Thursday, December 3, 2015

NCEP/NCAR November, cooler than Oct, but warmer than any earlier month.

The Moyhu NCEP/NCAR index for November was 0.513°C, down from October's 0.567°C, but easily ahead (by about 0.1°C) of any other month in the record.

The warmth first shown in the October index corresponded to the rise later shown in GISS, and I would expect similar behaviour here. On the GISS anomaly base, the November NCEP index was 1.04°C.

18 comments:

  1. Looks like the 2nd highest in GISS is .97C. Could it tie or top that? Probably not. Will soon find out.

    ReplyDelete
  2. Think GISS anomaly will be in the range 0.94-1.00 ( based on regression of NCEP against GISS), so there is some probability that it will be the second hottest month.

    ReplyDelete
  3. If we also asked CFSR:

    Oct:http://www.karstenhaustein.com/reanalysis/gfs0p5/ANOM2m_monthly/ANOM2m_GFS_monthly_equir.php?date=Oct%A02015&file=ANOM2m_CFSR_GFS_1510_monthly_equir.png
    Nov:http://www.karstenhaustein.com/reanalysis/gfs0p5/ANOM2m_monthly/ANOM2m_GFS_monthly_equir.php?date=Nov%A02015&file=ANOM2m_CFSR_GFS_1511_monthly_equir.png

    Is a little bit warmer in Nov then Oct

    ReplyDelete
  4. Yes there is a certain spread in the different compilations of reanalysis data..

    Difference November-October
    Karsten Haustein GFS/CFSR +0.03 C
    Karsten Haustein NCEP/NCAR -0.03 C
    Weatherbell CFSR/CFS v2 -0.04 C
    Moyhu NCEP/NCAR -0.05 C

    Regarding Giss, I believe that Oct will be adjusted to 1.07, due to warmer than preliminarily expected temperatures in Brazil, Greenland, West Antarctica and Iran. Giss November should be safely above 1.0, maybe 1.03 C?

    ReplyDelete
    Replies
    1. Small discrepancies can arise from different base periods. I use 1994-2013, and so the Nov-Oct difference includes the difference in averages for Nov and Oct in that time. If you move to, say, 1981-2010, there will be a slightly different average for Oct and Nov.

      Delete
    2. Just to make that more definite - GISS had means .548 and .58 for Oct/Nov in 1994/2013, and .379 .382 for 1981/200. Differences .032 a,d .003. So you could expect that, all else equal, my NCEP/NCAR index would show a Oct->Nov rise about 0.03°C less (or fall greater) than a Karsten index on 1981-2010. That is partly why I also quote a GISS-based 1951-80 number.

      Delete
    3. Yes, you're right, small differences (a few hundreths of a C) can be expected due to different base periods, or maybe slightly different layers, ie surface, 2 m, or sig 995.
      However, the difference between Karsten Haustein GFS/CFSR and Weatherbell CFSR/ CFSv2 is larger and may be caused by a lack of "GFS Model Bias Correction", explained under Additional Details in this page

      Delete
  5. I've been thinking around .94C for GISS, but 1.03 would be great for my goal for GISS 2015 ending up at .85+.

    ReplyDelete
    Replies
    1. December is a burner so far at http://www.karstenhaustein.com/climate . The first third of December, month-to-date plus forecast, show an anomaly up 0.24 C from the October average.

      Delete
    2. That is just totally congruent with my equilibrium climate sensitivity estimate of 1.000C to 1.001C.

      Delete
  6. Nick, my name is Bryan and this is my first time to comment here. I'm a meteorologist in Texas and I have been tracking the CFSR global temperature anomaly estimates output by the University of Maine Climate Change Institute (UM CCI) on a daily and monthly basis for several months now. I also have been following the CFSR output by WxBell. The UM CCI uses a reference period of 1979-2000, so I take their estimates and normalize them to a reference of 1981-2010 as sanctioned by WMO for climatology (or so I recall reading). WxBell already uses the 1981-2010 reference period as does the University of Alabama at Hunstville (UAH) for satellite estimates.

    The UM CCI estimates output each day are preliminary and are later adjusted after a month or two. So I have applied a linear adjustment (R-square = 0.95) to their daily preliminary estimates for November that should bring the estimates in line with final estimates for earlier months. For July through October they have final daily estimates that I used to compile monthly estimates, although I have found that when they later release final monthly estimates there are small discrepancies, possibly due to differing significant figures used (they have separate monthly estimates through June at this time).

