raindancewx

From what I remember, if Boston had actually gotten 3-6" or whatever it was last November at the official site, instead of nearly nothing, I would have been within 2-4" of what I had forecast for Boston. If you think it's 40" instead of 27", I'm just low instead of high by a few inches last year, so it doesn't really matter to me. Looking at the forecasts of others, I do think it would be prudent to find predictive variables that are objectively known ahead of winter for precip/temps/snow, other than say, ENSO or the PDO or whatever. Something like a precip pattern in Summer. You guys really live or die with the teleconnection indexes - AO, NAO, PNA, EPO - etc. You can probably predict...most of them correctly in any winter, but all is a tall order. My stuff is designed to be something like 1/3 data mining, 1/3 statistics, 1/3 global patterns. I find that there are like a 100 ways to match on any blend of ocean/atmospheric/solar variables, but it's much harder to have those match and get your weather to match day and date for an extended period - so that's what I try to do. Matching the highs/precip patterns for a long-time with the oceans/solar stuff seems to be a way to lock onto the MJO signal for a few months if nothing else.

Right is for a -4 to -12 SOI Sept-Nov with a 0 to -1 Oct PDO. Left is El Nino winter + an Oct PDO of 0 to -1.

PDO in October 2019 went the most negative it has been in six years. https://oceanview.pfeg.noaa.gov/erddap/tabledap/cciea_OC_PDO.htmlTable?time,PDO There are no cases of the PDO topping +0.5 from Nov-Apr after a reading of 0 to -1 in October for the 1931-2018 period. PDO was -0.45 in October 2019. El Nino October PDO Nov-Apr 1951 -0.32 -1.02 1963 -0.52 -0.88 1965 -0.36 -0.44 1968 -0.34 -0.81 1977 -0.61 0.50 2004 -0.11 0.47 2006 -0.05 -0.04 Similar SOI matches (-4 to -12 SOI in Sept-Nov) with a similar October PDO - Year Oct SOI SON 1932 -0.29 -5.7 1944 -0.40 -4.0 1946 -0.36 -9.0 1951 -0.32 -10.7 1963 -0.52 -9.9 1977 -0.61 -12.0 1990 -0.69 -4.5 2004 -0.11 -4.6 2006 -0.05 -7.2

Just for fun, since it is pretty rare - El Nino. PDO in October of 0.00 to -1.00 on the JISAO index (it was -0.45). https://oceanview.pfeg.noaa.gov/erddap/tabledap/cciea_OC_PDO.htmlTable?time,PDO El Nino October Nov-Apr 1951 -0.32 -1.02 1963 -0.52 -0.88 1965 -0.36 -0.44 1968 -0.34 -0.81 1977 -0.61 0.50 2004 -0.11 0.47 2006 -0.05 -0.04 Some very cold winters in the Southwest in that mix. 1963-64 is the coldest winter in 120+ years in New Mexico. 1965 is right behind it. 2006 is very cold too. 2004 and 2006 are also extremely wet. If you look at all October PDO values, 1931-2018, that were 0 to -1, November to April PDO values finish between +0.5 and -1.78. There are no strongly positive PDO cases. The years right around -0.45 in October tend to see the PDO stay between -0.5 and 0.0 for Nov-Apr. Overall, 18/24 cases of the PDO between 0 and -1 in Oct saw the PDO stay below average from Nov-Apr after an October value of -1 to 0. The Neutral/Negative PDO + El Nino composite is a cold West composite - we'll have to see what happens.

