In Dec 2004, Steig and Schmidt wrote
The Antarctic Peninsula, site of the now-defunct Larsen-B ice shelf, has warmed substantially. On the other hand, the few stations on the continent and in the interior appear to have cooled slightly (Doran et al, 2002; GISTEMP). At first glance this seems to contradict the idea of “global” warming, but one needs to be careful before jumping to this conclusion.
More recently, in Jan 2009, Steig wrote in explanation of Steig et al here
So what do our results show? Essentially, that the big picture of Antarctic climate change in the latter part of the 20th century has been largely overlooked. It is well known that it has been warming on the Antarctic Peninsula, probably for the last 100 years (measurements begin at the sub-Antarctic Island of Orcadas in 1901 and show a nearly monotonic warming trend). And yes, East Antarctica cooled over the 1980s and 1990s (though not, in our results, at a statistically significant rate). But West Antarctica, which no one really has paid much attention to (as far as temperature changes are concerned), has been warming rapidly for at least the last 50 years.
I was intrigued by the statement in Steig et al that “no one really has paid much attention to [West Antarctica ]”. How could this be?
The graphic shown below (Steig Figure 4e) shows temperature trends in Antarctica for 1979-2003 as calculated in Steig et all. The Antarctic Peninsula can be seen going off to the northwest. “West Antarctica” is the “hot” region adjacent to the Peninsula.
Figure 1. Excerpt from Steig Figure 4. Color scale goes from -0.5 to +0.5 deg C/decade.
One of the main theses of Steig et al is that a more accurate accounting of Antarctica yields an overall profit: the profits in the Peninsula and West Antarctica outweigh slight losses/trading dollars in East Antartica. Antarctic temperature trends are not something that I’ve previously taken a position on nor had I familiarized myself with the data. As someone approaching the matter freshly (as I mentioned before), it seems surprising to me that Antarctica wouldn’t have warmed along the rest of the world; it also seems to me that there are modeling issues specific to Antarctica and modelers might well get inconsistent results in this continent without meaning that everything that they ever did was worthless (though it might be worthless on other grounds).
I downloaded GISTEMP binary data and calculated 1979-2003 GISTEMP trends for 1200 km smoothing (250 km smoothing is much sparser) and obtained the plot shown below. (I haven’t figured out how to show the scale on the plotting software, but it’s -.5 to 0.5 deg C/decade (I used the package maps – but will experiment with some of Roman’s code in the future). I required 280 out of 300 months for a trend (the availability fell off very sharply.)
Figure 2. GISS Trends 1979-2003. Color scale goes from -0.5 to +0.5 deg C/decade.
Interestingly GISS uses exactly the same READER data set as Steig – see here where there are references to READER pages. (I’ve matched GISTEMP numbers to all the READER data for future reference.) To the extent that Steig claim that they can get the same result without satellite data, this means that it because they process the data differently than GISS – not that they are using different data.
Both maps show strong warming on the Peninsula, but GISS smoothing did not generate sufficient data in West Antarctica to show a trend one way or the other. Steig and Mann have deduced that this region with missing data in GISS is nonetheless warming strongly. Could be. If this were a mining map, one would certainly look for ore in regions adjacent to known ore and West Antarctica is adjacent to known Peninsula warming.
There are some other interesting differences: the “cold” area from 90-180E is mildly warming onshore in the Steig rendering. Would be interesting to know what accounts for the differences.
But first I’m going to look at what surface and AWS data is available in West Antarctica – to see what the data looks like before it goes into RegEM.
As a further observation on the inconsistency between GISS and Steig results for the Antarctic: Steig describes himself as an isotope geochemist. It’s nice that Steig’s taken time from his busy studies of isotope geochemistry to try to sort out GISTEMP data in the Antarctic – but isn’t this the sort of thing that GISTEMP should have done in the first place?
It would also be nice to hear from GISS on this: what do Steig’s results say about GISTEMP methodology? Steig didn’t analyse why GISS methods seem to have been incapable of discerning recent warming in West Antarctica (an unexpected sort of oversight), but shouldn’t there be some kind of effort at detailed reconciliation.
