Announcing the Winner of the “Guess the NOAA Anomaly” Contest

The anomaly has been released for May. The answer is: 0.5225.

Drum roll, please…

Congratulations to Bob Montle, who said:

My guess is +0.53

Not from the actual graphs but because the number seems to come out .5x(something) regardless of what common sense and intuition suggested before the actual posting.

For his efforts, he receives an “atta boy!” Great credit goes out to him for eschewing the urge to review maps and ocean data, like yours truly, and trying to make sense out of the temperature deviations on the maps and the anomalies released from satellites in order to hone in on the much-anticipated NOAA anomaly.

Clearly, he understands this process much better than the average layman, who might think that such indicators might make a difference.

Despite the fact that I was remarkably low in my guess, I will note that I followed up with the following in the comments:

My prediction is based on the observable maps and recent stats. If someone were holding a gun to my head and my life depended on it, I’d actually predict around 0.5, based on (1) the last few months of anomalies, (2) the fact that NOAA always seems high (I’m sure this is pure coincidence), and (3) the land masses missing will probably be assumed to have about a 5 degree Celsius anomaly.

But it really wouldn’t be all that imaginative to just pick an anomaly in line with the last 4 months. This way, if I’m right I look like a genius. If I’m wrong, I can chalk it up to conspiracy theory or something.

Oh, and the GISS anomaly was 55. It makes me laugh.

HadCrut has yet to be released.

April 2009 Update on Global Temperature – GISS

VACATION ALERT: I will not be posting for the next couple of weeks. The family and I are driving cross-country in a big honkin’ white Cargo Van. We are going to Delaware. “Why Delaware”" you may ask. It’s a valid question. I won’t get into all the details, but let’s just say that we choose our vacation destinations by drawing them out of a hat, under the theory that every place has something to offer. That theory is about to be tested… Anyway, just wanted to let you know of the upcoming hiatus.

On to temperatures…

GISS is always an interesting data set to review, because you’re never quite sure what you’re going to get. In a world where people like consensus, it is somewhat striking to me that we have general agreement between the RSS and UAH, and even the HadCrut data sets, and whenever there is an outlier it’s likely to be the GISS data set. Yet, when people quote statistics on global temperature, the GISS is the preferred set. This, despite numerous documented algorithmic data adjustments that have, over time, a non-negligible impact. Also, there is the real question of whether or not the bias of James Hansen enters into the evaluation of the data.

Nonetheless, it’s the reality we deal with that this is data that is looked to for policy decisions. So, if for that reason alone, it is worth keeping an eye on it so we can speak intelligently to what it is telling us.

First, let’s review the most recent data point:

March anomaly = 47 (in terms of .01 degrees Celsius). This was the lowest March anomaly since the year 2000. It was 18 lower than the March 2008 anomaly, but 6 higher than the February 2009 anomaly. This is quite different than UAH, where the movements were reversed.

Streak: This month’s lower year-over-year anomaly broke a streak of 6 consecutive months where the anomaly was higher year-over-year.

Rank: March 2009 ranks as the 11th highest anomaly in the data (since 1880, or 130 anomalies. 8.5 percentile.) Overall it ranks as the 95th highest anomaly in the data set of 1,551 values (6.2 percentile).

Average: The 12 month average ticked down from a recent high last month, and stands at 46.8, which puts it right at the level it was in January.

SLOPES

Overall slope since 1880 is 0.04700 hundredths of a degree warming per month. For those of you who prefer simpler numbers, that translates to 0.564 degrees Celsius per century.

We can extend the current trend line back to December 2001 that indicates flat temperatures (slightly negative).

