Friday, August 21, 2015

NOAA says July hottest month ever

here but I wish they wouldn't. In fact the anomaly was down from 0.87°C in June to 0.81°C. Oddly, that change is exactly what TempLS mesh now shows, while TempLS grid has the same drop (0.04°C) as GISS. Usually it is the other way around.

OK, so it is a warm year, and this was still the warmest July on record. The NOAA claim that it is the warmest month ever (also Tamino) is based on the annual cycle of absolute temperature, whereby ocean-cominated SH summers are cooler than NH, with less seasonal variation.

Why is this a silly point? The NOAA has a sensible discussion here on the reason for using anomalies in preference to absolute - see point 7. Yet they don't seem to be able to stick to it. They keep lapsing into quoting an annual absolute global temperature, and of course regularly quote a ConUS absolute average.

And it just gets them into trouble, pointlessly. The global absolute is got by adding the anomaly to an annual climatology (14°C) taken from a Phil Jones 1999 paper. But the average anomaly is known rather well, the climatology very poorly in comparison. So the sum is worth far less than the parts. Every now and then, a  sceptic raises the 1997 estimate of 62.45°F (16.92°C) for that year and says - see! the world has cooled over 2°C since. NOAA has been forced to add a feeble disclaimer to the 1997 report. But the sceptics are right (for once) to point this out. It just makes the NOAA look dumb. And of course the troubles caused with the absolute average for ConUS (in clumsy hands) are innumerable.

Back to July - we knew that the global absolute has that seasonal  cycle. It doesn't mean anything in terms of climate change, and isn't news. March had a very high anomaly, July less. But July will always exceed March in absolute.

That's one of the main things about anomaly - it contains the news. The information content about weather and climate change. If I tell you that it was 17°C here yesterday, you won't be impressed. The natural question is - what is it normally? Ie, what is the anomaly? And then you find that it is indeed quite warm for an August day.

You can see this news issue in a temperature map. If you see an absolute temperature map for July 2015, it looks like any other July. Sure, it tells you that Melbourne Fl is warmer than Melbourne Australia, and much other climatological information. But it doesn't tell you much about July 2015. For that you need the anomaly map.

NOAA knows all this. I just wish they would stick to it.

ps In other news, August so far is pretty warm. And Arctic Ice is still melting, with 2015 chasing 2011 for third place. Antarctic ice has entered a freezing pause, which may be linked to ENSO.






47 comments:

  1. You raise a good point. This may have seemed like an appealing idea to NOAA (hottest month ever!) and has certainly attracted a great deal of attention, but isn't as substantial a benchmark as has been hyped. A related lack of consistency also led to the scramble to explain the statistically-insignificant short-term temperature trends for the years following 1998.

    ReplyDelete
    Replies
    1. Yes, I agree about the short term. Consistency is important. And again, we have various explanations coming through for the supposed pause, just as the record warm readings roll in.

      Delete
    2. Attention may be a Very Good Thing (tm), even if the "Warmest Month isn't that significant an event, in that it puts global warming back in the news, and for the most part as a segment or article that is reported as fact, without the denialist talking points for "balance".
      I fully expect to wake up tomorrow to the Facebook posts of the denilaist talking points. It's be something about "How dare they say that(!) if the temperature was within the error bands of previous years".

      Delete
    3. One needs attention when saying things that are correct and important. Misplaced attention is a problem. I think the proper big story of the next 12 months will be the high temperatures of Nov-Apr, in contrast to the modest increase so far. But when we point out that Jan is very hot, it would be open to a sceptic to say, hang on, you said July was hottest ever. Is it hotter now?

      Hard to explain.

      Delete
    4. Measuring temperature variability exactly is a non-issue when the primary contributing factor is ENSO (and volcanic activity when it occurs). The real big scientific story will be the progress made in being able to accurately ad deterministically model the behavior of ENSO.

      So when the realistic ENSO models are in place and we can explain precisely the peaks and valleys in the temperature record, the focus will transfer on looking at the overall trend. Every curve will show temperature trends that are properly compensated for these natural artifacts, and the GCMs can finally become calibrated for the deterministic response.