Nate Mantua sent out the PDO data for October 2019: https://oceanview.pfeg.noaa.gov/erddap/tabledap/cciea_OC_PDO.htmlTable?time,PDO So...lowest PDO value since 2013 in any month. -PDO is strongly correlated to a cold NW/warm SE in winter if it continues. 2019-01-01T00:00:00Z 0.66 2019-02-01T00:00:00Z 0.46 2019-03-01T00:00:00Z 0.37 2019-04-01T00:00:00Z 1.07 2019-05-01T00:00:00Z 1.03 2019-06-01T00:00:00Z 1.09 2019-07-01T00:00:00Z 1.03 2019-08-01T00:00:00Z 0.38 2019-09-01T00:00:00Z 0.41 2019-10-01T00:00:00Z -0.45 2013** -0.13 -0.43 -0.63 -0.16 0.08 -0.78 -1.25 -1.04 -0.48 -0.87 -0.11 -0.41 2014** 0.30 0.38 0.97 1.13 1.80 0.82 0.70 0.67 1.08 1.49 1.72 2.51 2015** 2.45 2.30 2.00 1.44 1.20 1.54 1.84 1.56 1.94 1.47 0.86 1.01 2016** 1.53 1.75 2.40 2.62 2.35 2.03 1.25 0.52 0.45 0.56 1.88 1.17 2017** 0.77 0.70 0.74 1.12 0.88 0.79 0.10 0.09 0.32 0.05 0.15 0.50 2018** 0.70 0.37 -0.05 0.11 0.11 -0.04 0.11 0.18 0.09 0.26 -0.05 0.52

Your percentage listed for Boston is 80-27/80. Comparing the error to the theoretical total (80") doesn't make any sense. If I told you your kid was going to be 8 feet tall and the kid ended up 3 feet tall would you call that a 60% error? That's what you did. It's closer to a three fold error. I had 35" in my blend last year, and even that is a 30% error for Boston if you use real math. You picked average snow this year, so mathematically you don't have to worry anyway, it's more or less impossible to be as wrong as last year for Boston, so in that sense at least I'm sure you have a much better forecast. I don't really care what people forecast, but if you are going to grade yourself, you need to do (expected-actual)/(actual). You did (expected-actual)/(expected) which doesn't mean anything.

It's not perfect, especially in the middle period, but people not including 2018-19 as an analog to this winter are a bit nuts to me. Oct/Nov will probably be ~0.2C different in Nino 3.4? Close to a rounding error. I think the pretty similar to last year honestly, just shifted somewhat east - that's the certainly the impression from the Fall. Here is 2018 (10/01-10/15, 10/16-10/31, 11/01-11/11): Here is 2019 (10/01-10/15, 10/16-10/31, 11/01-11/11):

Take it or leave it but - 1) When you evaluate your forecasts, you should use (observed-forecast)/(observed) as your metric. You somehow have last year's call for 80"+ as only a 60-something percent error, when Boston only had 27". The math should be (80-27)/(27), i.e. your forecast was around +200% (triple) observed. Observed is the anchor, not the forecast. 2) I suspect December will be a lot warmer than you have in the North/East. Nino 4 is around 1.0C warmer than your composite. Even years with a lot of blocking aren't that cold in December when Nino 4 is that warm. My composite in my forecast had Nino 4 too cold too, so I mentioned I expected the East and North to be 2-3F warmer than depicted by the raw analogs. 3) It does amuse me that despite pretty different background states in some ways, you essentially forecast the same thing nationally two years in a row. Doesn't mean your wrong this time, just amuses me. 4) A lot of your very cold analogs for the East had much higher solar than this year: 1958, 1969, 1989, 2014. I think 1958 and 1989 are literally around 255 sunspots/month and 200 sunspots/month for July-June. This July-June is probably...3? Maybe 4? If you take out those four years, the cold signal is gone in December for the East.

