Thursday, January 7, 2016

Big rise in TempLS in Dec; record-breaking 2015 was 0.129°C hotter than 2014.

The TempLS mesh global anomaly (1961-90) rose by 0.07°C in December (from 0.894 to 0.964). That makes it the hottest month in the record, beating October by 0.04°C. And it makes 2015 by far the hottest year. The annual average was 0.727°C, exceeding that of 2014 (the previous max) by 0.129. This is based on 4200 stations reporting in December; Greenland and Mexico have not yet reported.

The rise was a little less than that for the Moyhu NCEP/NCAR index, which rose by 0.108°C from Nov to Dec. However, TempLS grid rose even more, by 0.164°C (from .832 to .996). And the annual mean, at 0.744°C) exceeded the previous record by 0.163°C.

In terms of implication for the major indices, as usual LS mesh links with GISS and grid with NOAA and HADCRUT. That would make a figure of at least 1.1°C for GISS - my guess would be about 1.15, and bigger rises in NOAA and HADCRUT. Of course all will have 2015 as a record year. GISS at 1.15 for December would put the annual average at 0.870°C, 0.127°C higher than 2014 (the previous record).

The breakdown for December showed a small drop in SST contribution, outweighed by rises in Asia, Siberia, and N America. The map shows a huge band of warmth around the 50°N parallel, through Eastern N America, Eurppe and Russia. Then also the El Nino region. Antarctica and Arctic were more normal, which is cooler than recent months. Changes in these regions are more strongly reflected in TempLS mesh than grid, which probably explains the greater rise in LS grid.

The satellite measures also showed strong rises - RSS from 0.428°C to 0.543°C in Dec. As Sou noted, that means that each of the last three months was the hottest ever (of that month) in the troposphere record.

I'll put the full report below the fold.


  1. Did the IPCC state the target anomaly for the first two decades of the 21st century? I believe it's .2C per decade in the SPM and an average of .2C per decade in another place, but to what anomaly should one add the .4C to get the target for 2020?

    1. The IPCC hasn't made any major statement on short term trends. They shouldn't, because GCM's don't claim to get their synthetic weather synchronized to Earth. There ius a passing remark somewhere that says about 0.2C/decade for some rather ill-specified period. I don' think this really qualifies as an IPCC "statement".

      But still, the long term trend should be approaching something like 0.2. In the trend viewer I marked 0.17 C/decade as a reasonable expectation for recent past decades. And while there are certainly periods for which it was not attained, there are equally recent periods for which it was exceeded.

      The Backtrend mode of the active plot on the data page shows this. If you go back twenty years, surface data generally has a trend exceeding 0.15°C/decade.

  2. Well, the SPM wording is pretty strong, and I would agree it was an unfortunate thing to have done.

    The 2001 mean is .5425C. Looks like 2015 will be ~.87C, which is looks very close .17C per decade... if this is how it would be done. I'm asking because another warmest year in 2016, which people are saying is possible, appears to put it within reasonable range of .2C per decade as of 2020.

    1. JCH,
      Sounds like you mean the AR4 SPM, where they said
      "For the next two decades, a warming of about 0.2°C per decade is projected for a range of SRES emission scenarios."
      It's not so clear whether that means an average of .2, or .2 in each decade. If you take it as GISS, and time from start 2007, then it is currently over 0.2 C/decade (.216 from 1/1/07. Current warmth could only increase that.

    2. IMO, it's ambiguous as to whether they meant from 2007, or the first two decades of the 21st century. On skeptic blogs they seem to think it means 2001 through 2020 as that looked like it would be very flat throughout... no longer the case. Anyway, if it reaches .4C by 2020, which is now entirely possible, it would silence one of the Climate Etc. arguments.

    3. If my memory is correct, the CMIP5 rcp8.5 model average trend for 2001-2030 is 0.27 C/decade. On the other hand, for a mixed model mean, with 29% land temps and 71% SST like the observational indices, the trend is 0.23 C/decade.
      If observations doesn't match the latter trend, it could be explained by natural variations, or that the actual forcings are weaker than the RCP.. (eg Schmidt et al 2014)

    4. The biggest problem with the 0.2C/decade number is that it doesn't specify an uncertainty bounds. It's pretty hard to interpret what it even means as just a central value, when you want to compare against data and the inevitable discrepancies that show up when comparing model to data.