      Delete
    5. If you can calibrate for the ENSO cycle (accurately model it), effectively you'd produced an anomaly series that doesn't depend this relatively short period but large amplitude variability.

      That goes in hand with what Nick is saying about sticking with the deseasonalized records.

      Delete
  2. Personally, I also prefer to stick with the anomalies, but I do not think that the uncertainty in whether this was the hottest month ever recorded is as large as the uncertainty of the absolute values themselves. Would not the relevant uncertainty be the uncertainty of the annual cycle? Which I would expect to be much smaller.

    ReplyDelete
    Replies
    1. Well, if you assign a number to the July absolute temp, that number is very uncertain. So NOAA said that the 2015 July average was 61.86°F, while they said in 1997 that the annual average was 62.45°F. And if you don't assign a number, it's harder to argue that July 2015 was the hottest (absolute), even though the seasonal cycle is known fairly well.

      But it is also a poor indicator of warming. It's using only July as a measure, since no other month is likely to overcome the seasonal barrier. So the coming hot ENSO months won't count, in that metric.

      And the seasonal cycle contains no information about the particulars of July 2015. If a global anomaly is high, that makes it somewhat more likely that I will experience a warm month. The information that the calculation process for the global absolute makes July warmer that January is no more helpful than telling me that July is aphelion (though it is indeed cold here then).

      Delete
    2. Yes, I think that is the main argument, the absolute temperature is a "poor indicator of [long-term] warming". That is the main reason to prefer to use anomalies. The seasonal cycle does not tell you about long-term warming.

      Delete
  3. The global absolute temperature is difficult to calculate, and this can probably only be done, in a somewhat synthetic way, by reanalysis.
    I checked with NCEP/NCAR data from KNMI Climate explorer ( the only alternative including July 2015) and the warmest month sofar is actually July 2011, the only month above 280 K. But that is global 2 m temps of course, not blended land/SST like the observational datasets..

    ReplyDelete
    Replies
    1. Sorry, something was wrong with the settings, the temperature of NCEP/NCAR July 2015 is 289.33 K, or 16.18 C (but still the warmest)

      Delete
    2. So Olof, to be sure here what you are saying, July 2015 is also the hottest month in the NCEP/MCAR +2m global data as well?

      Delete
    3. Sorry, a final correction, I hope :-) July 2011 is the warmest (289.33 K), July 1998 no 2, July 2009 no 3, and July 2015 no4 in NCEP/NCAR

      Delete
    4. Olof

      So just to be sure, July 2011 is the hottest ever month (on record), followed by July 1998 as the second hottest ever month, July 2009 is the third hottest ever month and July 2015 is the fourth hottest ever month. So ALL the top four hottest ever months occur in July? Is that correct?

      tonyb

      Delete
    5. I gave some numbers for anomaly here (table at bottom). It only goes back to 2010, and you can't compare different months for absolute, but you can compare July's. I got 0.24 for Jul 2011 vs 0.164 for 2015. So yes, looks like 2011 was the hottest, at least since 2009.

      As I've mentioned elsewhere, I think the NCEP/NCAR index may be lagging somewhat in times when SST is the driver, since it is air only.

      Delete
  4. Tonyb, here are top 6 on the NCEP NCAR list:
    1. July 2011 289.3297
    2. July 1998 289.3175
    3. July 2009 289.3168
    4. July 2015 289.2589
    5. July 2005 289.2529
    6. Aug 2005 289.2339

    From climatology July is 0.165 warmer than August, so it is not easy for an August month to become no 1. But this is just one reanalysis dataset, there are others with other ranks and figures...

    ReplyDelete
    Replies
    1. Oops, one more error, number 6 above should be August 2014, not 2005, but the temp is right. I think I'll stop here before drowning in corrections, there is an excellent service at KNMI Climate explorer for those who are interested...

      Delete
  5. Nick has August at .333 through the 21st. Can it get back below .300 by the 31st?

    ReplyDelete
  6. JCH, despite the recent dip I believe that August will stay safely above 0.30. GFS forecasts suggest a spike the last days of August, maybe as high as 0.45

    ReplyDelete
    Replies
    1. Yes Olof, I saw that. This bucket of bolts we call August 2015 appears ready to step forward and actually act like there is an El Nino.

