Monday, September 14, 2015

GISS up by 0.06°C in August

GISS is out, and the land/ocean anomaly has risen from 0.75°C in July to 0.81°C in August. This is a smaller rise than I expected. It is similar to the 0.07°C rise I posted for TempLS mesh, but that has since crept up to 0.09.

The result is surprising to me, because GISS Ts, the land stations version, posted a very large rise of 0.22°C. Yet the analysis I posted for TempLS showed most of the rise accounted for by SST, with mostly stable land data portion. HADSST3 also rose.

There were some other oddities. GISS is posting a little earlier than usual, but unusually on a US Sunday, or at least that is when my data monitor picked it up. Yet the file was dated Friday 11th, which would be very early indeed.

The pattern of heat/cold is similar to that I have posted for NCEP/NCAR. Warmth in Middle East/Africa, but GISS has it right through W Europe too. Warmth in S America, and of course El Nino. Coolish in Antarctica and Central Russia. Maps below the fold.

Update. Some have suggested the Arctic melt season has ended. But in the Jaxa SIE record, two recent days of melting has taken 2015 to below, not only current 2011, but the 2011 minimum. Unless data is revised, that probably puts 2015 third behind 2012 and 2007. I think passing 2007 is unlikely.

Here is GISS for August:



And here is TempLS mesh (spherical harmonics):




at

8 comments:

  1. ehakSeptember 14, 2015 at 5:07 PM

    I can understand your puzzlement. The land-ocean index is 0.01 lower than August 2014. The met-station indes is 0.08 higher than 2014. At the same time Ersstv4 is 0.06 deg higher this year.

    Doesn't add up. Looks glitchy.

    ReplyDelete
  2. Olof RSeptember 14, 2015 at 6:31 PM

    I agree with ehak, arriving at a similar conclusion after playing with the Gistemp map. Gistemp has been a little bit glitchy since the introduction of GHCNv3.3 and ERSSTv4.
    According to the Gistemp map, SST only have a poor coverage in the Arctic ocean, much less than the green SST-crosses in NIck's "reporting stations" map.

    Meanwhile, TempLSgrid has ticked up to 0.748, the warmest monthly anomaly ever...
    I'm looking forward to reports from other gridded datasets; JMA, NOAA, and Hadcrut..

    ReplyDelete
  3. Kevin CowtanSeptember 14, 2015 at 9:58 PM

    If I take the giss land-ocean grids (from the SBBX files) and integrate using a land mask, the land temp is down 0.02 from last year and the SST is up 0.03 from last year.

    That's different from the dTs index, which extrapolates over oceans, suggesting that the coastal land stations are warmer and the interiors cooler this year. The smaller SST increase suggests that GISTEMP is filling in cooler land temps onto sea ice areas. In fact the ice cover is lower this year than last year (maybe)? So there is less infilling from land -> ocean this year than last. That could be playing a role.

    So it may not be a glitch - it may just be down to how GISTEMP does blending. However that is pure speculation: it is testable but I haven't tested it.

    ReplyDelete
    Replies
    1. Nick StokesSeptember 14, 2015 at 10:27 PM

      Kevin,
      Yes, I've been integrating the 1200 km grid. I get a small change - a rise of 0.08 instead of 0.06. I've run spherical harmonic fits to try to locate the difference - nothing obvious. I'll post about this tomorrow.

      Actually, the main difference is indeed that GISS says the Arctic was cool in Sept, while TempLS says warm.

      Delete
  4. ehakSeptember 14, 2015 at 10:46 PM

    JMA is in. Up 0.07 from July. Record from last year crushed by 0.12 deg.

    ReplyDelete
  5. ClimfrogSeptember 14, 2015 at 11:12 PM

    JMA August anomaly is massive http://ds.data.jma.go.jp/tcc/tcc/products/gwp/temp/fig/aug_wld.png

    ReplyDelete
  6. Olof RSeptember 14, 2015 at 11:50 PM

    That's the warmest monthly anomaly ever at JMA, beating "old" Feb 1998 by 0.02. It's the warmest summer and seasonal anomaly as well...

    ReplyDelete
  7. JCHSeptember 21, 2015 at 1:58 AM

    OT: Nick's number for September so far is .367! Holy cow.

    ReplyDelete