raindancewx

I'm not expecting this winter to drop below 25.50C in Nino 3.4 If that's the case, we'll have gone every winter since 2010-11 without a La Nina below 25.5C - one of the longer time frames for that since 1950. There were some widespread rains over western NM and Arizona on 10/5-10/6 - that exact time showed up as a favored storm date in the Southwest in my analogs, so I think my outlook is on the right track with the timing variables I use. I've mentioned 2011 a lot in this thread for similar La Nina development. Let me give some reasons why I don't particularly like it as an analog for actual weather: 1) MJO timing is a bit off. We were at the 4/5 border on 10/1. For 2011, the MJO was at 5/6 on 10/1. 2) No Gulf Hurricanes. I don't mean US landfalls...I mean any hurricanes. At all. That's pretty weird and I don't like what it says about how the Atlantic is behaving. Probably not coincidentally, the AMO was also negative in winter 2011-12. Not looking that way this year. 3) 2011 was very cold in the Northwest in June (remember the 120s in Canada this past June?). When it did get hot int he Northwest in 2011, in September, it was fairly cool in 2021. 4) Arctic Sea Ice extent was considerably higher at the minimum value in 2021, despite following a much lower minimum year. On balance, that means Arctic patterns were more favorable for sea ice retention for a full year from 9/2020 (3.8m square km min) to 9/2021 (4.7m square km min) then from 9/2010 (4.6m square km min) to 9/2011 (4.3m square km min), especially given the warming up there since 2010. Long-term the lowest sea ice extent minimums are pretty correlated to the WPO in winter. It's not hard to find similarities to 2011 though - 1) Second year La Nina winter 2) warmer than preceding winter in Nino 3.4 3) rising solar activity 4) similar subsurface development (warm Spring before cooling) 5) record precipitation Summer Northeast in a year with tropical storm hitting NE US 6) very negative PDO My guess is 2011 will be a decent match to US precipitation patterns for a while (I'd give it a B for that) with the PDO state similar and the La Nina developing in a similar way. But I don't think it's going to be a good analog at all for temperatures. If you look nationally, it's definitely one of the best matches in the past century for precipitation in September.

Have to be careful with assuming the subsurface cold will translate fully to the surface. September 1983 was 26.2C in Nino 3.4 like this year. But the subsurface was -2.15 in September, much colder than this year. Still only a 26.0C event in winter. The coldest reading was 25.40C in November 1983 with that La Nina. The warmth in the PMM area and then by Indonesia (at and below the surface) also are going to get mixed in with the cold coming up at some point. Given how closely the subsurface has mimicked 2011 for Jan-Sept, the warm up below the surface should begin around Halloween, give or take two weeks. My guess would be November / December coldest at the surface again and then fairly rapid decay starting late January.

BAM for comparison. The joke with last winter is it'd be near impossible to be much warmer than December or much colder than February on a national basis. So two thirds of the winter forecast trend are kind of known already on a y/y basis.

Objective best matches for January-September highs and precipitation in Albuquerque heading into La Nina winters...rolled forward to February. Local blend of these six averages eight days 10F or colder, with two of the eight days 20F or colder. I'm warmer than this nationally, but it is fairly similar to what I have in my real analogs for February.

Might as well mention this here: Nino 3, Nino 3.4, and Nino 4 have tracked very similarly to a blend of 1967, 2001, 2011 at the surface for January-September. When you roll the blends forward, you get the following for winter. 24.95C in Nino 3 (-0.65C v. 1951-2010) 25.95C in Nino 3.4 (-0.55C v. 1951-2010) 28.00C in Nino 4 (-0.10C v. 1951-2010) Nino 1.2 has been running warm of the blend all year, and September was close to +0.5C v. the blend in Nino 3. So if anything, Nino 3 may be somewhat warmer than what I have. At some point, these blends will break - but right now actual data keeps coming in slightly warmer than the blend. So there is space for quite a bit of cooling when the cold water does reach the surface. At this point, I see no reason for the blend not to hold for the La Nina structure through at least December. I have a table showing the similarities of the three years to 2021 for the three zones in my outlook, I'll link it in a few days. Last winter was similar in Nino 3, colder in Nino 3.4 (25.58C) and Nino 4 (27.32C).