      Delete
    2. Olaf Olaf Olaf, it has taken a dip below .300, but looks now to be recovering the lost ground.

      Delete
  7. Hadcrut4 July is here now with 0.691, warmest July ever of course.
    Another interesting thing, deniers were complaining that 2014 wasn't significantly warmer than 2010. Now, based on seven months only, the Hadcrut4 2015 mean is outside the 95% confidence interval of 2014 (0.12 warmer), and 2014 is outside the 95% interval of 2015.

    ReplyDelete
    Replies
    1. Simply noticing that two measurements aren't statistically distinguishable doesn't automatically one into a climate denier.

      But theres is still the problem with using a generally agreed to biased measurement of anomalized atmospheric temperature to say that you've achieved a record.

      If you've only achieved a record in a series that is biased, how is that even interesting?

      Delete
    2. Usain Bolt's 9.58 in 100 m is only one realisation with the variables neuromuscular function, wind, temperature, air pressure, etc. How can we determine that Usain Bolt is the fastest runner with 95% statistical confidence? Probably not. Hadcrut has at least done 100 realisations, which makes it possible to compare different years statistically.
      The "warmest year on record" in the popular sense, is that with the highest median value of 100 realisations, set with the rules of Hadcrut4 calculation. Usain Bolt, 9.58, is the fastest 100 m realisation, measured by the rules of IAAF.

      Delete
    3. HadCrut4 is a generally acknowledged biased measure of the mean temperature of the Earth, due to the way it infills missing data. As we know, it effectively replaces missing grid points with the global average, which is a very poor method of infilling for higher latitude locations.

      So could you explain the scientific interest a record in a series that doesn't accurately measure the true temperature, when we have more accurate series? I don't see we've learned anything.

      HadCRUT isn't Usain Bolt. HadCRUT is the fifth fastest guy in the race, and you're comparing how fast he ran in this race, compared to previous ones, and have found he's set a new personal record.

      Anyway, the maximum temperature is at best a poor proxy for the real quantity of interest, which is the rate of warming. That is much more robustly measured using trend estimates.

      Individual data points suffer from the fact that short-period variability can produce large excursions. Noting that you're getting new high temperature records over time is simply a statement that there is an underlying positive temperature trend that is overlayed by short period variability.

      Of course you will see new high temperature records as we go forward: The Earth is warming due to the increase in atmospheric CO2 concentration from human activity. And it will continue to warm as long as we continue to increase the amount of CO2 in the atmosphere.

      And of course these new records will occur in periods where you see large positive swings associated with short-period variability. But that doesn't help you, once you've seen a new record, sort out whether this new record is associated with a sudden increase in the overall trend as opposed to larger than usual excursion in the short-period variability.

      Delete
    4. "Quoted uncertainties are computed by integrating across the distribution described by the 100 ensemble members, together with additional uncertainties described in the HadCRUT4 paper that result from measurement and sampling error and coverage uncertainty"

      Coverage bias not accounted for says Carrick.

      I guess Carrick will notify metoffice of his contradictory findings. In a paper submitted to a journal.

      Delete
    5. Thanks for the concern trolling, but it's already been written by the well known "denialists" Kevin Cowtan and Robert Way.

      Coverage bias in the HadCRUT4 temperature record

      Delete
    6. Carrick missed the point that metoffice esitimates the coverage bias. Check Cowtan and Way figure 6

      http://onlinelibrary.wiley.com/doi/10.1002/qj.2297/epdf

      Compared to the coverage bias estimated by the metoffice...

      Delete
    7. Carrick could of course also check this for C&W:

      ".. The Met Office have been reporting the existence of this bias since 2009, although the issue has not received widespread media coverage. The Met Office also provide uncertainty estimates for their temperature data: our results fall within the 95% confidence intervals of the annual data."

      http://www-users.york.ac.uk/~kdc3/papers/coverage2013/background.html

      http://www-users.york.ac.uk/~kdc3/papers/coverage2013/media_compare12.png

      Concern trolling perhaps

      Delete
    8. ehak, including the uncertainty in comparisons is something we agree on. It's why simply ranking the central values of these indices (as was being discussed on this thread and others) is just nuts.