My winter analogs included several years that featured snows in the Southwest around November 20th, way back when I did my winter forecast in early October. I weighted the years 1953 (x2), 1983 (x2), 1992, 1995, 2009 (x3), 2018. So assumed it was close to 50/50 we'd get snow here around 11/20. Looks promising so far on the models. https://www.scribd.com/document/429712619/Winter-2019-20-Outlook The Raw Snow Data: 2018 6 (Dec 2, Dec 26, Dec 27, Jan 1, Feb 19, Feb 22) 2009 11 (Oct 29, Dec 8, Dec 23, Jan 22, Jan 23, Feb 3, Feb 4, Feb 7, Mar 11, Mar 20, Mar 24) 1995 7 (Dec 17, Dec 18, Dec 31, Jan 1, Feb 1, Feb 2, Mar 6) 1992 13 (Nov 4, Nov 20, Nov 21, Dec 4, Dec 5, Dec 12, Dec 15, Dec 16, Dec 19, Jan 13, Feb 15, Feb 28, Mar 1) 1983 7 (Nov 19, Nov 26, Dec 25, Jan 14, Jan 16, Mar 18, Apr 26) 1953 9 (Nov 18, Nov 19, Nov 20, Nov 21, Dec 5, Dec 11, Jan 12, Jan 13, Jan 20) Snow Windows Albuquerque: Nov 18-21 Dec 2-5 Dec 15-19 Dec 25-Jan 1 Jan 12-16 Feb 1-4 Mar 18-20 The SOI support is there too. Over a 10 point drop from 11/8 to 11/10. Should put a big system in the SW US or Mexico around 11/20 12 Nov 2019 1011.94 1010.50 -9.17 -5.76 -8.39 11 Nov 2019 1011.42 1010.05 -9.61 -5.18 -8.48 10 Nov 2019 1011.08 1010.50 -14.64 -4.71 -8.51 9 Nov 2019 1011.85 1010.70 -11.01 -4.37 -8.38 8 Nov 2019 1013.13 1010.25 -0.00 -4.38 -8.23 7 Nov 2019 1013.77 1008.40 15.84 -4.69 -8.01

With the temperature down to 34F at 6 pm and low dew points, Albuquerque should be up to 11 nights of 32F or less since October 1 on 11/11. Tied with 1991 for fourth most since 1931 (10/1-11/11 lows of 32F or less).

Models had shown Nino 4 cooling. It has a bit. If that continues, a colder December is possible nationally. https://www.cpc.ncep.noaa.gov/data/indices/wksst8110.for Nino1+2 Nino3 Nino34 Nino4 Week SST SSTA SST SSTA SST SSTA SST SSTA 25SEP2019 20.0-0.5 24.8-0.1 27.2 0.5 29.7 1.1 02OCT2019 20.0-0.6 25.1 0.3 27.2 0.5 29.7 1.0 09OCT2019 19.7-1.0 24.8-0.1 27.1 0.4 29.5 0.9 16OCT2019 20.6-0.2 25.3 0.4 27.5 0.8 29.7 1.1 23OCT2019 19.7-1.3 25.0 0.1 27.3 0.6 29.7 1.0 30OCT2019 20.8-0.4 25.4 0.5 27.4 0.7 29.6 0.9 06NOV2019 20.7-0.6 25.2 0.3 27.2 0.5 29.3 0.7 26SEP2018 20.2-0.3 25.5 0.6 27.3 0.6 29.3 0.6 03OCT2018 21.3 0.7 25.6 0.7 27.4 0.7 29.5 0.8 10OCT2018 21.1 0.4 25.6 0.7 27.3 0.6 29.5 0.9 17OCT2018 21.1 0.3 25.9 1.0 27.6 0.9 29.6 0.9 24OCT2018 21.3 0.3 25.9 1.0 27.7 1.1 29.8 1.1 31OCT2018 21.4 0.2 25.9 0.9 27.8 1.2 29.8 1.1 07NOV2018 22.1 0.8 25.8 0.9 27.4 0.8 29.5 0.9

This is going to verify...in Mexico, and no it isn't relatively common because of the elevation.

In Albuquerque, Nino 3.4 temps / solar conditions / Nino 3.4 temps in the prior winter are pretty predictive for highs in Oct-May. Once October is in you, you can anchor the data to what was observed then too. Last year it was a strong match in all non-March months. This year, I have a decent match for the sun, and ONI (ONI will probably be around 27.1C) and a good match for October. But I don't have a good way of getting ONIp and the sun to work together. But so far, the blend is holding in November. Will be curious to see how it goes.

Latest European run has Nino 4 cooling, Nino 3.4 and 3 warming through November, then cooling, and then Nino 1.2 warming into January before cooling after. It's kind of a "warmth drains East" look. The dashed line tends to verify at the extreme end of the sea surface temperature plume in most months, so I think it's fairly safe that November and December at least see Nino 3.4 above El Nino thresholds. There isn't much warmth below the surface after that though, so some cooling is likely. This event probably won't count as an El Nino by CPC, but it probably is going to count by the Jamstec definition (165E-140W) definition. By my older thresholds (1951-2010), it will probably count, since 26.5C is the average for me I use for Oct-Feb anomalies, while CPC is higher in Fall. The SOI response, right around -8 for the past 90-days, is also consistent with a weak El Nino Modoki setup.

