raindancewx

Here is an interesting thought for the winter: When is the last time, two same sign ENSO events produced the same outcome in the winter? In a row I mean. ENSO: 2009 and 2014? No. 2014 and 2015? No 2015 and 2018? 2018 and 2019? 2006 and 2009? 2004 and 2006? 2012 and 2013? 2003 and 2012? 2017 and 2016? 2016 and 2011? 2011 and 2010? 2010 and 2008? 2007 and 2008? 2005 and 2007? 2000 and 2005? My point with this is that 2017-2018 was a very hot/dry Western winter. So expecting the same outcome again is probably wrong, despite what CPC/IRI have.

I went with 2012 and 2007 as the main analogs in my forecast because those events were colder east in the Nino zones in Fall, and they have very low sea ice. But the cold in the Nino zones spreads out later on. The cold core below the surface has been about 140W for a while, so it doesn't really fit into any of the three blends listed above. My sense is the average departures for the four zones in winter may be pretty consistent against the 1951-2010 averages, probably around -1.0 overall, despite the possibility of a colder peak than that between now and December in Nino 3.4. I'm a believer in some form of the storm depicted for the NE later in the month - I've been expecting that for a while. I lean toward it being a strong Nor'easter that originates over relatively warm water rather than a hurricane or Perfect Storm type of deal. SOI - still amazingly similar for this brief negative period. 2007 has not been a good match for October so far, but to me that's an MJO thing, not because the La Nina is that different. Oct 12-15 10.7, -0.2, -5.5, -6.3 in 2007 Oct 13-16 10.3, -2.3, -7.9, -6.1 in 2020 The coming period with the cold dump in the middle of the country is similar to October 2012. Several of the cold-ENSO years following two El Nino winters are also pretty similar to this October so far - notably 1978 and 1988. 1959, another cold ENSO following two El Nino winters is also showing up a lot in the CPC 6-10 and 8-14 analogs. October may end up looking a lot like a blend of 1978, 1988, and 2016, all years following two-cold ENSO years. The model depictions with Montana and other areas of the NW, 5-10 above normal in October look wrong now with the highs in the 30s in the coming day for places like Billings (88 a few days ago - I live in a desert and we haven't topped 87 here this month for reference). This is my forecast for anyone who missed it. So far, close to 300 people have read it. Pretty happy with that actually. https://www.scribd.com/document/479516526/Winter-2020-21-Outlook,

What was up with the weird sexual comments about clouds earlier? I know it was you Chinook... It's not the 9/9 storm, but pretty impressive cool down in Billings from 88 earlier in the month to snow this week.

Pretty healthy La Nina with the warm version of the -PDO in September. I'd really like to see more of the North Pacific cool off with this event. This is probably the best shot at that for the next few years. The AMO actually cooled a fair bit in September, still very high AMO value though. We're almost in a warmed up -AMO, -PDO configuration, if you assume the oceans would be ~0.2C colder than now without the slow warming since 1981-2010. The Atlantic would look pretty cold with that extra 0.2C gone.

