Friday, March 16, 2018

GISS February global temperature unchanged from January.

GISS cooled, February anomaly average was 0.78°C, same as January (GISS report here). That differs from TempLS mesh, which rose by 0.06°C, as did the NCEP/NCAR index.

The overall pattern was similar to that in TempLS. Warm in the Arctic (very) and Siberia, Eastern USA, and also a band from Nigeria through to W India, bit warmest around the E Mediterranean. There was a band of cold in Canada below the Arctic extending into the US upper mid-west, and in Europe. Both show an interesting pattern of mostly warm patches in the roaring Forties.

As usual here, I will compare the GISS and previous TempLS plots below the jump.

Here is GISS

And here is the TempLS spherical harmonics plot

This post is part of a series that has now run for six years. The GISS data completes the month cycle, and is compared with the TempLS result and map. GISS lists its reports here, and I post the monthly averages here.
The TempLS mesh data is reported here, and the recent history of monthly readings is here. Unadjusted GHCN is normally used, but if you click the TempLS button there, it will show data with adjusted, and also with different integration methods. There is an interactive graph using 1981-2010 base period here which you can use to show different periods, or compare with other indices. There is a general guide to TempLS here.

The reporting cycle starts with a report of the daily reanalysis index on about the 4th of the month. The next post is this, the TempLS report, usually about the 8th. Then when the GISS result comes out, usually about the 15th, I discuss it and compare with TempLS. The TempLS graph uses a spherical harmonics to the TempLS mesh residuals; the residuals are displayed more directly using a triangular grid in a better resolved WebGL plot here.

A list of earlier monthly reports of each series in date order is here:
  1. NCEP/NCAR Reanalysis report
  2. TempLS report
  3. GISS report and comparison with TempLS


  1. I see you caught Wonderin Willis's latest "How Not To Model The Historical Temperature"

    Note the fit he used
    Temperature = 2012.7 * Orbital – 27.8 * Snow Albedo – 2.5

    If Snow Albedo is related to CO2 because of the black carbon emissions coating the snow, then this is not an independent factor and the Wonder Boy is "snowing" us once again propelled by his twisted psychoses.

    I live in a northern climate and have been cross-country skiing since I could walk. One thing that every XC skier understands is the slickness of the snow. Back in the day, at this time of the year one would get dirty old crust snow blackened by the pollution precipitates that would collect over the no-snow days and amplified by remelts. This would really slow a skier down so we are sensitive to that kind of grit fouling up the wax (only thing worse is skiing over dog crap).

    But you know what has happened in the last few decades? This stuff is really disappearing from the snow surface as particulates are disappearing from the emissions. You don't see the dirty snow anymore and skiing is great. That is partially what the inflection is conveying and the dip in the temperatures is probably related to a gradual reduction in these pollutants since the 50's as scrubbers and such were added to the smokestacks.

    1. To be fair, that was actually his example of how not to model the historical temperature.

    2. Yea, Willis is most certainly clueless about how to mathematically model any of these topics. Take an example: we all know that ocean tides can be modeled with a great deal of precision simply by applying the lunisolar forcing profile to the tidal equations. In some cases, hundreds of frequencies may be used depending on the accuracy desired. Willis would probably sneer at this approach as he has no feel for physics on this level and he probably has no intuition about what tides are, probably as clueless as Trump is about the ocean.

      Downthread in the comments to that post, take a look at how Willis goes after Ian Wilson, where Ian is simply trying to open up the analysis to consider how other forcings, and specifically lunisolar forcings, may fit in:

      Willis is really the most intellectually incurious and thin-skinned person I have come across. The stuff he is studying is well out of his league and he puffs up when anyone challenges him.

    3. I should add that Willis' example is a great example of the clueless strawman argument, straight out of Kansas. He builds something up just to tear it down, but without the slightest clue as to what he's doing.

  2. Apparently Curry thinks she can predict El Nino occurrences:

  3. I found something interesting and relevant, thought I'd put it here. For some reason that I forget I was looking at differences between OISSTv2 and ERSSTv5 in the tropics and noticed there is a big sudden tick upwards in the difference plot in 2017. Doing the same check with the raw ICOADS data using Climate Explorer I found a similar tick there too. However, it was not present in HadSST3 or ERSSTv4 (which is still being updated monthly).

    As far as I can see the most obvious factor is which ICOADS dataset is in use. ERSSTv5 uses ICOADSv3, whereas HadSST3 still uses ICOADS 2.5, recently supplemented by ERDDAP drifting buoy data as per the note at the top of this page. I would expect ERSSTv4 is still using ICOADSv2.5, and they probably haven't bothered to take the step of adding ERDDAP given that this version is effectively deprecated at this point.

    Note that Climate Explorer still uses the label ICOADS v2.5, but clicking on the metadata link reveals it's actually v3.0. Looking around I can't find any way to access a v2.5 dataset. It could be that HadSST3 is technically accessing v3 but just not using the newer data sources - e.g. ARGO floats - in that release.

    Anyway, the upshot is that it seems likely 2017 was biased slightly high in the ERSSTv5-based datasets, at least relative to other recent years, and it appears to be due to some problem with the newer data sources in ICOADSv3.

    1. Thanks, Paul
      That is a big jump. I could rerun TempLS with ERSST v4 to see the effect.

  4. Nick and PaulS,
    The switch from ERSSTv4 to v5 was significant for the ranking of years in TempLS mesh, making 2017 second warmest instead of 2015.

    I have considered ERSSTv5 as an improvement since it includes Argo data, giving much better coverage in areas sparselely sampled by ships and drifting buoys.

    I made a chart similar to Pauls, but with the the difference from Argo-only data.
    And yes, Argo makes the difference being clearly warmer in 2017 compared to the others. ERSSTv5 follows Argo best, probably because it includes Argo data. Era interim skin SST is quite near and may also have Argo influence.

    The difference may not only be explained by Argo coverage, Argo "surface" is from 5 m whereas drifters measure close to the sea surface.

    1. Thanks Olof,

      One thing I would need cleared up is whether Argo 5m is included in ICOADSv3. I thought I'd read that it was, but then the ERSSTv5 write-up says they obtained Argo data separately from ICOADSv3. The reason that's important is that, as shown in my original plot, a raw ICOADSv3 average shows a similar divergence from OISSTv2, ERSSTv4 and HadSST3. If Argo data isn't included in any of those datasets it can't be the primary cause of this divergence.

      I've made some additional comparisons for NH and SH Extratropics. In this case I've compared ERSSTv5 with ERSSTv4 because there seemed to be some seasonal differences when comparing OISSTv2, which obscured the pattern.

      NH Extratropics
      SH Extratropics
      and Tropics again for reference

      The divergence appears to be similar in the NH Extratropics, but not apparent at all in the SH Extratropics. I would have thought if Argo were the issue it would be most apparent in the SH where it has a bigger proportion of total observations. Instead the geographical pattern seems more indicative of some change in more traditional measurement platforms.

      One interesting thing about your 5m Argo difference plot is that it shows Argo data consistently varying less with ENSO than the SST datasets. We would therefore expect a dip in the difference plot in 2017, due to the La Nina, which is what we see. However, ERSSTv5 appears to be more consistent with previous La Nina dips than the other major datasets. If Argo is not in ICOADSv3, and therefore not the primary cause of divergence, this comparison with the apparently homogeneous Argo network would suggest ERSSTv5 is actually in the right here, not the other datasets.