Don't look now, but with the MJO different, 2013 now once again looks completely different to 2019 in November. It doesn't look like 2018 or 2014 either. Passing resemblance to 1951, 1967, 2012 though.

Today looks a lot like the November MJO composite for phase six, with the warmth in Billings, Denver, Kansas City, etc, but the East Coast cold, and the heat struggling to push into New Mexico. The push into 7-8-1 later in the month should warm up the East quite a bit - this month probably won't finish anywhere as chilly as it looks right now. A lot of the US is 6F or more below average in November to date.

This event may not last too much longer. I think October will eventually be warmed up on the CPC data to 27.3C or so. After that, Nov-Dec look El Nino warm, safely. Jan/Feb/Mar still kind of iffy. Heat is only down to 100m. I'd say we are about due for a) a Strong La Nina and b) a flat Neutral (26.5C) - will be interesting what happens if the warmth doesn't get replenished. NOAA PDO crashed in October. Will be curious to see the JISAO/Mantua value for October. https://www.ncdc.noaa.gov/teleconnections/pdo/ 201907 0.42 201908 -0.16 201909 0.03 201910 -0.81

I don't "expect" my winter analogs to work in the neighboring seasons, but I do like to see relatively close matches to make sure I am on the right track. I had the cold West idea for October in the winter analogs (the blend is from 10/1, although I didn't release it until 10/10), with the cold shifting East for November. Not bad. The pattern was more amplified than my analogs, probably in part because the PDO went more negative than I had it. I'll never understand why New Mexico never shows up cold on these maps. Albuquerque was 2.7F below the 1951-2010 average high in October 2019 (68.3F v. 71.0F).

MJO looks about two weeks faster to reach phase six in November than last year. Will be very interesting to see if it stalls in Phase 1 again, like in October. That would probably warm up the South and make the West / Plains frigid again, although shifted a bit east of the late October outbreak.

This event does look a lot colder than last by SSTs, it's kind of interesting that the SOI response has been better so far. That may just have to do with how cold it is in Nino 1.2 relative to Nino 4. That 29.5C v. 20.2C spread in October has got to be a record. Nino 4 is literally almost as warm as October in the Super Nino of 2015-16, while Nino 1.2 is literally almost as cold as in October 2017-18. The 2017-18 temperatures in Nino 1.2 reached levels that hadn't been seen since 1980 in some months.

Nino 4 was 29.5C in October (4th warmest since 1950). Observed record warmth is 29.84C in Nino 4 in October. 1969 29.32 1982 29.44 1987 29.58 1991 29.47 1994 29.45 1997 29.34 2002 29.41 2004 29.46 2006 29.45 2009 29.68 2015 29.84 2018 29.71 Those are your Nino 4 years that hit at least 29.3C in October. Once you get above 29.5C or so, none of the years feature a cold December in the East, the coldest above 29.5C is 2009, and it is average, despite strong counter signals to Nino 4. The map last year looked like a composite of the 29.3C+ Decembers prior to it. Nino 4 is cooler than 2018 though - and it may cool a bit more relative to 2018 by December, so I do think December will be slightly cooler than last year in the East. MJO should get to phases 6-7-8-1 later in November. I'd imagine a pretty big warm up is coming with it to prevent overall November numbers from getting too out of control cold. All of those phases are warm in the South overall in OND. The subtropical jet is waking up, and the Indian Ocean has been alight with convection recently, both tend to precede warm ups.

The MJO now looks like it will be a pretty coherent force for at least the next week. We're probably going to get into phase 8 around mid-month, which times up well with the recent huge SOI crash for a big storm somewhere in New Mexico or Colorado. Phase 8 MJO is pretty wet in November here I'm pretty sure. We should get some rain this week too, will be the first November precipitation of any kind for Albuquerque since 2016.

