By Dr. David Whitehouse, the Global Warming Policy Foundation
Today’s Times says, “Nasa analysis showing record global warming undermines the skeptics.” However, a closer look at the information which the Times bases its headline on shows that a combination of selective memory and scientific spin play a large role in arriving at it.
The conclusion is based on a new paper written by James Hansen and submitted to Reviews of Geophysics. The paper released by Hansen has not been peer reviewed, and he admits that some of the newsworthy comments it contains may not make it past the referees.
Hansen claims that, according to his Gisstemp database, the year from April 2009 to April 2010 has a temperature anomaly of 0.65 deg C (based on a 1951 – 1980 average) making it the warmest year since modern records began. It is a fractionally warmer than 2005 he says, although an important point to be made is that statistically speaking, taking into account the error of measurement and the scatter of previous datapoints, it is not a significant increase.
The Nasa study said: “We conclude that there has been no reduction in the global warming trend of 0.15-0.20 deg C per decade that began in the late 1970s.”
This is a selective use of a trend line that joins a datapoint in the late 1970’s with the most recent one ignoring the details in the data inbetween. The fact is that one could have taken a datapoint a decade ago and tied it to the same point in the late 1970’s and deduced an even greater rise in temperature per decade. So another way of describing the data is that the rate of increase has actually declined.
Another point to be made is that an increase of 0.2 deg C per decade, if it is real and sustained, is 2.0 deg C per century, an increase not that unprecedented in the climatic record of the past 10,000 years, and substantially less than the widespread predictions of a higher increase.
In the Times article, the Met Office in the form of Vicky Pope, said that their data showed that the past year was “just below” the 12-month record achieved in 1998. Remember, 2009 annual temperature was, according to the Met Office, statistically indistinguishable from every year between 2001–2008.
Vicky Pope then says that Nasa might be right because the Met Office had underestimated the recent warming detected in the Arctic! There are few weather stations in the Arctic and the Met Office, unlike Nasa, does not extrapolate where there are no actual temperature readings. It is curious to hear this given the criticism that Met Office scientists have expressed in the past about the way the Gisstemp dataset is pieced together this way!
Vicky Pope does say however that, “the Met Office continues to predict that 2010 is more likely than not to be the warmest calendar year on record, beating the 1998 record.” This is also a curious statement since she adds that Met Office analysis showed that the four months to the end of April were probably the third warmest for that time of year.
In only the past few weeks however the Met Office has been saying something different.
In the Sunday Times of May 23rd Vicky Pope says that 2010 could be the hottest year on record due to the current El Nino. She also says that the 2010 January – April temperature was the seventh warmest on record meaning that out of the past ten years (allowing for the 1998 El Nino) most of them have been warmer during the January – April period, though not statistically so.
In the Sunday Times article Kevin Trenberth, head of climate analysis at the National Center for Atmospheric Research in Colorado, adds what is missing from the article mentioned earlier: “We have seen rapid warming recently, but it is an example of natural variation that is associated with changes in the Pacific rather than climate change.”
In the Times article poor journalism is compounded with scientific spin from James Hansen’s article to give a misleading impression about the state of the science and what the data actually shows. It will be interesting to see if 2010 breaks any records in the Gisstemp or Met Office datasets. If it does the next question to ask would be, is it statistically significant as one would expect the occasional high point due to errors of measurements causing measured datapoints being scattered around a constant mean (the case post 2001). It would be highly misleading and scientifically fraudulent to look at one datapoint that is higher than the rest yet within the error bars of the previous years and say, “look, a record.” This will not undermine the skeptics but science itself.
Feedback: david.whitehouse@thegwpf.org
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
The attached graph illustrates my previous post.
http://www.woodfortrees.org/plot/hadcrut3vgl/from:2000/to:2011/trend/plot/hadcrut3vgl/from:2000/to:2011/plot/uah/from:2000/to:2011/plot/uah/from:2000/to:2011/trend/plot/gistemp/from:2000/to:2011/plot/gistemp/from:2000/to:2011/trend/plot/rss/from:2000/to:2011/plot/rss/from:2000/to:2011/trend
Is 2010 heading for a record? Absolutely!
