Anyone make a hottest year market yet?

@ScottSupak likely will be hottest per GISTEMP (much more confidently than NCEI,ERA5 dataset questions which I put both as a point estimate in the neighborhood of maybe 65-70%).

Comparison of polymarket bins to my model

Polymarket's 95% seems a tad overconfident in comparison to my model, likely because they are seeing point estimates for the temperature roughly ~0.1C above last year's temp (1.25):

GIS TEMP anomaly projection (January 2024) (corrected, assuming -0.066 error, (absolute_corrected_era5: 13.150)):
  1.359 C +-0.077

The point estimate I have for above is around 87% of being > 1.255 from the model I use, but using the Z-score from that CI puts it not breaking the record at ~ 1.35 std.deviations away (9%), corresponding to 91% chance of it breaking the record. Enough variation in remaining days to shift the middle point +-0.015 C (+- 0.1 C offset for remaining days).

This questions bin for reference:

@parhizj how did you get to ~0.1C above from last years January?

@LeonardoParaiso you do a ERA5 daily temps -> GISTEMP month model:

say the ERA5 temp for this month will be 13.221 C (use GEFS or ERA5 or some other model to extrapolate rest of month). Calculate the average temps for January for ERA5 then calculate the same for GISTEMP LOTI; the difference will give you a rough estimate of how much to offset the 13.221 C. You can be more sophisticated and add error correction terms (like the more warm ERA5 temps the more we should reduce the temp as this is a proxy for the bias of GISTEMP which fixes the ice free area so it tends to underestimate the ERA5 temps in the current modern, warmer climate) in which case for a simple linear model for January I get an adjustment of -0.068 C bringing the ERA5 temp down to 13.153 C. This produces then a GISTEMP temp of 1.361 C for my model.

This is > 0.1C the January 2023 temp of 1.24. I recall the January 2024 temp itself is probably on the border of almost reaching 1.25 i.e. ~1.244 from the recent gistemp run I have done) so this puts it at around 1.361 - 1.244 = 0.117 C higher:

GIS TEMP anomaly projection (January 2024) (corrected, assuming -0.068 error, (absolute_corrected_era5: 13.153)):
  1.361 C +-0.074

@parhizj Why do you still have positions between 1.19 and 1.30 if your gistemp runs go so high? Also, you dont take into account any of the ERA5 SSTS? Can you elaborate on why?

@LeonardoParaiso That is (roughly) the probabilities I get for it.

I only use the ERA5 global temps for this model (https://sites.ecmwf.int/data/climatepulse/data/series/era5_daily_series_2t_global.csv) as this is just simpler. If I wanted to try to go for the best accuracy I would redo all the ERA5 global 2m temp averages myself trying to regrid them according to the (ice-free) subboxes present in the gistemp baseline (this takes way too time and is a waste of compute for these predictions).

As for using the ERA5 SSTS I don't know how I could effectively use them (are you thinking of combining the ERA5 land temps and the SSTs separately and trying to infill like GISTEMP? You still would need to redo all the years to get a model to match to GISTEMP, and this seems much more complicated and I don't know how much benefit there would be.) Otherwise, more speculatively you could try to refine the error terms for SST instead of using a simple linear correction model? (again I don't know how much benefit there would be). In all there is not much motivation to refine the model I have given the uncertainty already present in the data itself (gistemp itself has an uncertainty of ~0.05C already).

@parhizj I see what you mean now. I just don't understand that, by now, knowing the that GISTEMP will be around 1.36C with the uncertainty of ~0.5C how can you put the odds of much lower than that so high? Like the interval of 1.15-1.29. Are you counting on La Niña adjustments or SST adjusments like i said?

Temps look relatively flat and constrained for rest of the month

A +-0.2 offset range is probably too large, so anything within +- 0.1 range looks reasonable to me as a meta prediction... (anything between the green and pink lines as most likely at the moment):

This one time I computed for the Polymarket's bins as well to see how their spread compares to mine for reference and as usual it has a much tighter distribution. For reference polymarket is currently at 1.30-1.34 at 40% and >1.34 at 45%, suggesting something along the lines of the green or orange line; this implies an offset of about -0.1 C which is within range and I don't necessarily outright disagree with but I think a more flat 0.0C offset (purple line) is more likely given the forecast temps seem to mostly be flat rather than a trend rising or flat):

Edit (fixed graph):

Polymarket's percents for Jan. (diff bins from this question):

>@parhizj
Peak recent anomaly 1.33C for Oct and La Nina monthly values
2024 7 0.21
2024 8 -0.07
2024 9 -0.15
2024 10 -0.28
2024 11 -0.14
2024 12 -0.62
possibly starting to go down more steeply but maybe it is a while longer before that starts to have its effect?

ENSO cooling effect should be outweighing greenhouse gas slow steady effect?

I guess above sort of reasoning should carry little weight versus having data for January temps?