Does it bug you that October 2011 looks a bit like the actual temperature pattern to date for this month? September kind of does too. I don't actually like it as an analog for US weather, just for how the La Nina has developed at and below the surface. If you buy the CPC stuff it looks cold in the West for at least a little while. Lots of cold coming. My analogs were split on the WPO phase in October, and that's important out here in the fall, but they all have it positive in November. My thing with 2001 is it has a relatively early SSW but had a warm US November which I expect. But following the warm November, the -NAO in December kind of didn't do much for the East in 2001. Similar to last year in that sense. I think Weatherbell and crew have something like a -WPO, -NAO, -AO December look for the East and that's why they have it cold. But the colder ENSO years with that kind of look are not that cold since 2000, it's 2000, 2001, 2005. Solar stuff I find is mostly important for mountainous areas with weird atmospheric properties. Cold air is dense - and the mountain tops are generally colder than valley bottoms outside of inversion setups. But elevated air is also thin. So I think of it as if we have "thin" cold air - and so the sun impacts cold air more on a mountain in New Mexico than a place like Boston or San Diego or Chicago or Memphis or Jacksonville. I have some proprietary maps on my PC of global cloud patterns by month with statistical signifiers for meaningful changes, and the solar changes do impact cloud patterns globally at pretty remarkable statistical strength. If I remember right the wave-lengths below violet, which is the lowest wave-length color (hence 'ultra violet') vary by 6-8% from the solar minimum to the solar maximum via direct observations. I've also seen research saying that solar activity directly influences the Indian Monsoon as well as the other monsoon circulations. I've found similar things here. August is the wetter month in high solar years, July is the wetter month in low solar years. There are then meaningful differences in which parts of the cold season are wettest based on when the monsoon was strongest.

As far as determining the ENSO type, the easiest way is just to use an imprecise filter and then find the deeper colors. Look at how last year changed for instance: I still haven't found a better match on a month to month basis at the surface to 2021 than a blend of 1967, 2001, 2011. That blend is 25.95C in Nino 3.4 in winter, but closer to 28.0C in Nino 4, and then around 25.0C in Nino 3 for winter. Each month has been pretty close in each zone. If anything the blend is too cold in Nino 3 since it was warmer in September, with Nino 1.2 running warmer consistently. I have a table in my winter outlook - I'll show how close that blend is in a few days.

These are the Februaries I mentioned that match closest to Jan-Sept 2021 highs & precipitation locally, in La Nina years: My analog blend for February isn't that cold. But it's actually not that different of a look.

I have a pretty wild March. This is not the blend I'm using in each month, but I do think this will work in March. Also, this winter to me screams fluky snow events in the South. It won't be the same as last year, but each of the years I'm using has one to two meaningful snow events almost down to the Gulf Coast even though I have the South pretty warm.

My forecast for winter is just about done. I'll link to it here in a few days. Euro still has a weaker La Nina than last year with the new update. I'm expecting Nino 3.4 to fall as low as 25.4-25.7 in Oct/Nov before rebounding warmer in winter. Last year fell to 25.28 in November before warming. The Euro still uses 1981-2010 as the baseline, so CPC figures will come in 0.05-0.10C colder on the ONI site using the warmer, newer baseline. I classify La Ninas by how strong they are in winter. So to me, last year was a weak event, since it was only "moderate" in Oct-Nov, and weak the rest of the duration. This event may never actually drop below -1.0C even against the warmer recent baseline. I still like 25.75-26.15C for winter overall, but expecting 25.95C.