Will it be record warmth? Absolutely NOT!
Will it be a record of hyperbole and spin? Definitely!
Will my eyes continue to hurt from rolling? Positively!
I was driving home tonight and this story made the 6 pm National news broadcast on the radio. They mentioned NASA but did not reference Hansen in any way. While the people who visit this blog can quite rightly scoff and laugh at such questionable science, main street media audiences are lapping this alarmism up! The general population are buying it because no other opinions are getting through.
Anthony- you can’t get to ‘down under’ quick enough as far as I’m concerned.
BYO sick bag! PS. Have you tried to get an interview on ABC Melbourne Radio while you’re here? KJ
Amino Acids in Meteorites says:
June 4, 2010 at 7:14 am
climatepatrol
June 4, 2010 at 2:30 am
I didn’t know he was retiring. So he’s getting out because he sees the handwriting on the wall that the earth is cooling and no matter what he does to manipulate the data he can’t compensate for it? Or what is his reason given?
He’ll be 70 next year, seems a reasonable age to retire, I expect I’ll retire by then.
Matt V., you picked a relatively short time range, and noise has a bigger effect over a short range. Take a look at a 30 year range (about as far back as I can go to still get satellite data): http://www.woodfortrees.org/plot/hadcrut3vgl/from:1980/to:2011/trend/offset:-.15/plot/uah/from:1980/to:2011/trend/plot/gistemp/from:1980/to:2011/trend/offset:-.24/plot/rss/from:1980/to:2011/trend
The agreement for RSS, GISS, and HADCRUT is remarkable. Only UAH is a somewhat off from the others. The big picture says that Hansen’s work is well done, and he does not deserve the excoriation he receives from followers of this blog.
matt v. says:
June 4, 2010 at 7:43 am
The attached graph illustrates my previous post.
http://www.woodfortrees.org/plot/hadcrut3vgl/from:2000/to:2011/trend/plot/hadcrut3vgl/from:2000/to:2011/plot/uah/from:2000/to:2011/plot/uah/from:2000/to:2011/trend/plot/gistemp/from:2000/to:2011/plot/gistemp/from:2000/to:2011/trend/plot/rss/from:2000/to:2011/plot/rss/from:2000/to:2011/trend
——————–
Again your comparisons are invalid because you have plotted the four temperature anomaly series without adjusting them to a common baseline period. Read ” Comparing temperature anomalies – getting the baselines right” in the Notes section at http://www.woodfortrees.org
Below is a link to a chart showing 1980-2010 trends for the four series on a common baseline period:
http://www.woodfortrees.org/plot/gistemp/from:1980/offset:%20-.24/trend/plot/hadcrut3vgl/from:1980/trend/plot/rss/from:1980/trend/plot/uah/from:1980/trend
Wren says: June 4, 2010 at 10:44
Again your comparisons are invalid because you have plotted the four temperature anomaly series without adjusting them to a common baseline period.
Below is a link to a chart showing 1980-2010 trends for the four series on a common baseline period:
http://www.woodfortrees.org/plot/gistemp/from:1980/offset:%20-.24/trend/plot/hadcrut3vgl/from:1980/trend/plot/rss/from:1980/trend/plot/uah/from:1980/trend
Wren, can you post anything without fudging the data? Why the GISSTEMP offset?
Owen says:
June 4, 2010 at 10:13 am
Matt V., you picked a relatively short time range, and noise has a bigger effect over a short range. Take a look at a 30 year range (about as far back as I can go to still get satellite data): http://www.woodfortrees.org/plot/hadcrut3vgl/from:1980/to:2011/trend/offset:-.15/plot/uah/from:1980/to:2011/trend/plot/gistemp/from:1980/to:2011/trend/offset:-.24/plot/rss/from:1980/to:2011/trend
The agreement for RSS, GISS, and HADCRUT is remarkable. Only UAH is a somewhat off from the others. The big picture says that Hansen’s work is well done, and he does not deserve the excoriation he receives from followers of this blog.