The NAO thing I use is designed on the assumption that the changes are random, and so it should be hard to find a good match to any given set of changes. What I do find is, if the NAO has several years with similar changes, the blends work much better if the prior Nino 3.4 SST is the same. So if you just looked at 2015-16, you'd see 1968 is an amazingly close match in April to May, and March to September, in terms of how the NAO changed. But, 2015 followed an El Nino and 1968 followed a near La Nina - so the NAO match fails. This is one reason I weight ENSO order pretty highly when I analog stuff. Here is 2015: 2015 1.79 1.32 1.45 0.73 0.15 -0.07 -3.18 -0.76 -0.65 0.44 1.74 2.24 1968 0.13 -1.29 0.40 -1.08 -1.76 0.33 -0.80 -0.66 -1.92 -2.30 -0.93 -1.40 So you've got near identical March to September and April to May NAO changes. That combo doesn't exist in very many years. But the 2015-16 winter NAO was very positive, and 1968-69 was very negative. So a blend of 1992/1994, which has near identical changes to 1968 in the NAO, but nearly matches the prior Nino 3.4 reading, produces a much better outcome. 1992 -0.13 1.07 0.87 1.86 2.63 0.20 0.16 0.85 -0.44 -1.76 1.19 0.47 1994 1.04 0.46 1.26 1.14 -0.57 1.52 1.31 0.38 -1.32 -0.97 0.64 2.02 1992: -1.31 (Sept-Mar), 0.77 (May-Apr) Nino 3.4 prior (26.76C) 1994: -2.58 (Sept-Mar), -1.71 (May-Apr) Nino 3.4 prior (27.64C) Blend: -1.95 (Sept-Mar), -0.47 (May-Apr) Nino 3.4 prior (27.20C) --> DJF NAO of +1.11 2015: -2.10 (Sept-Mar), -0.58 (May-Apr) Nino 3.4 prior (27.18C) --> DJF NAO of +1.31 Basically, you have to match on the two change periods first, but it works better if you get the prior Nino 3.4 reading to match. For 1950-2019, the r-squared for March NAO to the following DJF is about 0.10 - it's not too strong. None of the months are individually, but the two changes in combination are actually pretty hard to replicate even with 70 years of data, you usually end up with less than three good matches. My rule is no match should take me more than five minutes to make by eye-balling the data. The data also relies on only the monthly NAO readings from 1950-2019. So pre-1995, there really isn't enough data to get good matches, since I only use years that would have been available and known to make the NAO matches. Not every year can be replicated super well with the SST prior matching. But I find it gets the NAO sign (neutral, positive, negative) right almost all years, and is +/-0.4 for the seasonal DJF value 18/25 times since 1995. The average error is about 0.36 for DJF, which is somewhat better (~40%?) than guessing the 1995-2019 NAO winter value each year.

For what its worth: SOI - Oct 12-15 10.7, -0.2, -5.5, -6.3 in 2007 Oct 13-16 10.3, -2.3, -7.9, in 2020 Really close. In 2007, southern CO had precip 10-days after the crash. ABQ did not. But you have almost identical magnitude/timing for the crash if nothing else

I'm not sure that the lack of relationship since 1990 is a climate change thing though - I think it could conceivably revert back. But the Typhoon Tip guy had asked about if it changed in more recent years. I don't use October to try to predict the NAO. I've had much better success using the change from April to May and March to September as a blend to predict the NAO in winter. March to September is probably indirectly an indicator of the total sea ice extent change, and April to May is just a decent indicator for whatever reason. I'm pretty sure there is academic literature supporting May as a predictive period for the NAO though, I think that's where I got that part of it.

I posted the graphs in the New England thread for anyone who wants to see, but there actually is a real relationship between the Oct NAO and the DJF NAO if you are in a winter following an El Nino. It is a weak relationship, and it vanishes since 1990. But it is there if you use the 24 winters following an El Nino from 1950 to 2019. In the winters not following an El Nino, the relationship is very near 0 in a correlation sense. Long-term, in an October after an El Nino, the correlation (r-squared) to winter is close to 0.2. Probably real, but not precise enough to be useful. Believe it or not, the Spring months have stronger r-squared relationships to the DJF NAO, which is why it is interesting we had the first -NAO April in forever in 2020.

Going into this winter I developed some experimental indicators to predict the timing of La Nina cold shots well in Albuquerque. The two strongest are: ACE Index in La Nina predicts the mid-Dec to mid-Jan high pretty well (+/-3.4 at 82%) in Albuquerque. The current ACE index favors a 47.2F mid-Dec to mid-Jan period (average is 46). My analogs are somewhat colder than 47 but within the 3F range. The timing of the first 90F high in May or June predicts the Nov-Jan high pretty well too. The earlier the first 90F high is, the colder Nov-Jan is in La Ninas. This implies Nov-Jan should be about 1.5F below average, +/-2.6F at 85%. I went with a pretty warm November here. That's either spectacularly wrong, or it's going to be frigid in December or January if the formula and analogs are right.

The Jamstec has in October 2020 what it forecast in October 2016 for 2016-17, and that ended up a cold NW winter, which I take as a decent sign for my outlook. I did look - the Oct correlation for the NAO v. DJF disappears even on the "after El Nino years" thing I did before since 1990 or 2000, although it is less than 10 years since 2000 and not a good way to use correlations at that point. Once the coming cold dumps into the NW, this October will probably to look a lot like the composite of some of the prior Octobers following two El Ninos ahead of a cold ENSO winter. April/May are actually better at predicting the NAO in winter than October is, despite how far out they are. We had a -NAO in April this year for the first time in forever, so that makes it unlikely to me that recent years would be super similar.