Climate Audit 
49 Comments
Here’s code for making the GISS map. Requires package mapproj. I haven’t doublechecked for turnkey-ness but it’s copied from my working script and is close to turnkey. I sometimes have things on my console (as readers know). [Note: the h-function is being rendered incorrectly by WordPress; I’l post up ASCII version]
source(“http://data.climateaudit.org/scripts/station/giss/read.binary.txt”)
download.file(“ftp://data.giss.nasa.gov/pub/gistemp/download/SBBX.Tsurf1200”, “temp.dat”,mode=”wb”)
#about 41.5 MB
handle < – file("temp.dat", "rb");
Data <- read_GISS_data_SBBX(handle); #thanks to Nicholas
close(handle);
#length(Data[[2]]) #8000
count= sapply(Data[[2]], function(x) sum(x[[1]]==9999) ) #1123
grid.info=make.grid.info(Data)
grid.info$lat=apply(grid.info[,c("LATS","LATN")],1,mean)
grid.info$long=apply(grid.info[,c("LONE","LONW")],1,mean)
#information about gridcells
temp1=(grid.info$lat =1979) &(time(giss.grid)<2004)
count=apply(!is.na(giss.grid[temp,]),2,sum);hist(count)
hist0=hist(count,breaks=seq(0,300,20))
g=function(x) {temp=(time(x)>=1979) &(time(x)<2004);fm=lm(x[temp]~c(time(x))[temp]);
g=coef(fm)[2];g}
Trend=rep(NA,ncol(giss.grid) );
for(i in (1:ncol(giss.grid))[count>280]) Trend[i]=g(giss.grid[,i])
Trend=10*Trend;range(Trend,na.rm=T) # -1969:20000.4618992 0.4535873
#1979-2003 -0.5947240 0.6810578
M=length(Trend)
#make plot
library(mapproj)
#library(GDD)
col1=c(“#0000BB”,”#0014FF”,”#006BFF”,”#00C2FF”,”lightblue2″,”lightblue1″,”grey85″,
“lightgoldenrodyellow”, “lightgoldenrod”,”#FF8700″, “#FF5B00″,”#D70000″,”#800000”)
K=length(col1)
breaks0=seq(-.5,.5,1/K); #breaks0[length(breaks0)]=max(Trend,na.rm=T)
h= function(x) { if (is.na(x)) h=NA else {
if(x<=max(breaks0)) h= min((1:length(breaks0)) [x<breaks0 ])-1 else h= length(breaks0);
h[h==0]=1}
h}
grid.info$trend=Trend
y=rep(NA,M); for(i in 1:M) y[i]=h(Trend[i])
grid.info$class=y
##PLOT
# GDD(file="d:/climate/images/2009/steig/giss.1200.gif",type="gif",h=420,w=420)
map('world',proj='azequidistant',orient=c(-90,0,0),ylim=c(-90,-60),fill=TRUE,col="white")
a=seq(-180,150,30); for(i in 1:length(a)) lines(mapproject(list(x=rep(a[i],181),y= c(-90:90) )),col=4,lty=2)
a=c(-180,0); for(i in 1:length(a)) lines(mapproject(list(x=rep(a[i],181),y= c(-90:90) )),col=4,lty=1)
a=seq(-50,-80,-10);
for(i in 1:4) lines(mapproject(list(y= rep(a[i],72),x= seq(-177.5,177.5,5) )),col=4,lty=2 )
for(i in 1:M){ polygon(mapproject(list(x=c(grid.info$LONW[i],grid.info$LONW[i],grid.info$LONE[i],grid.info$LONE[i]),
y=c(grid.info$LATS[i],grid.info$LATN[i],grid.info$LATN[i],grid.info$LATS[i]) )),col=col1[y[i]],border=col1[y[i]])
}
map('world',proj='azequidistant',orient=c(-90,0,0),ylim=c(-90,-60),fill=FALSE,add=TRUE,lwd=2) #irrelevant
points(seq(-.5,.5,1/(K-1)),rep(-.45,K),pch=15,cex=3,col=col1)
mtext( seq(-150,150,25),rep(-155,13),breaks0,font=2,cex=.8)
a=seq(-180,150,30); for(i in 1:length(a)) lines(mapproject(list(x=rep(a[i],181),y= c(-90:90) )),col=4,lty=2)
a=c(-180,0); for(i in 1:length(a)) lines(mapproject(list(x=rep(a[i],181),y= c(-90:90) )),col=4,lty=1)
a=seq(-50,-80,-10);
for(i in 1:4) lines(mapproject(list(y= rep(a[i],72),x= seq(-177.5,177.5,5) )),col=4,lty=2 )
title("1979-2003 Trend: GISS 1200 km")
# dev.off()
Is it clear yet whether this study used new data or whether it was a reanalysis of previous datasets?
Steve: Old data. The temperature data is the same as GISS. The AVHRR satellite data was done by Comiso, but re-analyzed.
So if an isotopic geochemist is qualified to do this work what reason do they have for complaining when statisticians take a crack at it as well? Other than trying to protect their turf and results from poaching by outsiders that is.
That’s about as subtle as a brick to the head.
Steve: Huh?? That’s what they said. They said that the warming in West Antarctic offset slight cooling/no change in the rest of the continent, yielding overall a slight warming. “the continent on the whole is warming,” Mann said. “This is the first study to demonstrate that.”
http://www.livescience.com/environment/090121-antarctica-warming.html They said that they did more accurate accounting: “People were calculating with their heads instead of actually doing the math,” said lead author of the study Eric Steig, of the University of Washington in Seattle. “What we did is interpolate carefully instead of just using the back of an envelope.” They are within their rights to try to do a more accurate accounting; I’m within my rights to examine their accounting policies. I don’t understand your beef.
Re: Jon (#4),
You’ve done it before. It’s tired and transparent.
