland1]

Ahhh makes sense. Sorry for all the questions, this is definitely something that could help me progress as far as my learning goes. Something that still doesn't click is how you go about eliminating all the dates that don't fit the specifications. Once you have dates eliminated for each indicie, then do you go through and find the repeating dates for all the selected indicies some how? Thank you for the help once again, sorry to be a pain

As an example, let's say one chooses all the AO values between -3 and -2. Then after the sort, that's all one is left with. For purposes of a broad composite, that would suffice. Sometimes, if one is looking for even closer matches, one would have to check each date or set of dates (if there is a sequence) and match the 500 mb pattern for those date(s) with the forecast 500 mb pattern. Those that have a reasonable match would be retained. Others would be dropped. At times, though, that leads to a very small sample size and high amount of uncertainty. Furthermore, there is a degree of subjectivity involved. It should also be noted that I'm quite hesitant to roll composite patterns forward for more than 10 days. Too many differences among individual dates occur, greatly increasing the uncertainty involved.

[thenewbigmack]

As an example, let's say one chooses all the AO values between -3 and -2. Then after the sort, that's all one is left with. For purposes of a broad composite, that would suffice. Sometimes, if one is looking for even closer matches, one would have to check each date or set of dates (if there is a sequence) and match the 500 mb pattern for those date(s) with the forecast 500 mb pattern. Those that have a reasonable match would be retained. Others would be dropped. At times, though, that leads to a very small sample size and high amount of uncertainty. Furthermore, there is a degree of subjectivity involved. It should also be noted that I'm quite hesitant to roll composite patterns forward for more than 10 days. Too many differences among individual dates occur, greatly increasing the uncertainty involved.

I was wondering how far you were willing to roll the composites forward. I know understand how you go about this... I just need to learn how to sort the data in excell. Benn a while since I used that =P I'm sure this might take a while to get down, but once I get it, I'm sure it will become much easier. I'm assuming the same applies when you are sorting multiple indicies, but just do each one individually. Taking a look at 500mb patterns also seems like it may be very helpful, but like you said, being too percise and you're left with a very small sample size. Great work

[donsutherland1]

With an Arctic outbreak possible for the eastern U.S., after a period of notable cold in the Pacific Northwest and on the Plains, probably just after mid-month (perhaps a few days after the AO bottoms out), the following are some maps from past extreme cold in the East (1950-2010). For purposes of focusing on extremes, I used cases in which the minimum temperature was < 10°F in Boston, New York City, Philadelphia, and Washington D.C. on the same date.

500 mb Pattern 5 days before:

500 mb Pattern for those cases:

Surface Temperature Anomalies 5 days before:

Surface Temperature Anomalies for those cases:

850 mb Temperatures 5 days before:

850 mb Temperatures 3 days before:

850 mb Temperatures for those cases:

Several quick points:

1. The cold generally enters the U.S. over the Northern Plains/Great Lakes and then comes eastward. There are exceptions, including direct discharges into New England.

2. 5 days before the extreme cold, there is a large area of 850 mb temperatures < -25°C in Canada; the core of the coldest surface temperature anomalies appears to be aimed at the Northern Plains/Great Lakes region and even covering part of the Northern Plains (the extended range on the GFS ensembles has that "look")

3. 3 days before the extreme cold, there is both a large area of 850 mb temperatures < -25°C in Canada and also a significant area of -30°C to -28°C 850 mb temperatures in Canada; the extreme cold is overspreading the Great Lakes region

4. On the day of extreme cold, the worst of the cold (< -25°C 850 mb temperatures is centered over central/southern Ontario and Quebec); the -20°C 850 mb isotherm extends from eastern Ohio, across almost all of Pennsylvania, across much of New York State just north of New York City, and into New England; the eastern half of the U.S. and all of southern Ontario/southern Quebec are covered by much below normal to below normal surface temperature anomalies; the most extreme anomalies are centered over the Ohio Valley.

These are things to look for as the timeframe for the possible Arctic outbreak draws closer. Even as the 1/4/2011 6z run of the GFS temporarily delays the Arctic outbreak, its ensemble mean has maintained remarkable continuity with earlier ensemble runs as to the timing of the possible severe cold. The ensembles also remain in good agreement with the pattern one would reasonably expect given the forecast teleconnection indices.