You two guy’s must think you’re fooling someone. But that’s the alarmists’ gameplan, isn’t it. Delete the offset.
And don’t tell me to “show trends”. It was done to reduce the temperature discrepancy between satellite and surface.
Owen says:
June 4, 2010 at 10:13 am
The agreement for RSS, GISS, and HADCRUT is remarkable. Only UAH is a somewhat off from the others. The big picture says that Hansen’s work is well done, and he does not deserve the excoriation he receives from followers of this blog.
_________________________________________________________________________
This graph says Hansen does deserve the excoriation he receives: http://i31.tinypic.com/2149sg0.gif
Wren
There is nothing wrong in showing the four different temperature dataset’s spread for the last 10 years . The 10 years was picked because NASA was similarly arguing that the last decade was the warmest decade. They did nat say that there was a lot of static in their 10 year data.I was merely showing that there is spread in the latest data sets for a similar 10 year period . You picking the last 30 years , that is a different time frame and the spread is different again.Perhaps what applied the last 30 years , no longer applies the last 10 years . Your argument that data sets are similar for both periods is not valid. They two sets of analysis for two totally different timeframes. I am not saying your time frame is wrong , just that it does not invalidate mine either.
Tim Clark says:
June 4, 2010 at 11:09 am
Wren says: June 4, 2010 at 10:44
Again your comparisons are invalid because you have plotted the four temperature anomaly series without adjusting them to a common baseline period.
Below is a link to a chart showing 1980-2010 trends for the four series on a common baseline period:
http://www.woodfortrees.org/plot/gistemp/from:1980/offset:%20-.24/trend/plot/hadcrut3vgl/from:1980/trend/plot/rss/from:1980/trend/plot/uah/from:1980/trend
Wren, can you post anything without fudging the data? Why the GISSTEMP offset?
======
Because if you want to compare anomalies that have different baseline periods, you need offsets to put them on a common base. Without a common base, the comparisons will be wrong.
The baseline periods for the four temperature anomaly series are:
UAH and RSS — Jan 1979 – Dec 1998
HADCRUT — Jan 1961 – Dec 1990
GISTEMP — Jan 1951 – Dec 1980
The offset adjustments (- 0.15 for HADCRUT , and – 0.24 for GISTEMP) put these two series on a common baseline period with UAH and RSS. For more on the subject, see “Notes” at http://www.woodfortrees.org
Now I see what was bothering me. You forgot to offset Hadcrud.
matt v. says:
June 4, 2010 at 12:06 pm
Wren
There is nothing wrong in showing the four different temperature dataset’s spread for the last 10 years . The 10 years was picked because NASA was similarly arguing that the last decade was the warmest decade. They did nat say that there was a lot of static in their 10 year data.I was merely showing that there is spread in the latest data sets for a similar 10 year period . You picking the last 30 years , that is a different time frame and the spread is different again.Perhaps what applied the last 30 years , no longer applies the last 10 years . Your argument that data sets are similar for both periods is not valid. They two sets of analysis for two totally different timeframes. I am not saying your time frame is wrong , just that it does not invalidate mine either.
====
The main issue is you are comparing different anomalies without putting them on a common baseline period. By not doing so, the differences you see are largely a result of difference in the bases rather than real differences.
Offsets adjustments for putting the four anomalies on a common baseline are explained at http://www.woodfortrees.org
From the linked graph with trend lines for the different anomalies already on a common baseline, you can insert any starting year you like. If you start with the year 2000 and compare that graph with your previous results, you will see a very different picture.
http://www.woodfortrees.org/plot/gistemp/from:1980/offset:%20-.24/trend/plot/hadcrut3vgl/from:1980/trend/plot/rss/from:1980/trend/plot/uah/from:1980/trend
Tim Clark says:
June 4, 2010 at 12:40 pm
Now I see what was bothering me. You forgot to offset Hadcrud.