      That said, it's not clear that the recent upward excursion of HadCRUT will remain within uncertainty bounds of C&W's reanalyses.

      Also, when you compare two series, you don't simply look at whether their uncertainty bands overlap to decide whether they are statistically compatible, so what C&W said in your quoted text is a bit misleading.

      Delete
  8. The ranking is of course interesting when you include the uncertainty bounds. Which includes coverage bias.

    And this from Carrick:

    "But there is still the problem with using a generally agreed to biased measurement of anomalized atmospheric temperature to say that you've achieved a record.

    If you've only achieved a record in a series that is biased, how is that even interesting?"

    Problem for Carrick is of course that the others series also achieves record this year. Including CW. And difference between CW and hadcrut is smaller this year than 2010 which is the former record in CW. That means the recent excursion in hadcrut is more within uncertainty bounds of CW. And the difference between the former record in hadcrut might very well be outside the uncertainty bounds of hadcrut. Most will find that interesting. It tells us something about the temperature this year in comparison with former years. It will be a record "by a mile".

    ReplyDelete
    Replies
    1. ehak, No, it's not a problem for me at all if other series obtained a record for this year. I fully accept that the Earth is warming and that humans are playing a major role in recent warming.

      I advocate using the best data and best methods, and letting that inform us of what is happening. It doesn't help us to use erroneous data to set policy questions, even if it makes some of our tummies feel better to talk the particular series that happen to show the largest warming this month.

      There's just no point to using squirrelly data, be it MBH98 as opposed to more modern proxy reconstructions, or temperature series that are not truly global, like HadCRUT4.

      Delete
    2. This is not about squirreling data. Hadcrut has its uncertatinty estimates and CW with better coverage is well within those estimates. There is no need to downlplay the fact that hadcrut (like the other series) will crush the former record t. Even if you do not find that interesting.

      Delete
    3. "Squirrelly" refers to substandard data and there's every reason to put less weight in substandard data.

      Delete
    4. Of course hadcrut will underestimate the warming because of coverage in arctic. Like other gridded datasets. And btw hadcrut is not the dataset that shows the greatest warming. CW show greater warming than hadcrut.

      But one might wonder why Carrick never discovered the uncertainity estimates for hadcrut. One might even wonder if Carrick has discovered that CW use hadcrut data.


      Delete
    5. Anybody with the reading comprehension of a fifth grader would know I'm aware that C&W uses HadCRUT in one of their reconstructions (the other being the NCDC series).

      The whole point of C&W is to extend HadCRUT and NCDC to global.

      And of course C&W has a faster warming rate because it includes the otherwise missed Arctic coverage (which exhibits well known polar amplification).

      That has never been in dispute. For anybody with a brain anyway, something I'm starting to wonder if you possess.

      The point, as anybody but a complete schlub knows, is warming rates are measured by trends not single monthly temperature values. That's why C&W can be lower than the uncorrected HadCRUT series at the moment, but still exhibit a faster warming rate.

      And it's why only an incompetent scientist would ever advocate looking at monthly temperature records (which are corrupted by unforced variability) to study where the climate is warming or not.

      Delete
    6. What was your concern Carrick? This:

      "I advocate using the best data and best methods, and letting that inform us of what is happening. It doesn't help us to use erroneous data to set policy questions, even if it makes some of our tummies feel better to talk the particular series that happen to show the largest warming this month."

      Bizarre.

      Who has been using hadcrut4 this month to set policy questions?

      And this:

      "But theres is still the problem with using a generally agreed to biased measurement of anomalized atmospheric temperature to say that you've achieved a record.

      If you've only achieved a record in a series that is biased, how is that even interesting?"

      When Olof noted that hadcrut4 was outside the uncertainty bounds of earlier years. Those that includes coverage bias.

      Delete
    7. You gotta understand how Carrick operates. Physics is always secondary to him as long as he can score political points or frame an agenda.

      Delete
    8. Yeah it's clearly bizarre to advocate looking at the best data. And clearly there's a political agenda behind that.

      What a pair of moronic comments.