[Ji]

what about snow donald?

[donsutherland1]

I was wondering how far you were willing to roll the composites forward. I know understand how you go about this... I just need to learn how to sort the data in excell. Benn a while since I used that =P I'm sure this might take a while to get down, but once I get it, I'm sure it will become much easier. I'm assuming the same applies when you are sorting multiple indicies, but just do each one individually. Taking a look at 500mb patterns also seems like it may be very helpful, but like you said, being too percise and you're left with a very small sample size. Great work

I will extend them out 5-10 days, but rarely longer unless there is strong consistency among the individual dates. Only then, I might look another 5 days out. I might also take the most consistent dates and roll them out another 5 days. However, I don't go beyond 15 days under any circumstances. But that's a value judgment on my part. Also, it should be noted that the composites give general ideas and insights. Exact details could well differ. Hence, while I remain reasonably confident that Seattle will likely see multiple days with lows in the teens (and probably one day with a low below 15°F) and probably 3"-6" of snow for January given the pattern, I will wait for the event to draw closer before even speculating whether something even colder might occur. The GFS has some single digits. Given the rarity of that event, I want to see how things evolve closer to the event before making any such calls on such extreme cold. I would take a similar approach with respect to possible subzero cold in parts of the East. The pattern is one that has, in the past, produced single digit readings, so the idea of at least one single digit reading in let's say NYC or Philadelphia appears reasonable if the pattern delivers as I think it will. It's far too soon to speculate on whether any of the East's major cities from PHL to BOS might have a subzero low. Chicago very likely will have such a day. Detroit might also have such a day. Ottawa almost certainly will.

[donsutherland1]

what about snow donald?

Right now, that's a very tough call with the La Niña. There might be an opportunity for your area in the January 11-13 timeframe, but that's not yet assured. I am cautiously optimistic about that possible system.

The upcoming Friday/Saturday system/inverted trough is unlikely to bring anything more than flurries, if that. Some part of New England and possibly central New York State probably has the best chance at seeing 4" or more.

In the general pattern that is evolving, there might also be some overrunning opportunities beyond the January 11-13 timeframe, perhaps even a system that brings some light snow as it ushers in the Arctic air.

[thenewbigmack]

I will extend them out 5-10 days, but rarely longer unless there is strong consistency among the individual dates. Only then, I might look another 5 days out. I might also take the most consistent dates and roll them out another 5 days. However, I don't go beyond 15 days under any circumstances. But that's a value judgment on my part. Also, it should be noted that the composites give general ideas and insights. Exact details could well differ. Hence, while I remain reasonably confident that Seattle will likely see multiple days with lows in the teens (and probably one day with a low below 15°F) and probably 3"-6" of snow for January given the pattern, I will wait for the event to draw closer before even speculating whether something even colder might occur. The GFS has some single digits. Given the rarity of that event, I want to see how things evolve closer to the event before making any such calls on such extreme cold. I would take a similar approach with respect to possible subzero cold in parts of the East. The pattern is one that has, in the past, produced single digit readings, so the idea of at least one single digit reading in let's say NYC or Philadelphia appears reasonable if the pattern delivers as I think it will. It's far too soon to speculate on whether any of the East's major cities from PHL to BOS might have a subzero low. Chicago very likely will have such a day. Detroit might also have such a day. Ottawa almost certainly will.

Great thinking and analysis! Good to hear your believe seattle will in fact see some decent snow this month. I myself am up right by the canadian border so things tend to be a little colder where I'm at, however, would love to see everyone share the love. I've been learning weather for about a year, and I really would like to learn how to better forecast the medium to long range. Using the models, especially with not much agreement lately, isn't exactly the best way to make a forecast This method you and I have been discussing seems to help out a lot. I don't have excel on this computer, however I'm download openoffice right now which is pretty much the same thing, I will just need to learn how to change the date format and sort through the data as you said. Interesting times coming up for many!

[donsutherland1]

Great thinking and analysis! Good to hear your believe seattle will in fact see some decent snow this month. I myself am up right by the canadian border so things tend to be a little colder where I'm at, however, would love to see everyone share the love.

