raindancewx

Members
  • Content Count

    1,957
  • Joined

  • Last visited

Everything posted by raindancewx

  1. Pretty sure the IOD was positive last year too by Fall anyway, although not as much as this year. The Jamstec has it falling off later. The main thing with the IOD is it is a way to keep relatively high pressure over Australia, which helps the SOI stay negative or neutral.
  2. The (relatively) cold backwards C by the Alaskan Coast around that blotch of warmth is a pretty canonical -PDO, especially with the warm waters east of Japan. 2014 was the opposite - super warm by the coast of Alaska, and cold east of Japan. Will the -PDO ish look now last all winter? I kind of doubt it. Just because Nino 1.2 falling below 20.0C in October is a leading indicator that the PDO will fall. It did on the weeklies. But now it looks like it will warm up. So it probably reverses, at least somewhat, by December or January. Even with the El Nino last year, the Nov-Apr PDO only recovered to +0.5 from +0.3 in 2017-18. With a weaker El Nino this year, it could easily be closer to 0. The trend has been down. I think the PDO is weak enough that the SE ridge won't get squashed or destroyed for any length of time, and when the subtropical jet is strong, and the SE ridge is strong, that is the best shot at major storms (cut offs) for where I am, as the two battle it out. Nov-Apr PDO 2013 0.38 2014 2.07 2015 1.70 2016 1.06 2017 0.30 2018 0.51 With the NAO negative, the winter -PDO correlations seem to work about perfectly for October so far.
  3. Looking back at what the Weather Channel forecast in August for Fall really gives some sense of how difficult this winter may be to nail. They really were quite close on September, I'd give it a B- (from mid-Aug), but their October forecast is going to be fairly opposite, with the very warm South, somewhat warm NE, and very cold NW. I don't really get why they expected the pattern to flip so quickly? I know its gotten colder since the early days of October when it was in the mid-90s as far as Philly...but still. I generally find their forecasts verify better than say, CPC or Weatherbell, and they're definitely better than WXRISK or Accuweather.
  4. Nate Mantua PDO value came in today: https://oceanview.pfeg.noaa.gov/erddap/tabledap/cciea_OC_PDO.htmlTable?time,PDO Sept 2019: +0.41 Aug 2019: +0.38 Sept 2018: +0.09 I believe October will be substantially lower than September, but we will see.
  5. I used different analogs, but I agree on the mid-winter warm up for the East. The years I picked had the NAO negative in December, neutral in January, then positive February. More generally, the big warm Sept-Oct periods in the East usually feature some kind of late warm up in winter it seems like. My pure analogs are actually very cold in December, but I think it will be warmer (somewhat) in the East with Nino 4 near record warmth than the raw blend. I also agree on a pattern with some huge storms - I think there are somewhere between 3-5 historic storms Nov-Mar nationally, it's just where? Dorian, the recent record blizzard in the NW a few weeks ago, and the relatively rapid shifts in temperature profiles in the Fall all hint at it. I think the culprit for the cold shifting positions is likely the changes in the PDO. It appears to be heading negative, with the cold ring next to Alaska developing around a warm tongue east of Japan right now. The Modoki El Nino look right now is also likely to break down, which favors the West later, as you say.
  6. Not quite as warm as last year....but pretty close now in Nino 3.4. I think the Jamstec is too cold in the SE, but I'm broadly on board with what it has. A Modoki El Nino (ish) with a neutral or -PDO (colder right by Alaska for SSTs, with a warmer tongue east of Japan. I really struggle with places in the mid-south going from +8 to +12 in October, after a warm September, to a very cold winter like the model shows. The Jamstec is a bit drier than I am in the Rockies (I think we're near to slightly above average, it has near- to below).