The period of time where a best-fit line can be drawn that indicates no change in global temperature is not 100 months long. The last time we had a stretch of 100 consecutive months where a negative trend line could be fit was the period beginning April 1988. In recent history, during the stretch of time for which we consider the warming to be measurable, the longest stretch where a horizontal/slightly negative trend line occurs is from January 1987 through April 1997. That spans 10 years and 4 months. If our current front anchor remains at December 2000 as a starting point, we will not reach that until March 2011 – 2 years from now. So we still have a ways to go to declare that this is beyond any previous hiatus in temperature rise during a potentially warming period. It could be less time than that if some cooler temperatures are forthcoming, driving our starting point back in time a bit. Supposing we would reach that point, the previous time a flat or negative trend line of that stretch occurred is the period beginning June 1969. At this point, it seems get pretty iffy whether or not the trend line can be explained away. But we’re not at that point quite yet.

Here are some notes on the different trend periods:
60-month: Current slope = -0.1123 (-1.35 degrees Celsius cooling per Century). Kind of hovering around the same level as the last four months (ranged from -0.1086 to -0.1151). Here is the comparison to its peak slope, period ending May 2007.

February 2009 Update on Global Temperature – GISS

I didn’t look at GISS last month, so here’s a quick update for the January month-end numbers:

The overall trend since January 1880 has a slope of 0.047, which corresponds to warming of 0.564 degrees Celsius per Century.

The furthest back we can go to see a flat or cooling trend is January 2001, 8 years and a month.

The stats:

1. The anomaly for January was 52, which was actually not as large a jump as the satellite data showed. 
2. The historical rank of that anomaly, though, still shows it to be the 5th warmest January on record and the 62nd highest anomaly out of 1,549.
3. This is the fifth consecutive year-over year warmer anomaly

 
The rest of the charts show the various slope change charts, which shows how the various slopes have changed since the last peak.

The 60-month slope values have trended down since early 2004, with some cyclical fluctuations on the way. The current slope value of -0.1147 represents 1.37 degrees cooling per century. After a small bounce, this latest value reversed a bit.

The 120-month slope values have trended down overall since early 2002. The current slope value of +0.1535 represents 1.84 degrees warming per century. This is a slight decrease from previous, but in the range where it's been holding for a few months. It is anticipated that this will decline later this year as some low front-end anomalies fall off.

The 180-month slope values have trended down since 2007. The current slope value of +0.1489 represents 1.79 degrees warming per century. This has declined continuosly in that time and is at its lowest value since the period ending December 2002.

The 240-month slope values have generally trended down overall since early 2004. The current slope value of +0.1554 represents 1.86 degrees warming per century. The most recent slope vlaues have dropped quickly after a period of level slopes for over a year.

The 300-month slope values have trended down overall since early 2007, though the trend over the last year or so is actually positive. The current slope value of +0.1579 represents 1.89 degrees warming per century. Other than a dip in mid-2008, the slope level has hoverred around the 0.158 mark since October 2007.

The 360-month slope values have trended down overall since 2005. The current slope value of +0.1339 represents 1.61 degrees warming per century. This has declined steadily for over a year and is now at the lowest value since the period ending September 2002.

December 2008 Update on Global Temperature – GISS

I had intended to do a full write-up on this, but it’s late.  I am a victim once again of an agonizing Packers loss in a game they had no business losing.  Ugh.

But I did get the charts updated for your viewing pleasure, so I’ll get them up here.

What I will note is that the GISS anomaly is 58, and that the October number increased from 55 to 58, as well. According to GISS, the last 5 years have not cooled as quickly as every other temperature measure says we have. Also, while all other temperature measures tell us we’ve been flat dating back to 1997, the GISS flat period only goes back to January 2001.

I just present the data. Others have spent a great deal of time lamenting the differences in GISS results versus other results. So, with that, here are the charts:

November 2008 Update on Global Temperature – GISS Charts

In my previous post, I discussed the jaded history of this month’s GISS anomaly, and some initial verbal comments.

In this post, I present some charts using the updated GISS numbers. The data source is found here. (And I just checked, and amazingly enough the numbers stayed the same from last evening’s post!)

Anyway, here we go…

Overall trend since 1880 is around 0.6 degrees per Century

We can extend a trend line back to December 2000 with a negative slope. All further extensions back are positive.

The current 60-month trend line is negative, though it has increased slightly since last month.