      Delete
  9. C&W is a truly global index, which is a good thing, but that doesn't mean that C&W is right by ranking 2010 warmer than 2014. There are other "global" datasets, BEST, Gistemp loti, and TempLS mesh, and they rank 2014 no 1. BEST and TempLSmesh (like C&W) are not biased by the seemingly unfair GHCN down-adjustment of Arctic stations, and BEST share the same SST dataset with C&W. Thus, the difference in rank is likely a result of method uncertainty and clearly within the error bounds.

    OT maybe, but I just had a discussion with a climate skeptic elsewhere. He claimed that the satellite datasets had a higher "quality" than the surface datasets (a ridiculous claim stemming from Ole Humlum). I checked with uncertainty ensembles from RSS and C&W. The trend 1979 through 2012 was 1.34 +/-0.68 for RSS TLT, and 1.78 +/- 0.13 for C&W (C/ century, ensemble median +/- 95 % CI). Spencer and Christy have not yet made a similar uncertainty evaluation of their products ( lack of guts or scientific objectivity?).
    Anyway, in the process I noted that the Hadcrut4 trend for the period was 1.62, being outside the 95% bounds of C&W (disregarding possible effects of autocorelation).

    ReplyDelete
    Replies
    1. Olof, I agree that that improved converage doesn't mean that C&W is right. It's just likely to be less wrong than a series that is obviously non-global.

      While like you I think the recent high temperatures in the various indices as evidence of global warming, I still advocate using more robust metrics such as temperature trends. When you look at noisy data, there's always the real danger that you'll misinterpret noise (short period unforced variability in this case) with signal (long-duration secular trend).

      I think the satellite data are better for what they measure accurate, which is global coverage and short period variations. Skeptics and loose marbles like Chris Monckton really like the satellite series, esp. RSS, because it's showing a slower warming rate (similar to how until just recently the people on the other side apparently preferred GISTEMP, because it was showing the most rapid warming rate).

      Anyway, that's why MERRA + HadCRUT makes more sense than MERRA by itself.

      As Robert Way has pointed out, it's a risky business to trust the long-period tend in the reconstructions, and I think this is doubly true for satellite reconstructions, which as you know, have to piece together data from multiple satellites, than often having decaying orbits.

      Anyway, in the process I noted that the Hadcrut4 trend for the period was 1.62, being outside the 95% bounds of C&W (disregarding possible effects of autocorelation).

      Typically adding auto-correlation between the series just makes the statistical disagreement worse. On the other hand, when the series are agreeing "too well" based upon the assumption of uncorrelated noise, auto-correlation in the data is usually an explanation for it. You might have noticed Nick employing an argument based on autocorrelation recently on other threads.

      Delete
    2. I've always liked GISS because it's made in the USA. Nationalism.

      Delete
    3. Hm... BEST, GHCN, RSS and UAH are made in the USA too. What a phony nationalist.

      Try harder.

      Delete
  10. BEST has little to do with the government. It's from Berkeley!! RSS and UAH are not accurate at measuring the GMST. NASA is heavily involved in the defense of the country, as is NOAA. When my father's regiment captured Mt. Suribachi, scientists for an organization that ended up a part of NOAA were moving right up the slopes with the leathernecks. The Marines put up a flag. The scientist put up some sort of scientific device. It was mere yards from the flag.

    ReplyDelete
  11. Well BEST is paid for by the federal government, and depending on the terms of the contract, they may actually own the product.

    But are you saying your father was on Iwo Jima?

    I'm not surprised about the scientist going up with the soldiers. In my experience, many field researchers are adrenalin junkies.

    ReplyDelete
    Replies
    1. On a slightly more serious note, I agree with you on UAH and RSS not being good measures of surface air temperature. Though it's interesting that the recent warming blip hasn't shown up on them (yet). Possibly that's one way to distinguish underlying secular variations from regional scale fluctuations that don't reflect global phenomena.

      But I'm not convinced that the SST products are ideal either (esp. when people start scrutinizing monthly records). As I said at some point recently here, I think the SST has been given too little attention.

      As Nick would say "it's what we have". And it's probably adequate as long as you don't get too carried way.

      Delete