    The preliminary UM CCI estimate for November is 0.54C compared to October at 0.52C (normalized to 1981-2010) and compared to 0.458C and 0.495C respectively from WxBell. WxBell and UAH both show a slight decrease from October to November as you are showing, so it will be interesting to see what the final UM CCI estimates show. The preliminary November UM CCI estimate of 0.54C is the second highest monthly estimate, compared to the highest of 0.56C in January 2007, for records back to 1979. In the WxBell data, the October estimate of 0.495C is second to January 2007 at 0.555C and November ranks fourth behind March 2002 at 0.481C, and December 2003 at 0.466C. My best guess after assessing the many factors leading to uncertainty in these estimates is that they are probably not more accurate than about plus or minus 0.3C to 0.5C since 1979 with an increasingly greater uncertainty for estimates farther back in time. Consequently, I do not have much confidence in small changes and trends less than the estimated uncertainty.

    For more information, see the last several posts on my WordPress Climate Concerns blog where I have also looked at trends and compared estimates from UM CCI, UAH, NCEI, and BEST:
    https://oz4caster.wordpress.com/

    ReplyDelete
    Replies
    1. Bryan,
      Thanks for your comment. I'm glad someone is noting the CCI numbers. I don't know why these always seem to be presented as image only. Olof above also notes a small rise in CFSR for November. It does seem to be a slight outlier.

      I've been reading your site. I'm surprised at the large divergence between historic GFS and other indices. I plotted the NCEP/NCAR index here, and it seemed to track the others fairly well. I'll try to download historic GFS sometime, but it is a huge amount of data.

      Delete
    2. Thanks Nick. Even though I have used the GFS for weather forecasting since its inception, I am not very familiar with how measurement data are used to initialize the model runs and also I'm not familiar with all the work that is done in the "reanalysis". I will check back in with the CCI December results.

      Delete
  7. I would say 0,98 - 1°C for GISS in november. But now I'm looking at this december temperatures : unbelievable. According to NCEP CFSv2, it could be up between +1,10°C and +1,15°C for more month to date in december. Ok, it's only 4 days but GFS predicts more warming for the days to come... It could well reach a ridiculous +1,20°C "GISS equivalent" for the 1/3 month to date from what I calculated.

    ReplyDelete
    Replies
    1. Just saw the NOAA PDO Index is in at .47... way down. The AMO may be up again.

      Delete
  8. Because now TempLSgrid and TempLSmesh are out, I can predict the Nov GISS value to 1.00.

    @Nick:But here in the table the dates are (and maybe the temperatures) now confused.
    http://www.moyhu.blogspot.com.au/p/latest-ice-and-temperature-data.html#L1
    It starts with Dec 11, then Jan 11, then Feb 12 ... Dec 12, Jan 12, Feb 13.

    @KC: Yes, the Dec temperatures are high so far and if it remains, the GISS Dec would be 1.2. But even as the El-Nino seems to decays faster, the GISS Jan 16 and Feb 16 anomaly maybe also above 1.0. And the annual anomaly of the second El-Nino-year 2016 maybe higher than the first, as in 1997/98.

    ReplyDelete
    Replies
    1. SOI didn't look like El Nino in november but it's going down again, though not very fast. So I don't know if El Nino has already peaked or if it will go down again as In january 1998 and jan 1983.
      If you adjust temperatures of 1997/98 and 1982/83 to 2015 mean, El Nino would have peaked to about 1,2°C in feb 1998 and 1,15°C in jan 1983. So maybe the "giss equivalent" I was talking about is not so weird... If not in december, maybe in january or february 2016 indeed.

      Delete
    2. Not SOI. MEI performs lower than I expected in the last month. In 1983 and 1998 the value was above 2.0 for several month. I think MEI will decay faster in 2016. So with this and the warm Dec 15 it will be more difficult for 2016 be warmer in GISS than 2015. Now I expect 1.1, 1.0, and 1.1 for Dec 15, Jan 16, and Feb 16 respectively. This would give 0.86 for 2015. But even with the faster decaying El-Nino 2016 maybe warmer than 2015 and maybe also than 0.9.

      Delete