I get the same thing as you for the NAO going negative in December. But I also think it might go negative in late-February and March for a time. The Southwest had a much wetter and colder Summer than last year, which does conceptually fit with a warm winter out here in a cold ENSO year (see 2005, 2013, 2017, etc). Nonetheless: I have to tell you, when I objectively match and rank precipitation totals and highs for each month from January to September, there is a strong signal for another huge cold wave in February in the Southwest if you filter on closest weather in La Nina years. When i ran the blend of the best matching La Ninas year to date, I got ~8 days 10F or colder in February, with ~2 days 20F or colder in February for my location, and that's including some of the "warmer" Februaries looks. I think 2-3 of the six closest La Nina years had temperatures 30-40 below average on a couple days. Other things sort of support it too. A lot of times February gets stupid cold in the Plains following a +AO, +WPO, +NAO look in November (1978, 1979, 1992, 1993, 2020 as examples). June cold snaps in the Southwest (I used highs under 70F in Albuquerque from 6/15-7/15) also tend to support cold in mid-Nov to mid-Dec or mid Feb to mid-March. Also, the mid-Dec to mid-Jan period is highly sensitive to the total Atlantic ACE Index here. Roughly ~20 points of ACE adds (removes) 1F off the high for that period. I am not convinced at all we hit 180 ACE again this year. That said, the major hurricane days are very highly correlated to dryness out here (more than ACE), so I'm actually introducing a -40% precipitation spot in the Southwest for the winter, with 13.75 major hurricane days already (8.75 for all of last year, despite 180 ACE).

I actually don't talk about the EPO much because where I live it has literally has no predictive value for anything for temperatures. I do agree with the Typ guy though, you can generally figure out what the EPO will do if you get the WPO right. The WPO kind of bullies everything around more than having a direct impact on the US most of the time. I saw someone say I like 1967, 2001, 2011 - I do -but only for the La Nina development. Those years as a composite are not real close to what I expect for winter at all, especially for March. I have this for winter, but I am using different weightings for the individual months this year, since some of the years are much better for precipitation than temperatures nationally. I would argue knowing I have this for winter is almost useless since it varies so much month to month, even within the months. The hindcasts for 1995-2020 on my NAO thing have the average DJF error at +/-0.36. I have a slide showing each of those 20+ hindcasts in my winter forecast. Is that useful? Sort of - it can be - rarely misses by more than 0.60 in that time frame. So when it sees a big + or - signal, it's very unlikely to be wrong. But it's not some Word of God thing for the month to month readings. For quick review, it's the NAO change in Apr to May blended with the NAO change Mar to Sept, anchored by similar prior ENSO winters. So last year I did 1961, 2003, 2007 as a blend. Had the -NAO in December, the +NAO in February, and the composite missed the -NAO in January, even though Jan 2004 was negative. Anyway, this year, it's a blend of 1967, 1976, 2000, 2001. All four negative - pretty negative at that - for NAO in December. Jan/Feb both neutral for the NAO. March probably negative, but this method has no skill in March. (i.e. it's no better than a guess in March). Long-term, if you look at February, the biggest/highest +NAO years are never after cold Nino 4 winters in the prior year. If you look back to February 2013, that's the last -NAO February, and it followed a 27.30 or something Nino 4 winter in 2011-12. That's the last cold Nino 4 winter before last year. There is some potential for -NAO conditions to last longer than last February or more intensely but briefly. I don't think the NAO index will matter much though in February if I'm doing my outlook right. But the long-term correlation for DJF Nino 4 and the following February NAO value is around 0.06 r-squared for 1950-2020, which isn't strong, but you have no real outliers. The most fun month to me on a national scale is March if I did this right. That month looks cold, tornadic, and snowy to me.

I'll link to my outlook here when I am done. I was checking overall snow patterns in the individual years yesterday. Each of the five analogs I am using had fluky deep-South snow events. The real crazy thing is they all are Northwest of the landfalling systems in the Gulf - typically Louisiana and Texas. My best guess is if there is a fluky deep-South snow event, the timing would be early December (12/1-12/10) or late January (1/21-1/31). We'll see. The snow map showing the fluky pattern will be in my outlook.