Date Tahiti (hPa) Darwin (hPa) Daily Contribution 30 day Av. SOI 90 day Av. SOI 14 Oct 2020 1012.74 1010.20 -2.26 12.44 8.84 13 Oct 2020 1014.49 1010.00 10.31 12.58 8.77 12 Oct 2020 1014.54 1009.90 11.28 12.34 8.52 11 Oct 2020 1014.50 1009.75 11.99 12.18 8.26 10 Oct 2020 1015.51 1010.40 14.31 11.82 8.15 Yay. Euro/GFS now have some rain down here sometime around 10/24. This is similar to when the dry spell in Fall 2007 broke up here, if it verifies. The Jamstec has a very warm/dry winter nationally. Still not 100% on the strong La Nina. Then it has an El Nino in 2022. The forecast for the US is actually pretty similar to what it forecast in 2016, when it was actually quite cold in the NW.

https://oceanview.pfeg.noaa.gov/erddap/tabledap/cciea_OC_PDO.html Nate Mantua / JISAO PDO fell off more in September. January -0.23 February -0.68 March -0.82 April -0.57 May 0.09 June -0.08 July -0.38 August -0.28 September -0.70

Your snowfall in NYC is pretty tied to the ACE Index in La Nina years. The MJO has different wavelengths and magnitudes each year, and a lot of La Ninas won't have any substantial MJO waves. Below 150 ACE in a La Nina, NYC is 2 for 17 at topping 30" of snow, which is what I'd consider a "good" season for NYC, instead of just near average, in the 1931-2019 period. The 2016-17 La Nina was arguably a Neutral and and had 146 ACE, so it hardly counts (30.2"), and 1938-39 isn't really a likely outcome for the winter. At the current 123 ACE, you'd expect 25 inches of snow or so, based on the 10 closest ACE years. I do expect the ACE to go up a bit more, but probably not above 150. The 10 La Ninas above 150 ACE since 1931, feature five snow years above 40" in NYC, compared to 0/17 for the under 150 years..

I've been doing more research on the cold-ENSO winters following two El Nino winters in a row. Pretty severe cold likes to show up in the West around 12/16-1/15. 1931 is similar. The 1978, 1988, 2016, 2016 blend of Octobers looks a bit like how this October may finish after the cold dumps into the NW too.

The NAO Oct v. Dec-Feb NAO thing actually is pretty unreliable most of the time. It's only in the Octobers following an El Nino winter that it becomes somewhat reliable as a negatively correlated indicator for winter. It's still only like r-squared of 0.2 though. But in the other years it's completely useless. It's why I don't use it in my main NAO prediction method. It has ~weak skill in years following El Nino if you're charitable. It's probably fairly safe this year since the October NAO will likely be -1 or lower.

These guys have a pretty good track record. They like similar snow in the West to last year. That'd be ~5-8 inches at the airport here, and 8-12 for most of the city. My raw analogs had 4.8" for Albuquerque, but my statistical regression for snow had 8.3". So I went with 7", but 5-8" seems pretty reasonable. Albuquerque can do well in a strong La Nina, 1988, 1973, were good here. But it's more likely we finish below average like 2007 or 2010. GFS also does kind of seem to see something around 10/24...fingers crossed.

Nino1+2 Nino3 Nino34 Nino4 Week SST SSTA SST SSTA SST SSTA SST SSTA 02SEP2020 18.9-1.6 23.6-1.3 25.8-0.9 28.4-0.3 09SEP2020 19.5-1.0 23.4-1.4 25.7-1.0 28.5-0.2 16SEP2020 20.0-0.4 23.6-1.3 25.9-0.8 28.2-0.4 23SEP2020 19.6-0.8 23.6-1.2 25.7-1.0 28.1-0.5 30SEP2020 20.1-0.5 23.8-1.1 25.6-1.1 27.9-0.7 07OCT2020 19.5-1.2 23.4-1.5 25.5-1.2 27.8-0.8 Still like a blend of 1995/2007 on the weeklies. The 1999/2011 events are close at the moment too. Way stronger/colder than 2017, 2005, 2008 at this point. Somewhat stronger than 2016 at this point, but that event peaked in November. Way warmer than 2010 still. 10OCT2007 18.8-1.9 23.3-1.6 24.9-1.8 27.7-1.0 04OCT1995 19.6-1.0 23.9-1.0 25.6-1.1 28.1-0.5 The subsurface actually looks like it is plateauing. Fall 2012 actually looks pretty similar to this Fall so far too, which makes me wonder if the low-sea ice cold-ENSO signature is showing itself again. The Sandy like system depicted to hit the NE in the BS range on the GFS is kind of a hint at that too. Pretty sure Canada had a late hit in 2011 too. These are the closest SST matches for September for the four Nino zones. I put the temperature pattern locally to the right of the matches. You can see it gets pretty cold in Dec/Jan in a lot of them. Colder air is coming into the NW in the coming days, ala 2007 and various others, despite arriving a bit late. But a lot of these years actually do have the hot West October: 1978, 1988, 1952, 1999, 1950/1959/1966/1967 sort of have it but without the expansiveness/severity of the heat. It is interesting to me to see 1959, 1978, and 1988 show up - those are three of the only cold ENSO years in the last 90 years to follow two El Nino winters in a row.