—————–
Tim, you are right. I forgot to offset hadcrut. The link to the corrected chart is
http://www.woodfortrees.org/plot/gistemp/from:1980/offset:%20-.24/trend/plot/hadcrut3vgl/from:1980/offset:%20-.15/trend/plot/uah/from:1980/trend/plot/rss/from:1980/trend
wren
I get the same spread with your offsets. http://www.woodfortrees.org/plot/gistemp/from:2000/offset:%20-.24/trend/plot/hadcrut3vgl/from:2000/offset:%20-.15/trend/plot/uah/from:2000/trend/plot/rss/from:2000/trend
matt v. says:
June 4, 2010 at 1:53 pm
wren
I get the same spread with your offsets. http://www.woodfortrees.org/plot/gistemp/from:2000/offset:%20-.24/trend/plot/hadcrut3vgl/from:2000/offset:%20-.15/trend/plot/uah/from:2000/trend/plot/rss/from:2000/trend
=====
No, the magnitude is different. In the graph with the offsets, the spread starts at about 0.07 in 2000 and ends at about 0.09 in 2010, a difference of 0.o2. In your graph without offsets from your earlier post(see below), the spread starts at about 0.27 in 2000 and ends at about 0.35, a difference of 0.08.
http://www.woodfortrees.org/plot/hadcrut3vgl/from:2000/to:2011/trend/plot/hadcrut3vgl/from:2000/to:2011/plot/uah/from:2000/to:2011/plot/uah/from:2000/to:2011/trend/plot/gistemp/from:2000/to:2011/plot/gistemp/from:2000/to:2011/trend/plot/rss/from:2000/to:2011/plot/rss/from:2000/to:2011/trend
The notion of offsets is as basic as it gets. I’m surprised that so many here have difficulty understanding what offsets are and why they are necessary for comparisons.
The big picture remains: 4 different calculation methods of an average global temperature from two very different types of data (direct on-site thermometer readings and microwave radiances) are all in substantial agreement. The temperature record is solid, and it is best interpreted over the long term.
WREN
I can see where the diffrence has risen in our posts . I started my original article comparing “the least square trend line slopes” which later seemed to have morphed to “spreads ” of the four different data sets. The least square trend line slopes of the four different data sets remain as per my original article for the 10 years in question and they are different.
Regardless, GISS fills in Gaps with the positive anomaly, in addition, I agree with JB that
“They have adjusted down non-satellite-era temperatures (1951-1980) and are now comparing them to current satellite-era temps AND THEN CLAIMING IT’S THE WARMEST EVER.”.
Also, GISS deleted (or at least I heard) the arctic temp anomalies. What would be their motive behind that? Sounds Like they’re covering up the evidence. After all, we all know AGW is all about the money……right……?
matt v. says:
June 4, 2010 at 4:50 pm
WREN
I can see where the diffrence has risen in our posts . I started my original article comparing “the least square trend line slopes” which later seemed to have morphed to “spreads ” of the four different data sets. The least square trend line slopes of the four different data sets remain as per my original article for the 10 years in question and they are different.
====
The slopes before the offset adjustment are different than the slopes after the offset adjustment. Note that the RSS trend line moves from below the HADCRUT trend line to above it after the adjustment. Perhaps if you can print the two graphs, the differences would be easier to see.
The four anomalies have less similar temperature changes in the short-run than in the long-run, as we have seen by comparing the offset-adjusted trends for 2000 – 2010 and 1980-2010, and as can be seen by comparing the trends for either period to the changes in the last 12 months.
http://www.woodfortrees.org/plot/hadcrut3vgl/last:12/offset:-0.15/plot/gistemp/last:12/offset:-0.24/plot/uah/last:12/plot/rss/last:12
keep working Wren, keep working, maybe you’ll eventually convince at least yourself
matt v. says:
June 4, 2010 at 4:50 pm
WREN
I can see where the diffrence has risen in our posts . I started my original article comparing “the least square trend line slopes” which later seemed to have morphed to “spreads ” of the four different data sets. The least square trend line slopes of the four different data sets remain as per my original article for the 10 years in question and they are different.