  7. Just for reference: Nino 3.4 is now very close to last year. K - 273.15 = C for reference. So the 27.0C is orange, because it is 0.5C or more above the 1951-2010 October average in Nino 3.4
  8. At the end of the day, you can try to forecast a season using two basic methods: 1) Find predictive variables and guess what they will do (this seems to be what most of you do). That will work most of the day, but you'll get situations like last year, where 1963 matches on most of the variables that are predictive but end up completely different by ignoring weather clues, like the super hot October 1963 compared to the very cold October 2018. My analog system, prior to incorporating weather matches had 1963 as a top three match last year...but it fell off to like 15th with weather included in Fall. (2018-19 was 27.40C in Nino 3.4 for DJF, 1963-64 was 27.35C in Nino 3.4 for DJF) 2) Find predictive variables. With the predictive variables find a blend that matches the weather for a long-period, with the variables and weather still heading in the "correct" direction together going forward. This is what I try to do. I'm really looking forward to seeing the forecasts this winter - there are so many more things that can go wrong this winter compared to last winter: 1) Solar - there is a non-negligible chance of a rapid increase by mid-2020 2) PDO - it looks like it is heading more Neutral, and maybe legitimately negative, but it also may snap positive. My analogs assumed near neutral, on either side. People see the Blob and think "2013" but US weather really hasn't matched well to 2013 at all since Summer. October has been a bit closer, but I think it is a case of two trains passing each other from opposite directions. In 2013-14, the PDO was warming from years of being in the predominantly negative phase. In 2019-20, it is cooling from years of being in the predominantly positive phase. 3) ENSO structure. Modoki El Nino since September. But Nino 1.2 is losing the cold pool below it, Nino 4 warm pool below it is moving east. 4) NAO. Pretty rare historically to have any streaks <=-0.3 more than four months, less than 2% of all possible outcomes. October is negative. November probably will be too. 5) SOI under -10 in September is like a ~95% El Nino indicator for winter. For what its worth, the new Jamstec (October) does have an El Nino Modoki look continuing in winter, with a cold East. The dry slot from TX to MI, with New England and the West Coast warm is pretty much what I had in my forecast.
  9. Nino 1.2 is currently under 20.0C in October on the weeklies. The monthlies usually come in warmer for Nino 1.2...still, if the month finishes under 20.0C in Nino 1.2, a -PDO is likely for winter. Since 1950, 17/18 years with a <20.0C Nino 1.2 in October have finished with a -PDO for Nov-Apr. With an El Nino forming, that can be an incredible pattern nationally. The temperature pattern for October (super cold NW / super hot SE) is consistent with the PDO correlations to winter when the PDO is negative. This is the composite for when the PDO finishes under 0 for the Nov-Apr period, in an El Nino (it is the same idea - cold West / mild East if you use only 1950+ years)
  10. The first map is a correlation map, so the yellow/green colors mean positive correlation, i.e. when the PDO is negative the NW is cold. The blues/purples mean negative correlation, so when the PDO is negative, the SE is warm. The opposite is also true, positive PDO means cold south (negative correlation), positive PDO means cold northwest (positive correlation). The cold ring of waters by Alaska and then the warm tongue of waters east of Japan - that is the canonical -PDO image
  11. I'm starting to think the PDO may drive the pattern this year. The October split so far, very hot SE, very cold NW, is consistent with a -PDO in winter if you look at the correlation map I posted in the ENSO thread. The zones that are most correlated are currently the hottest/coldest so far. The cold ring forming by the Alaskan coast surrounding the warm tongue east of Japan is the canonical -PDO look. Really aren't any recent El Ninos (27.0C+ Nino 3.4) regimes with a -PDO, but it has happened. Nino 1.2 is under 20.0C for October to date on the weeklies (19.85C). In 18 <20.0C Nino 1.2 Octobers, 17/18 the PDO finished below 0 on average from Nov-Apr, and the highest value was in 2005-06 at +0.14. Generally speaking, a -PDO favors a cold NW, a neutral PDO favors a cold Central, and a positive PDO favors a cold SE in winter. In 2017, the weeklies were ~19.5C for October in Nino 1.2, but the monthly data came in at 20.2C. So this may all be moot anyway, since I base stuff off the monthlies. Blending the PDO in Mar-Aug with Nino 1.2 SSTs in October is highly predictive, and the best match for that looks to be 1996/1988 averaged together for the PDO, for -0.1 for Nov-Apr. That's for a 20.0C October Nino 1.2 after a +0.83 Mar-Aug PDO. If the PDO hadn't been +0.8 in Mar-Aug, I think the -0.5 PDO depicted below for a 20.0C Nino 1.2 in October would be about right. When the PDO is warm mid-year, and then Nino 1.2 is very cold in October, you look for Nov-Apr to be more negative generally (so in 2017-18, with the cold Nino 1.2, we went from +0.6 Mar-Aug to +0.3 for Nov-Apr).