The UK October output is essentially what I forecast. In the SW you tend to get ~near constant incoming precipitation divided by outgoing precipitation in each ENSO state if you calculate it annually on the Koppen classification scale. So the 1999 Summer was very cold here, and wet. Something like 40 days >=90F, compared to 90 or so this year. The following winter was hot and very dry to end up at the ~La Nina constant. I'd expect a pretty different outcome from that year given the much hotter/drier summer. 2011-12 would be interesting but I don't think it's a good match for the eastern US. I'm not a fan of 1998 either, mainly because it is a ~4C drop in Nino 3.4 from a Super El Nino, with high solar activity, and the start of a three year La Nina.

The Euro is showing a brief pressure pattern reversal by Darwin, Australia and Tahiti in two days. Relatively high pressure by Darwin and relatively low pressure should crash the SOI. Would likely dump a storm into the SW later in October. The rule is, ten days after the SOI crashes by 10 points in one day, there is a low between Los Angeles and Dallas, somewhere between 30-38N. It works around 90% of the time locally in Oct-May, and then doesn't work at all once the highs prevent any storms from entering this part of the country in Summer. The new Jamstec should be out soon. Some of the models have fairly high amplitude MJO phase five now for a while. Hope we see that again in Nov, Dec, Jan - can be quite cold in the SW later on.

The Euro has been showing a brief period with high pressure over Darwin and low pressure over Tahiti in a couple days. That'll probably crash the SOI pretty hard. So that'll likely put a storm over the SW sometime around 10/24. That time frame has been showing up in some of the methods I like as the approximate end to the dry spell here.

Your area was kind of weird when I did my winter forecast. St. Louis did pretty well for snow, but areas closer to the Missouri/Arkansas border really got hosed for seasonal totals. Really sharp cut off. Same thing in the NE. NYC did OK, but Philly and south pretty bad. I had 1995, 2003, 2007, 2012, 2019 as analogs, but with 2007 at five weight and 2012 at double weight. Most of the indicators I can think of in La Nina favor a cold West / warm East look, but it's strongest for the NW/SE. One thing I'd watch is that systems sometimes tend to dive unusually far south at the exact spot where hurricanes have hit. So the obvious spot for that this year would be Louisiana...and places north of Louisiana. That kind of showed up in the analogs too, with Texas and Florida much hotter than the Central Gulf Coast states. I think you had unusual snow events in New Orleans, Houston, etc. in years like 2005, 2008, and 2017, but I'd have to look.

That UKMET run is almost identical to what I had in my forecast. Makes me a little nervous because I don't really buy the models having skill this far out. Warm/Wet is the idea for New England though. Most places seem to have little trouble getting to their seasonal snow numbers in the US if they get snow every year and are less than +2F v. their long-term average highs I find, with above average precipitation. For truly cold locations, it can be even warmer than +2F if the highs are still predominantly in the mid-30s or colder it seems like. One thing I noticed locally - there is at least some notable difference in precipitation timing from some of the big hurricane (ACE) years already. The past month has had no rain here, but you had pretty major precipitation events 9/10-10/10 in 1995, 2005, 2017, before getting essentially nothing after for eight months. Dry periods in the past month do show up in my analogs though. Of the recent three La Nina super hurricane seasons, 1995 is closest here, so the dry period started later September than this year, but you had a big system around 9/9 like this year. The 2005 and 2017 9/10-10/10 periods both had huge rain events late September, completely missing this year, one of the reasons I dislike those years (~1.5-2.0" rain 9/28 or so in both). The 2010 La Nina also had a big late September rain event here, but it had much lower ACE than the other three, and far more precipitation too.