Here is the most recent peak value for a 60-month trend value for comparison against the current trend line.

This chart shows how the slopes have changed since the peak value. We have been in a period of declining slopes.

Here is the 120-month trend line as it currently stands.

This chart shows how the 120-month slope values have changed over time. We appear to be in a period similar to 1914-1947. The current similar period of constantly positive 120-month slopes started in 1979. If the cycle actually duplicates, we would cross into negative territory in 2012.

The decrease in 180-month slopes has been very steady. Note the r-squared fit of 0.9665. However, this is only since the beginning of 2007.

This chart shows the 180-month slope values over time. Note the 35 year period from 1915 to 1950, where slopes remained above zero. The values generally increased until 1945, and then in the next five years declined until they went negative. Our current period of positive 180-month slopes started in 1977. We see the peak value of the current stretch of positive slopes in 2007 at this point. If history repeats itself, it would mean that we will see this trend line go negative in 2012. This works on both the 35 year time frame and the last 5-year decline from the peak time frame.

The above chart shows the 240-month slope values over time. We see a period of entirely positive slopes from 1920-1955 on a 240-month basis. Our current period of positive slopes started in 1979. If the 35-year period is repeated, we would see negative 240-month slopes in 2014. However, the current period's magnitude has remained at its highest level for a longer stretch than the previous cycle. Before, it took 9 years from the peak before we saw negative slopes. The current period seems to be lagging that.

Since people seem to gravitate towards the 30-year trends for some reason, here is the chart showing the current linear trend for the last 30 year period.

The 30-year slope values have declined over the last (almost) four years, as shown above. It's had its phases, but the overall trend is down in the most recent years from a peak value of just above 0.15.

Here is a chart of the 30-year slope values over time. Most of the time since 1926 has been in positive territory. We saw an extended period from 1926 to 1963 with positive slopes. The peak occurred in 1946, so it took 17 years after the peak to see (slightly) negative 30-year trend lines. In the current period, the positive slopes began in 1974. The peak was reached in 2005 at this point, so it's about the same length of time as the previous cycle. So, we shouldn't expect to see a negative trend line for the 30-year period for another 13-15 years. If this would occur, it would require continued cooling period into at least the early 1920s.

All statements above with regard to putting dates to slope values is simply conjecture based on simple observation of the similar previous cycles in the early 20th century. It is not a prediction.

November 2008 Update on Global Temperature – GISS

Well, I’m finally getting to the temperature analysis on GISS. Before I start with that, let’s discuss the interesting goings-on with Hansen’s team, which if you are a temperature anomaly watcher, I am sure you have already heard about. This is well covered in different places already, and for much more detailed reading on what all went down this month please check out these Watts links:
Watts link to initial GISS release and subsequent discovery of issues with the data.
Watts link to the updated GISS data discussion.
Watts link to story on kudos for finding the errors to Watts commenters.
Watts link to interesting discussion on the Russian anomalies and surface stations.

And for the perspective from Climate Audit:
CA link to initial wonderment about the Russian anomalies.
CA link to NASA blaming NOAA.
CA link to discussion on corrected data and a “warm” Siberia.
CA link to questions about other potential issues in Russia.
CA link to discussion on who should be blamed and who loses more credibility.
Final CA link on discussion of what caused the problem in the first place.

I was disappointed that our friend Jeff at The Air Vent didn’t supply his insight on the matter. His lack of poltical correctness combined with interesting statistical analysis makes for some good reading. He was the first one to bring the GISS flap to my attention.

OK, so after you read through all those posts, you’ll know the summary. But for those who want the nickel summation, here it is: the GISS data was originally released and the stunning conclusion was that October 2008 was the warmest October in recorded history. An interesting exclamation given the anecdotes around the globe of early snows and freezing temperatures. Not to mention that the satellite data showed no such thing. Well, some of our geeky friends on the internet decided to actually take a look at some of the data, especially when the Russian map showed temperatures that averaged over 10 degrees Celsius above normal.