@ChristopherRandles I've been focusing on my own project and haven't had time to go through the comments on real climate, but I did read somewhere (news site?) that essentially climate scientists haven't reached a (detailed?) consensus on attribution for the extreme 2023-2024 anomalies:

https://www.realclimate.org/index.php/archives/2024/05/new-journal-nature-2023/

https://www.realclimate.org/index.php/archives/2024/12/nature-2023-part-ii/

The paper mentioned in the second post attributes most of it to change in albedo, between mostly northern mid-lat and tropics but still far away from filling in the details:

"What they can’t do using this methodology is partition the albedo changes across cloud feedbacks, aerosol effects, surface reflectivity, volcanic activity etc., and even less, partition that into the impacts of marine shipping emission reductions, Chinese aerosol emissions, aerosol-cloud interactions etc. So, in terms of what the ultimate cause(s) are, more work is still needed."

In the most recent post below though covering different baselines Gavin mentions something interesting regarding the anomalies though:

https://www.realclimate.org/index.php/archives/2025/01/2024-hindsight/

"With much of the focus on the longer-term records, it seems to have flown under the radar a little how oddly the MSU/AMSU records have been behaving over the last year or so. As with the surface records, the satellite products (UAH, RSS, NOAA STAR) all have 2024 and 2023 as the warmest and second warmest years, but unlike the surface records, 2023 was not such a outlier (~0.06ºC above 2016), while 2024 was huge (with records broken by ~0.32ºC). Additionally, the time over which the peak temperatures have lasted (17 months or so) is much shorter than the peaks around 2016 or 1998 (7 months). I don’t have much insight into why this is happening, but it might hold some clues about the drivers of the recent anomalies."

@ChristopherRandles

"ENSO cooling effect should be outweighing greenhouse gas slow steady effect?"

2020 tied for hottest year at the time, even tho it was the first year of the triple dip Nina.

@ScottSupak
Over 10+ years steady GHG effect tends to swamp other effects.
Over 1 year ENSO can swamp other effects, it is a short term effect.

For the 2020 year you picked a la Nina did develop but there were mild El Nino conditions at start of year. Being the first year makes it a little weaker of an effect than following two years when La Nina existed throughout the year.

2020 was also first year of 70% sulfur emissions reductions on marine fuel. Also a rapid effect so perhaps biggest warming effect in any one year from this.

So on an annual basis, while ENSO can swamps other effects, it does not always do so if the ENSO effects are weak/only part of year and/or a few other effects from GHG, marine fuel emissions, solar ... can combine to outweigh ENSO.

If instead we look on a monthly basis we can see than by Dec 2020 when La Nina had had a chance to kick in, the temperature was lower than each of the previous 5 Decembers.

@ChristopherRandles yes, of course, and so the question is how strong will this Nina be, and how strong will effects that can counter it be?

It’s very likely this La Niña will be weak, with the Niño-3.4 index unlikely to reach -1.0 °C for a season. This is based on computer model guidance and how late in the year La Niña conditions emerged. ENSO events peak in the northern Hemisphere winter, and there’s just not a lot of time for La Niña to strengthen.

@ScottSupak yes, so what do we see:
GHG steady as normal
Marine fuel still warming but by reducing amounts
Solar probably still marginally increasing warming effect
Hunga Tonga both warming and cooling effects now reversing back towards trend but probably by reducing amounts.

Basically unlike the 2020 situation there isn't a whole lot of significant new warming effects for January 2025 to counter Nina effects.

NINO3.4 anomaly dropped to -0.62 in Dec from -.014 for Nov (for latest data I am looking at monthly data rather than 3 month averages)
https://www.cpc.ncep.noaa.gov/data/indices/sstoi.indices
that is quite a sharp decline (certainly compared to recent 3 month averages drifting slowly lower) but I would point out there is also some uncertainty in timing in when this flows through to global average temperature. It might well be too soon to expect a strong effect from this in January.

Using monthly ENSO data rather than 3 month averages is probably a bit dodgy, longer period averages work better - which is probably admitting there is unpredictable residual noise. On monthly change in temp, this unpredictable residual noise might swamp the expected difference between ENSO and other known effects. Hence we could get a temperature increase despite an expectation for Nina cooling to outweigh GHG, solar and other known effects. For estimating January temp data, January temp data is probably more reliable than estimating what changes we expect from Dec.

That is perhaps just a more detailed explanation of why I wrote "I guess above sort of reasoning should carry little weight versus having data for January temps?"

@ChristopherRandles we also see arctic sea ice well below trend, so more heat absorbed with all the feedback effects there; a possible increase in ghgs due to magat policies; a decent chance the nina is weak; and other hot spots in the ocean making up for lower SSTs from the nina.

@ScottSupak It is dark in the arctic, max ice reached ~March. So little solar radiation being absorbed now - it is more of a chance of that excess surface temp heat dissipating to space as conditions return to more like normal at the ice max.

Manufacturing returning to USA would likely be good re energy efficiency and mix and less transport. Alternate outcome of tariff war reduces demand. Not sure I see an outcome with increase GHGs, but perhaps. However, even supposing there is increases in GHG, it takes a long time to change the forcings which are accumulation of effects over at least 20 years.

Do hot spots have a tendency to return to more normal?

Other than those points ...

@ChristopherRandles For what it's worth, statistical models early this year were giving estimates (with caveats) of 6% and below for 2025 being hottest https://x.com/hausfath/status/1877822767090852044

@StevenK 6% being the estimate from Berkeley Earth https://berkeleyearth.org/global-temperature-report-for-2024/

(of course, these were before January temperatures)