https://www.scribd.com/document/479516526/Winter-2020-21-Outlook That's why I have for the winter. The gist of it is: +NAO La Nina - but likely negative in Dec/Mar - either briefly but severely, or weakly, but persistently. A blend of 1961, 2003, 2007 is damn close on the predictive periods I use to get NAO values in winter, and very close in the four Nino zones, and in the four nino zone transitions y/y. Also, in the 24 winters following an El Nino since 1950, the winter NAO tends to behave opposite of the October NAO pattern. The -NAO pattern October now favors a +NAO winter. As for the look of winter, most of the things I can think of that should be independent point to something less amplified but similar to 2016-17. +NAO La Nina winters (>0.4 DJF) in aggregate are cold West / warm East. It's more reliable NW/SE, it's especially true if you throw out the super hurricane seasons like 2017. All cold-ENSO events following two El Ninos in the last 90 years follow the above pattern to at least some extent. These are 1931, 1959, 1970, 1978, 1988, 2016 as a blend. I threw out 1942 and 2005, with different looks because they follow three El Nino winters, not two like the others. If the models are right, and it is a strong La Nina, the blend of the sub 25.0C DJF La Ninas is also cold West/warm East (1973, 1975, 1988, 1999, 2007), with the La Ninas still following the patterns - highest ACE year (1999) is the hottest, and the La Ninas following El Ninos are much better for the West (1973, 1988, 2007). La Ninas following an 80-160 ACE hurricane season tend to be cold NW, warm SE, again mixed SW/NE. The West will roast generally in the highest ACE La Ninas: 1995, 2005, 2017, etc. La Ninas following El Ninos are all mild-cold - somewhere - North of a San Francisco to NE/SD line, in each of the last 10 times that scenario has happened. This is more of hot TX pattern than a hot SE or SW pattern. The lowest sea-ice cold ENSO winters starting 2007 are all cold in the West, while the higher (relatively) sea-ice cold ENSO years are all cold in the East. This is 2007, 2012, 2016 (2020?) for the West, and 2008, 2010, 2013, 2017 for the East, with 2011 on the border. The threshold seems to be 4.3 million square km for the minimal sea-ice extent in September. We were 3.7 million this year. The +QBO La Nina idea does not support the above ideas. But the sample is so small, that I'm not real worried about it. Especially since rising QBO from Summer to Winter with a La Nina does appear to gel fairly well with a cold NW idea. The warm October in the NW is fairly consistent with years like 1988 as well, despite being rare in a La Nina.

https://www.scribd.com/document/479516526/Winter-2020-21-Outlook That's my winter outlook. Finally finished updating and caught most-all of the typos. It's a much colder winter than last year overall, but it's still not "cold" for many people. I have fairly limited areas wetter than average too. I think the PDO will fluctuate a lot, but should be primarily pretty negative from Nov-Apr with how cold Nino 1.2 is. What we're seeing now is some cooling east of Japan (toward a +PDO), but also cooling by NW North America (toward a -PDO). You should at least see continue cooling by NW North America if the usual build up of cold in a La Nina happens in Western Canada.

Here it is: My 2020-21 Winter Outlook for anyone who cares. https://www.scribd.com/document/479516526/Winter-2020-21-Outlook My winter look is ultimately not that different from: a) Cold-ENSO years following two El Nino winters. Since 1930, these are 1931, 1959, 1970, 1978, 2016 as a blend. The 1942 and 2005 La Ninas don't "look" like the others, but I think it is because they followed THREE El Ninos, and not two. b) Cold ENSO years since 2007 with very low-sea ice. The cut off seems to be <4.3 million square km sea-ice extent in September. c) Strong La Ninas. The strongest La Ninas (<25.0C Nino 3.4 DJF) are 1973, 1975, 1988, 1999, 2007. The blend of those years is similar to A & B. Don't know that we get that strong, but it doesn't really impact directly. The crappier strong La Ninas still follow the other patterns - 1999 is the hottest and highest ACE for instance. The years after El Ninos are generally better in the West, and so on, whereas 1975/1999 followed La Ninas. d) +NAO La Nina winter composite. I find if a winter follows an El Nino winter, you want the NAO to be positive in October for a -NAO winter. If you have a -NAO October, you tend to get a +NAO winter. That said, I think you'll see the NAO go negative for at least a time in December and March. e) Middling ACE years. The middle ACE years in La Ninas tend to be colder in the West, hotter in the East.