      Olof: If observations doesn't match the latter trend, it could be explained by natural variations, or that the actual forcings are weaker than the RCP.. (eg Schmidt et al 2014)
      This goes back to my comment on the uncertainty in the trend...I think another big question is over the magnitude of the feedbacks in the models. I haven't seen a serious attempt to set bounds on just how much variability in feedback nature actually allows, but I'd be surprised if it wasn't at least a factor of two.

  3. To reuse a phrase I recently saw in a comment from one low-level global warming denier, yes, "some parts of the world have been experiencing a milder than usual year so far".

    That level of innumeracy and ignorance is shocking, but at least it's slowly declining to a minor nuisance, the damage done.

  4. Nick,

    On your...

    ... page you have a broken link ...

    ... should be ...

    I might also have some questions on your NCEP reanalysis data files ...

    Basically about the NCEP reanalysis product (1979-present) versus the ERA Interim product (1979-2014).

    Tripped over this stuff at Woy's blog ...

    Particularly this graphic from the NCEP monthly time series ...


    1. Everett,
      Thanks - I've changed the UAH link to point to the directory. It's likely they will coninue to chenge the file name.

      I'm not sure of your question on NCEP vs ERA Interim. I've chosen NCEP/NCAR basically because it is promptly updated, with files of manageable size.

      On the comparison of Decembers to past years, you may notice that I don't try to make such comparisons with the NCEP/NCAR index (highest in record, etc). I use them mainly to provide current day data. A reason is illustrated here. Reanalysis is model output, and the methods vary with time, with varying correction applied. As I understand, they don't usually re-do historic calculations. So going back, different years may be differently calculated. A correction for this may or may not have been applied. Here is an illustration of the substantial difference.

    2. Tanks for the reply Nick.

      I'm not exactly sure what the cci-reanalyzer website is showing for the "CFSR (1979-2015)" time series. I thought that that that the NCEP time series.

      That page allows you to select from different datasets and if you choose any specific month then you can download a *.csv containing the entire monthly time series (in absloute temperature ib degrees C).

      I've now plotted up both your anomaly data and the cci-reanalyzer "CFSR (1979-2015)" time series (converted the temperatures to anomalies using your 1994-2013 baseline). The two time anomaly series are very different, specifically between say 2005-2015).

      Anyway's I'm getting rather sleepy now, so perhaps I could send you a trimmed down Excel spreadsheet showing the differences (I'm kind of the technical opinion that your numbers are correct).

    3. Everett,
      Mine is NCEP/NCAR, which is different from CFSR. I posted here about checking the monthly data with the official version. I think the NCEP/NCAR is much more consistent with the regular indices - there is a comparative plot here. I think the CCI numbers have been adjusted to remove biases - it seems that is expected. I don't know whether that makes agreement with indices better or worse. It seems to be mainly to improve short-term forecasts.

      If you would like to email, it's nastokes and westnet and com and au.

    4. The CFS reanalysis at the UMaine site is CFS until early 2011 and CFSv2 since. There are differences in grid resolution and the surface physics package between CFS and CFSv2. The combined CFS/CFSv2 series has cooling of 0.2 to 0.3C in the 2010/11 period vs other re-analysis and surface temperature series. CFS tracks the other re-analysis products well before 2009, similarly CFSv2 tracks the other series after early 2011 but I wouldn't compare post-2011 to pre-2009. Of course some, who want to show a cooler trend, gravitate to the combined series.


  5. I tried to calculate an approximate value for the GISS Dec 15 temperature by linear regression, as I have done for some previous month. This time the difference due to how and if to account for the annual cycle differences are large approx 0.1 C. So I get different best estimate predictions for the GISS Dec 15 temperature form 1.08 to 1.19, depending on the dealing with the possible annual cycle differences between GISS and TempLS.

  6. BEST lo has reported the latest data, December is 1.070 C, up 0.062 from the previous record in Oct. The year 2015 is 0.784 C, up 0.137 from the previous record in 2014

    1. Since BEST is the first global index to conclude 2015 (except TempLS of course), I present an apples-to-apples comparison, in the spirit of Cowtan et al 2015, of observations vs models .
      I would say that the BEST 12 month running mean is back in the middle of the model track, and will likely continue to rise towards the upper bound for the next 7-8 months..

  7. January 2016 continues to post some big numbers on Nick's NCEP tracking system.