I agree. I hope that the Pacific Northwest will experience some great winter weather.

I've been learning weather for about a year, and I really would like to learn how to better forecast the medium to long range. Using the models, especially with not much agreement lately, isn't exactly the best way to make a forecast...

In such cases, one should be cautious about model details and place greater weight on the ensembles until the situation becomes clearer. I highly recommend that your read Wes's (Usedtobe) discussions. He is as good as one gets. One can always learn from his work. AmericanWx and CapitalWeatherGang are fortunate to be able to have access to his talent, knowledge, and expertise.

[thenewbigmack]

I agree. I hope that the Pacific Northwest will experience some great winter weather.

In such cases, one should be cautious about model details and place greater weight on the ensembles until the situation becomes clearer. I highly recommend that your read Wes's (Usedtobe) discussions. He is as good as one gets. One can always learn from his work. AmericanWx and CapitalWeatherGang are fortunate to be able to have access to his talent, knowledge, and expertise.

I agree... although this pattern has not acted "Nina-like", I'll be happy if we can share the cold, in a back and forth kind of manner (or at least thats what it seems like). I will definitely give his work a read through, if you have any suggested articles or posts, please let me know. As far as the ensembles go, you like putting more faith in them for the LR? They seem to have been much more consistent with this approaching pattern than the operational runs have been. I have heard people say before that the operational runs kind of "catch up" to the ensembles, but not sure what to think as far as that goes

[donsutherland1]

I will definitely give his work a read through, if you have any suggested articles or posts, please let me know.

One can run a search at AmericanWx for "usedtobe" and look for his more detailed discussions. He's also written papers and they can probably be found online through Google.

Two papers that I have found invaluable can be found at:

http://www.hpc.ncep.noaa.gov/research/snow2a/snow2a.pdf

http://www.hpc.ncep.noaa.gov/research/mcs_web_test_test.htm

As far as the ensembles go, you like putting more faith in them for the LR? They seem to have been much more consistent with this approaching pattern than the operational runs have been.

During periods of high uncertainty, the ensembles are very helpful. La Niñas generally have a greater amount of undertainty than El Niños given the fast flow, need for phasing for big East Coast events, etc. The current situation has a lot of uncertainty. One saw how yesterday's 18z run of the GFS basically took away the Arctic attack (both for the Pac NW and eastern U.S.). The 0z run brought it back. The 6z run has the severe cold in the Pac NW but has delayed the arrival of the severe cold in the East. The ensembles have been more consistent and they fit what one would reasonably expect from past patterns similar to the one that is evolving.

[donsutherland1]

After having reviewed the 12z GFS, I believe my general ideas for January remain on track:

1. January 2011 will likely have a more limited than usual area of warm anomalies for a moderate/strong La Niña event.

2. The Pacific Northwest will have some notable winter weather, including a period of severe cold and opportunities for snowfall, including Seattle and probably Portland, too.

3. Arctic air capable of sending the temperature into the single digits from at least Philadelphia northward is likely to move into the East just after mid-month. Whether it comes all at once or in pieces remains to be seen. The lowest readings could occur in the January 16-21 timeframe (a few days after the AO bottoms out). Ottawa will likely have one or more days on which the mercury falls to or below -20°C (-4°F), something that has not yet occurred this winter.

4. I am increasingly confident that both Boston and New York City will see monthly snowfall of 10" or more in January. That would fit the historical cases following excessive December snowfall.

5. The January 7-9, 2011 event will likely produce an area of light to possibly moderate snows across parts of eastern Pennsylvania/New Jersey and Southeast New York State. The highest probability of moderate or greater snows will likely occur over central or southern New England.

6. Another event around the January 11-13, 2010 timeframe should have some southern stream energy and will likely have a larger-scale impact than the January 7-9 system. Snow could fall from Richmond into New England. Details will still need to be resolved and the exact details of the storm will likely differ at least somewhat from those depicted on the 12z run of the GFS. What's important is that the potential exists for a larger event than the January 7-9 system. Whether or not it actually becomes the season's second KU storm, something that would be unprecedented during any moderate/strong La Niña event since 1871, is another issue that will have to be resolved with later guidance.