  12. My take on the very cold NW/very hot SE this October is that the PDO is going negative. Look at where the biggest anomalies v. the correlations are (granted, they are for winter, but with the NAO negative its a pseudo winter pattern now). Blending Nino 1.2 in October (currently under 20.0C on the weeklies) with the PDO in Mar-Aug (+0.83) to match 2019, you have to pick something like 1988-89 and 1996-97. I've hindcasted this method, it's remarkably strong as a predictor once Nino 1.2 is known in October. Analog PDO M-A Oct 1.2 PDO N-A 1988 0.86 19.81 -0.60 1996 1.06 20.31 0.38 Blend 0.96 20.06 -0.11 2019E 0.83 20.00 ??? This was the blend I used to estimate the PDO last year. Year Oct 1.2 PDO M-A PDO N-A 1963 20.96 -0.75 -0.88 2002 21.39 -0.24 1.69 2004 21.16 0.57 0.47 2004 21.16 0.57 0.47 Mean 21.17 0.03 0.44 2018 21.13 0.07 0.51
  13. All fear the PDO? The JISAO update isn't in for September yet, but the severe cold in the NW and severe warmth in the SE are both showing up in the zones that correlate highest to the PDO for winter. The cold ring by Alaska hasn't been there in a long time but it is trying to form.
  14. The cold pool near Peru is starting to surface so Nino 1.2 and Nino 3 cooled this week. Nino 3.4/4 were down a tiny bit. Against 1951-2010 averages in Nino 3.4, 27.0C is +0.5C, but CPC uses 26.75C, the 1985-2014 average, for Nino 3.4. Either way, Nino 3.4 for October is at borderline El Nino conditions - 27.15C or so, pending the rest of October...but lots of warmth below the surface is set to come up shortly. Nino1+2 Nino3 Nino34 Nino4 Week SST SSTA SST SSTA SST SSTA SST SSTA 18SEP2019 19.1-1.3 24.2-0.6 26.5-0.2 29.3 0.6 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 You can see that if we pop up to 27.15C or so in October, it's fairly close to the top 10 warmest Octobers already (27.58C is 10th since 1950). Nino 3.4 for the past three weeks has more or less caught up to 2014-15 at this point, but 2014-15 was much warmer in Nino 3/1.2 and colder in Nino 4. At this time of year, Nino 1.2 tends to lead what the PDO will do - so the very warm Nino 1.2 in 2014 preceded the +2.07 PDO reading for Nov-Apr. 17SEP2014 21.0 0.7 25.2 0.4 27.2 0.5 29.4 0.8 24SEP2014 21.2 0.8 25.4 0.5 27.1 0.4 29.3 0.6 01OCT2014 21.7 1.1 25.4 0.5 27.1 0.3 29.2 0.5 08OCT2014 21.3 0.6 25.5 0.6 27.1 0.4 29.1 0.5
  15. I went with a fairly snowy cold season for parts of the Plains, West and Northeast this year. Not many areas of the US came in with less than 75% of average snow. For New Mexico, the "average" snowline seems to be 7,000-8,000 feet - areas above that level will be snowier than average. The ocean temperature pattern globally in early October is pretty close to opposite of what the 2016-17 winter ended up, essentially a Modoki El Nino instead of a Modoki La Nina. If it was truly opposite to 2016-17, maybe it'd be severely cold in the Southeast, but I have the cold somewhat to the Northwest of the South. I do think with Atlanta and areas west of it currently 10F above normal in October after a hot September that a lot of nights in the South will be pretty cold as the warm air is replaced by cold/dry air. Longer term, the subsurface heat is draining in Nino 4 to some extent, and the wave of ocean heat is moving East, so you'll see less of a Modoki look as Nino 4 cools and Nino 1.2 warms. The PDO also looks like it is going to go negative for at least a little while which is not great for cold in the Southeast.