The obvious read on the cold season is it probably won't be as warm as last year nationally in Nov-Dec, or as cold as February in the Plains. Probably not as warm as March either. I actually still have January pretty warm though. December & March are my guesses for most different year over year from Nov-Mar. I do have February much warmer, but still pretty cold. Checked the other day - there were fluky deep South snow events in each of my analogs. Trying to figure out when the hell that's most likely to happen. My guess is early December or late January, but it's definitely interesting to see it. Can make the case for 1-2 incredible cold waves in winter in an otherwise warm pattern nationally if you look at years with similar Summers-Falls in 15-day increments.

Nino1+2 Nino3 Nino34 Nino4 Week SST SSTA SST SSTA SST SSTA SST SSTA 01SEP2021 20.3-0.2 24.7-0.3 26.5-0.3 28.4-0.3 08SEP2021 20.7 0.2 24.6-0.3 26.4-0.4 28.5-0.3 15SEP2021 20.6 0.2 24.3-0.6 26.3-0.4 28.4-0.4 22SEP2021 20.6 0.2 24.5-0.4 26.4-0.3 28.2-0.5 29SEP2021 20.5-0.0 25.1 0.3 26.8 0.0 28.2-0.6 With Nino 3 warming a lot the past few weeks I would bite on the current week as a Modoki La Nina look. But I also don't think the month will finish anything like this. In any case, still running much warmer than last year. The 40 or so years since 1950 with cooling in Nino 3.4 from September to Dec-Feb average about ~0.35C of cooling, with 80% of the cooling years seeing a 0.05C to 0.65C drop from Sept to DJF. 02SEP2020 18.9-1.7 23.6-1.4 25.8-0.9 28.4-0.4 09SEP2020 19.5-1.0 23.4-1.4 25.7-1.0 28.5-0.3 16SEP2020 20.0-0.4 23.6-1.3 25.9-0.8 28.2-0.5 23SEP2020 19.6-0.8 23.6-1.2 25.7-1.0 28.1-0.6 30SEP2020 20.1-0.4 23.8-1.1 25.6-1.1 27.9-0.8 Year to date for subsurface, 2011 is objectively the best subsurface match - I suspect pre-1979 years match better, but the data doesn't go back that far. You can see in the Ja-Sp (Jan-Sept) column that the total difference between 2011 and 2021 is about ~0.26/month (2.37 divided by 9 months), and it's double that all other years, even 2017 in second place. Year to date at the surface, Nino 3.4 and 4 are still remarkably close each month to a blend of 1967, 2001, 2011. Nino 3 is actually very close overall too, but September was much warmer than that blend.

Nino 3.4 came in at 26.21C. That's +0.32C from last year. Nino 4 is very similar to last year for September. To be honest, Nino 3 is actually the most interesting - the cold water below the surface is there. But at the surface, it is running warmer than years I don't even consider La Ninas winters like 1959, 1961, 1962, 1966, 1967, 1978, 1985, 1996, 2001. Nino 1.2/3 are both over 0.5C warmer than last year in September.

I'm still pretty happy with the blend I have for winter at the moment. You can see in the 5N-5S, 120W-170W area which is Nino 3.4, it's very close. It's a pretty strong match on the La Nina. There are other issues, but the IOD and AMO should move toward that blend of years with time. It's pretty hard to get a good match on the Nino zones at the surface for September. None of the individual years are great matches. 1959, 1985, 2000, 2001 are decent, but they have pretty big issues. Also have a backup blend for high solar activity. But not sure I'll need it yet.

Really not buying this La Nina as a big deal at the surface. It's getting late fast for it to catch up to last year. To be fair, last year was one of the ten coldest events in Oct-Nov before it weakened. But it was a 25.58C La Nina in Dec-Feb. At the end of the day, I consider 25.50-26.00C La Ninas in winter to be weak. 25.0-25.5 is moderate. <25.0 is strong.