After what really was a pretty simple analysis – comparing October temperatures against September temps – we come to find out that Russian temperatures are screwed up. September readings were duplicated to October. NASA had to pull the data, and recalculate. The anomaly dropped from 79 to 58. But then some further digging found more problems and they had to re-re-release it and now the anomaly stands at 55.

The source of the error was actually NOAA, who collects the data and does some initial adjustments to it. GISS then picks it up and does additional adjustments before releasing the final anomaly set.

There are a few questions that can be asked here: (1) Was there anything intentional here, to artificially elevate temperature readings; (2) who was at fault; (3) is there a good excuse for this occurring; and (4) does this call into question the entire data set? There are numerous other questions, but I’ll address these three.

First, as much as I appreciate the skepticism surrounding Hansen and GISS, let’s be fair here with regard to intent to deceive. I don’t consider these guys above spinning the data in a way that is deceiving, nor do I think they are above incorporating “adjustments” that work in favor of their preconceived hypothesis of warming temperatures. But spin is spin, and the adjustments are available for public scrutiny – and in fact have been scrutinized quite fairly. We can arguye about the validity of these and different people will differ in their opinions. But outright purposeful introduction of incorrect data is quite a different animal, and I don’t believe at all that this was intentional. First of all, it was so plainly obviously wrong that only a complete idiot would attempt to manipulate data so stupidly. So, one really has to believe that this was an unintentional oversight.

As for fault, it would appear that the inital and primary fault lies with NOAA. However, Hansen and team cannot pass all the blame. I find it remarkable that there is such a paucity of testing of the data that such a simple error could pass by both organizations. Particularly when they produce anomaly maps that were so extreme, wouldn’t it seem reasonable to just check the numbers for reasonableness? If I were to overlook such a glaring data error – whether I was the source of the data or not – my boss would have my head. This was negligence on both ends. It is all the more disconcerting that data quality such as this is lacking when the data set produced is such a widely used one. To answer the third question, there is NO good excuse for this having occurred. It is an embarrassment to both organizations.

Finally, it certainly does call into question the data. But this is not new. The adjustment process that continues to change historical temperatures going back a century have been questioned many times. The lack of surface stations in some areas, requiring extrapolation and smoothing of temperatures over thousands of miles in some cases reduces the data quality. But this latest debacle adds the element of data quality and control. Clearly, this has been lacking. If there is insufficient testing to pick up a duplication of records from one month to the next, then one can only surmise what other errors are in the data that are far less easy to detect.

In the end, though, whether we like it or not, this data is used widely. As such, I produce the temperature charts associated with it.

I’ll keep it a little shorter on this post since the most interesting part of this month’s GISS data is the story above. But here are the stats:

Anomaly = 55
Rank, overall = 49th warmest anomaly of 1,546 monthly readings; 5th warmest October, of 129.
The anomaly is basically in line with September (50) and previous October (54).

The 12-month average increased a blip from 41.0 to 41.1. Because past anomalies continue to change, the previous average in this set differs from the average I stated last month. It’s all kind of maddening, actually, but it is what it is.

Interestingly, even though an anomaly of 55 is a historically high one, it still keeps current trends flat. In fact, the period for which we can extend a negative trend line back is now to December 2000. So, even the GISS data set is approaching an 8 year period where a negative trend line can be fit. This extension backward occurred even though we are now in a neutral ENSO state.

Quick hits on the various trend lines: The 60-month trend is negative, of course, but the magnitude diminished with this anomaly. The 120-month trend line is positive but still declined, however. But this should jump around a bit in coming months. The 180-month trend is positive as well, but also declined to its lowest level since the period ending February 2003. The 240-month trend continues to stay at the slope range it’s been in since early 2007, and the 300-month trend is in the same range as it’s been throughout 2008. The 360-month trend line, though, continues to decline and is at its lowest level since the period ending August 2002.

Time has somewhat gotten away from me. The orange light is flashing on my laptop and I am ready to get some shut-eye. I do have the charts updated, but I will upload them tomorrow in a separate post.