CPC has ONI at +0.1 for ASO. They revised September warmer in the raw data. October came in +0.44C against their baseline. Against 1951-2010 means, it would be +0.69C, and it will probably be revised warmer next month anyway. Six weeks above 27.0C is an El Nino to me at this time of year. The 27.19C for October is still warmer than October 2014, even though the weeklies implied 27.3C Nino1+2 Nino3 Nino34 Nino4 Week SST SSTA SST SSTA SST SSTA SST SSTA 25SEP2019 20.0-0.5 24.8-0.1 27.2 0.5 29.7 1.1 02OCT2019 20.0-0.6 25.1 0.3 27.2 0.5 29.7 1.0 09OCT2019 19.7-1.0 24.8-0.1 27.1 0.4 29.5 0.9 16OCT2019 20.6-0.2 25.3 0.4 27.5 0.8 29.7 1.1 23OCT2019 19.7-1.3 25.0 0.1 27.3 0.6 29.7 1.0 30OCT2019 20.8-0.4 25.4 0.5 27.4 0.7 29.6 0.9 For the SOI stuff, you can look at the BOM data base and see if it has merit yourself. https://data.longpaddock.qld.gov.au/SeasonalClimateOutlook/SouthernOscillationIndex/SOIDataFiles/DailySOI1887-1989Base.txt Blizzard of 1993 was around 3/13 right? That's 31+28+13 days into 1993, i.e. 72 days. Here is the SOI in late Feb / early Mar 1993: 1993 61 1015.53 1008.30 14.95 1993 62 1013.67 1009.25 1.51 1993 63 1010.83 1008.50 -8.55 1993 64 1010.61 1008.80 -10.99 That 20+ point drop is about 10-days ahead of the blizzard. The extremely powerful storm this past March had a lesser drop, but it was there: 2019 60 1014.11 1009.00 4.79 2019 61 1012.17 1008.75 -3.30 2019 62 1012.02 1009.00 -5.22 This is the Perfect Storm time frame - (365 = Dec 31, -31 for Dec, -30 for Nov, -3 for Oct, -10 for the SOI lead time, i.e. 365-74 = 291?) 1991 290 1011.85 1012.15 -20.57 1991 291 1011.32 1011.90 -22.34 1991 292 1008.97 1011.55 -35.24 The system (closed low last I looked) coming into the SW this week ties in with the most recent drop. I generally look for a big storm over the SW in 10-days after a big SOI crash occurs in a short period (1-2 days) in El Ninos. But in non-El Ninos it is less reliable for the SW, and the reliability does seem peak from mid-Nov to mid-Apr, it has no use in Summer. 29 Oct 2019 1011.22 1010.05 -11.09 -4.59 -6.94 28 Oct 2019 1011.75 1010.10 -8.00 -4.19 -6.79 27 Oct 2019 1012.01 1008.90 1.42 -3.98 -6.5

Date Tahiti (hPa) Darwin (hPa) Daily Contribution 30 day Av. SOI 90 day Av. SOI 4 Nov 2019 1008.28 1010.80 -34.36 -4.51 -7.07 3 Nov 2019 1010.91 1009.65 -10.31 -4.05 -6.62 2 Nov 2019 1011.56 1009.15 -2.99 -4.37 -6.53 1 Nov 2019 1011.89 1008.60 2.61 -4.95 -6.6 Those 20 pt, single day SOI drops are trouble historically. Need to watch for a big time storm mid-month.

I'm aware that October 1948 is nothing like October 2019. Worth noting, the 135-165W, 30-60N zone was not warm in October like it now. It developed later. That's why that area of warmth is interesting to me, since it is west of 2013 - when it exists in 1948-49, even with the PDO, AMO, Solar, ENSO, IOD, and global SSTs way different, it still went to the pattern observed in October. The Oct 2019/Dec-Feb 1948-49 pattern only exists when that warmth is pronounced in the zone I listed. In 2013-14, you had more success pushing the heat out of the SE US compared to this year. The magnitude of the cold/heat in 2013 is also not really comparable if you use the same scale. A lot of states in the West reported their coldest ever October in 2013, with a lot of states in the South reporting top five hottest ever for October. I'm not trying to rain on parades, but I am seeing a lot of what happened last year, when people kept saying "2002" was similar to Fall, even though the October heat in the East and cold push were both more severe than in 2002.