  16. Definitely an idealized Modoki El Nino look for the past month, but the warm waters do seem to be pushing east into Nino 3 now. The areal extent of the cold to the east is thinning. You can also see the PDO going negative to some extent as the warm ring moves away from Alaska and the tongue that should be cold east of Japan in a positive PDO, is very warm. 1953-54 had a 23.25C DJF seasonal reading in Nino 1.2 - so I have that in my analogs to incorporate the very cold waters by Peru that will take a bit longer to warm up than Nino 3.
  17. I've done data mining on the NAO and the winter tendency will tend to match a year with similar April-May transitions AND similar March-Sept transitions. Years with those tendencies that match 2019 are 1990, 1993, and 2016 among recent years - all positive NAO winters. However, my winter analogs do imply one winter month will see a -NAO, despite the other two months being positive or neutral. Independently, Apr-May and Mar-Sept are both r-squared around 0.1 for predicting the NAO, but blended together correctly it is stronger than that. On the monthly NAO data, 1975 was a good match to last year for Apr-May, Mar-Sept, and it was a positive NAO winter like last year. There is also some tendency long-term (r-squared around 0.2) for Box C cold in the Modoki calculation (waters by the Philippines) to be a leading indicator of the NAO. I've linked to my outlook elsewhere - will be interesting to see how good or terrible it is in a few months.
  18. Nice and dry + cold = very cold lows for the rural areas of the high terrain in New Mexico.
  19. My analogs had parts of Oklahoma, Missouri and Kansas pretty snowy if you want to see. I included some slides at the end of my forecast about how well I did last year, so you can decide if it is worth paying attention to what I have or not. There is a slide that has snow totals for Oklahoma City, Tulsa, Little Rock and some other towns in the region if you are curious.
  20. I have a fairly warm November for the West, but I think we may sneak in a big snow storm for WY/CO/maybe NM too during that month.
  21. https://t.co/1yr0fAbUKK 2019-20 Winter forecast. I went with a Modoki El Nino turning to a more basin-wide event with time based on the subsurface. Snow maps, precip maps, temp maps - they are there. https://t.co/OwLKUbp8ka?amp=1 2018-19 forecast - did some images showing verification at the end. I'm not completely sold on this event being officially designated an El Nino, I think we only get 3-4 months at 27.0C in Nino 3.4, but its close enough honestly.
  22. The scribd link at the top has the 2019-20 forecast. Included the link to last year to show the images I verified at the image were present from October.
  23. I'm not sure Modoki stuff follows a cycle like ENSO overall, especially with Nino 4 warming faster than the zones to the east over the last sixty years. I do think it is interesting that these recent El Ninos have built West to East, a lot of historical El Ninos built East to West if that makes sense, so that might be a cycle. This is the stuff I was waiting for to finish my forecast. The trend is for Nino 3.4 to warm up somewhat from September readings, like it was last month - but Nino 4 is forecast to cool while 1.2/3 warm - so the look becomes less of a Modoki if all that verifies.
  24. I've basically finished my winter forecast using the updated analog weighting. Snow maps are a pain in the ass, I ended up looking at something like 40 or 50 towns to try to draw the boundaries correctly. I was hoping the ECMWF plume for Nino 3.4 would be out by now, but it isn't. If it isn't out tomorrow, will probably just post my forecast without it. Here is the subsurface trend for the past few months. The Nino 3.4 zone is completely filled up now with warmth. It does look to me like the warmth is moving East, so the Modoki look right now will probably fade later in winter, I think that is why my analogs are showing a big mid-winter thaw nationally for a month or so.
  25. It looked like a slightly below average snowfall season for Denver when I ran my initial simulation. But that blend assumed a Neutral with warm waters in Nino 3.4 staying west of 140W. It seems like they'll spread east enough to be an El Nino. Ultimately I think its a fairly seasonal winter for temperatures for NM and CO, but with a handful of very strong, wet storms. If the MJO gets stuck in phase 8/1/2 in winter, like it is now, the current system set to be super cold for the time of year should show up again at some point, even if its not til March, but further South and more moisture. My original weighting was 1953-54, 1983-84, 1992-93, 1995-96 (x3), 2009-10 (x2), 2018-19, but I can't really justify keeping 1995 that strong, even though Summer 1995 and the QBO are similar to 2019. I'm planning to have my outlook out later this week - I have the new weighting with the same years, it just takes a while to manually re-do the maps and data mining.