That’s why you shouldn’t look at climate trends over short periods (e.g. <15 years). Each dataset can be influenced slightly differently by different events. For example, satellite measurements respond strongly to ENSO events. If you choose a short time period which starts with La Nina and ends with El Nino (or vice versa) you can get a misleading impression of the underlying trend. GISS extrapolates over the arctic which means that during an arctic warming period, the GISS trend will be be higher than Hadcrut, say. Of course, this should all even itself out because as most posters on here are fully aware the sun is about to enter a grand minimum which will presumably result in the sort of arctic cooling which occurred between 1940 and 1970 (See GISS record).
In any case, as has been shown by a number of posters, these short term discrepancies between the datasets have very little influence on the longer term trends. The trends for all 4 main datasets are remarkably similar and consistent.
Just a quick word on 'offsets'. GISS calculate monthly anomalies which are relative to the 1951-1980 base period (Hadcrut uses th 1961-1990 period). UAH and RSS use the 1979-1998 period. Because 1951-1980 was a colder period than 1979-1998 then, quite naturally, GISS anomalies will be larger than satellite (UAH & RSS) anomalies. If we want to compare GISS (and Hadley) with the satellite readings then we need to use a common base period (i.e. 1979-1998). GISS temperatures for the 1979-1998 period were, on average, ~0.24 deg warmer than for the 1951-1980 period (Hadley was ~0.15 deg warmer than the 1961-1990 period). Note there are slight monthly variations but this gives a mean annual difference. Therefore if we want to compare GISS with UAH we need to knock 0.24 deg off the actual GISS anomaly (and knock 0.15 deg off for Hadcrut).
To see how it works let's look at the April 2010 anomalies, i.e. GISS +0.73; UAH +0.50. Now subtract 0.24 from the GISS anomaly to find the anomaly relative to 1979-98 and we get +0.49 (0.73-0.24), so UAH has an April anomaly which is 0.50 deg above the 1979-98 April mean while GISS has an April anomaly which is roughly 0.49 deg above the 1979-98 April mean. I say "roughly" because the 0.24 deg offset is an annual figure and it's possible that the April offset is slightly different. However GISS provides tools which allow us to check this. The following is an anomaly map for April 2010 relative to the 1979-98 period.
http://data.giss.nasa.gov/cgi-bin/gistemp/do_nmap.py?year_last=2010&month_last=4&sat=4&sst=1&type=anoms&mean_gen=04&year1=2010&year2=2010&base1=1979&base2=1998&radius=1200&pol=reg
The anomaly is in the top RH corner. It tells us the April was actually +0.51 deg warmer than the 1979-98 April mean. GISS is clearly up to no good here.
Amino Acids in Meteorites says:
June 4, 2010 at 11:40 pm
keep working Wren, keep working, maybe you’ll eventually convince at least yourself
======
I’m not alone in being convinced GISSTEMP and HADCRUT are on different baseline periods than UAH and RSS, and that offset adjustments are necessary if these series are to be compared. I think many others here are convinced, including Anthony Watts, and regular contributors Willis Eschenbach and Steve Goddard.
I’m also am not alone in believing comparisons of these four temperature series over short periods can be misleading. John Finn’s post does a good job of explains why such comparisons can be misleading.
Seeking a reply to TFN Johnson (June 3, 2010 at 9:47 am) I’ve been pointed to a global analysis using just 60 stations here: http://moyhu.blogspot.com/2010/05/just-60-stations.html
(ht — J at Open Mind — http://tamino.wordpress.com/2010/04/12/open-thread-19/#comment-42206)
Thoughts?