For 1950-2020, there is no winter in Nino 3.4 more than 1.0C colder than September. Not one. Fall can be pretty cold in that zone like last year. But Nino 3.4 is going to be warmer at the surface in September than last year when the monthly data is in. It's probably around 26.20-26.40C on the monthly data for September. The amounts are not trivially small or localized warming either at the surface - As far as ACE goes, post Sam, I'm expecting another 20 to 40 more ACE by the end of the season. Sam is going to be the last gunslinger for a little bit. We have maybe 45 days left before the potential in the Atlantic ends. Each day without a storm lowers the ceiling on the season fast. Also, since 1850, ACE has topped 70 like three times in Oct-Nov. So 200 is a real stretch for me. It took a cat 4 and cat 5 in November just to hit 180 last year. In 2017, for comparison, we had 175 ACE in September alone.

1983 is an example of a year that under-performed the subsurface though at the surface. 26.0C in Nino 3.4 in DJF 1983-84. Despite a -2.25 subsurface reading in October. A lot of that is washing away the 1982-83 El Nino heat. But you have a similar thing now - the heat from the March-June warm up below the surface has to get washed out. The La Nina last year was also weaker than 2010-11. So even though the subsurface is tracking like a colder 2011-12 event, the surface is tracking like a warmer 2011-12.

No real change at the surface from August-September. The subsurface in 2011 is an excellent match to 2021 for January-September. So it's likely the subsurface will stop cooling and flip to warming in October or early November, like 2011. This is the subsurface (100-180W, 0-300m down) comparison. Jan Feb Mar Apr May Jun Jul Aug Sept2011 -1.27 -0.22 0.50 0.58 0.47 0.39 0.06 -0.54 -1.012021 -1.02 -0.83 0.27 0.60 0.65 0.31 -0.40 -0.83 -1.27 October 2011 dropped to -1.26 in October, then reversed in November. It cooled again briefly in Dec-Jan, but October was the peak.

September finished slightly wetter than average here with the rains yesterday. Tentative precipitation analogs for Albuquerque, based on July-Sept objective matches are 1950, 1954, 1974, 1983, 2004 as a blend. That's a pretty mixed period for Oct-May - with some dry and some wet months. Matching on the wet September definitely moved the matches to wetter for Oct-May. The GFS still has a major low south of Kamchatka on 10/2 in the evening. Have to watch 10/19-10/23 in our area for something pretty big. That's actually pretty close to one of the recurring wet-periods locally (10/18 would be 46 days or so after the remnants of Nora, which followed the mid July and early June rains, and so on). The rain yesterday also followed at a 46-day lag from the mid-Aug and early July and late May rains here. The WPO doesn't flip phases super often, so could be in for a good pattern for a while. The historic snows last October down to Mexico were also fairly close to the 10/19-10/23 time-frame, but the setup looks pretty different even this far out.

The subsurface for 100-180W at the equator, 0-300m fell to -1.27 in September. That's colder than I expected. The simplest recent match that looks like the Canadian outlook is 2007, 2010, 2011, 2011 for July-Sept subsurface readings. (year, july, aug, sept) 2011 +0.06 -0.54 -1.01 2011 +0.06 -0.54 -1.01 2010 -1.36 -1.74 -1.93 2007 -0.48 -0.68 -1.03 Avg -0.43 -0.88 -1.25 2021 -0.40 -0.83 -1.27 The correlation for 1979-2020 September subsurface readings and winter SSTs in Nino 3.4 is very strong. Near 0.8 for r-squared. But it is still highly variable. When I ran it using the -1.27 figure, I got 25.54C as the winter estimate. But...the errors historically imply it's 25.54C, +/-0.75C at 90% confidence in practice. In other words, running "warm" of the 25.54C is probably feasible. Last year's actual winter reading ran "cool" of the 25.80C the September subsurface projected as an example. The average error from plotting September subsurface v. Nino 3.4 at the surface in winter is plus or minus 0.4C. Solar activity was over 51 sunspots for September - highest monthly since March 2016. So far, we're off to a devastating start for snow.

Don't think the new Canadian has the right strength (too cold) for the La Nina, but it does have a real -PDO look. Way more than the last run did. The new temperature outlook for October is pretty close to the winter analogs I have with a +WPO - I did add one degree because the average time is 1988.