Steve: I don’t understand your objection to the linked post. Is this the sentence that you object to:
Re: Jon (#7), To repeat Steve, what is your beef? Look at the difference, look at the methods of calculation that have been openly published here and then make a clear comment about what you see is wrong. Then we’d all be the wiser.
Re: Jon (#7), etc,
It appears that Jon objects to Steve’s use of financial metaphors and parallels in the accounting of station data. If so, it’s bizarre, since the whole point of this blog(and Steve will please correct me if I have it wrong) is to audit, in the business sense, the data, and the statistical means & methods used in climate research.
Of course, Jon could eliminate the confusion by clearly stating his objection.
Arn,
“The team used this relationship to extrapolate what the satellite would have recorded over the whole continent had it been in orbit when the weather station record began in 1957. Once the group reached the period when the satellite was in orbit, they checked the extrapolated values against the actual measurements to confirm that the method was sound. In the end, they generated a 50-year record of temperatures across Antarctica.”
http://earthobservatory.nasa.gov/IOTD/view.php?id=36736
Extrapolate …
– snip
I fail to see the negative implications of warming on much of the Antarctic continent from a practical point of view. It is simply way too cold, even during the current height of austral summer to melt any of the ice across a vast majority of the continent. I think there is a disconnect between practical interpretation of the study and what is reported in the agenda-driven media: for instance, what does this mean in terms of warming during the austral winter, when the ice-edge circumnavigates the Southern Ocean as far north as -65 degrees latitude? Plus, the study says nothing (also written as NOTHING) about interannual variability from the Southern Oscillation (EL NINO), as well as a host of other natural climate variability.
So, even if there is a profit or loss over some parts of the continent, does that have any physical significance?
There have been many such studies (which I won’t embarrass the authors here) that do not analyze the impact of natural variability as a mechanism affecting temperature, tropical cyclone activity, or cryosphere changes. However, they are NOT ignorant of the impacts, they just can’t tell us yet, and oftentimes flat out say so, “we did not account for natural variability but we think it is … “(negligible, small, big, who knows because we didn’t check). Some of this came through in various interviews in newspaper articles (i.e. Trenberth).
Much of the reason natural variability is such a mystery is related to the inability of climate modelers to recreate much of it, due to sparse data sources or just plain bad model physics, resolution, etc. Thus, this study is a good first start of probably considerable future research on Antarctic climate change. With the stimulus bill’s 400$ million dollars in climate change research, I am sure plenty of funding will be available for out of work climatologists to relocate to Antarctica.
Re: Ryan Maue (#6),
I do not think the authors are concerned about negative or positive implications. They are looking to justify the AGW theory in as many parts of the globe as possible. This will become another hockey stick type of discussion. They will say their stats were correct even if errors are found.
If “East Antarctica cooled over the 1980s and 1990s” how does it end up coloured yellow in Figure 4e (1979-2003)?
Visually, I don’t see any points of similarity between the two maps that couldn’t happen by chance alone. If you assigned trends randomly to the existing stations, rather than use real data, and generated multiple maps, I suspect the resulting maps would have more in common than the two maps here.
So … the East Anatarctic has been cooling since 1969. 40 years of cooling totally contradicts AGW.
However, generated and extrapolated data suggests it warmed from 1957 to 1969.
Steve: I’ve observed elsewhere that Antarctica is unique and presents one-off modeling. It is not my opinion that these outcomes ” totally contradicts AGW”. I ask readers not to get carried away and I do so one more time.
I’m wondering about the -0.5 to 0.5 per decade. Seems like a huge range. Maybe that should be for the time interval 1979-2003?
This GISS map page defines their trend method in the following quote.
Here is a southern hemisphere view of their trend map which I cropped from the polar view of their global trends map. I rotated it 90 degrees clockwise to put the prime meridian running top to bottom.
The .43 figure on the lower right is the global 25 year trend. No SH trend is greater than +-1 from 78S to the pole. The majority are less than +-0.5 over 25 years. At -77s -143w is the closest to the pole with a value greater than +-1 and it’s a huge jump from the adjoining grid-box being 1.9944 compared to the 0.4432 value in box -77s -145w.
In figure 4, huge steps from blue color to yellow or red appear all along the borders of the antarctic. That looks like some error in the processing algorithm.
GISS colors (Figure 2) are continuous at the land/sea border.
Abstract from Chapman and Walsh, “A synthesis of Antarctic temperatures”, Journal of Climate, vol. 20, issue 16, p. 4096, 2007
The paper in its submitted version can be found here. I haven’t seen Steig’s paper, but it would be interesting to compare both papers.
Steve you hyperlinked Doran and Gistemp right at the start. Doran is not available: has it just been removed? – thanks.
snip – policy
Great, we have a data set (how many sensors?) covering a large area called Antarctica showing a relative trend wrt to a chosen time period. I’d like to see some scrutiny of this latest hoopla about Antarctica’s trend and these ‘new’ findings, so thanks for this post Steve. Of course the Mannians take this ‘latest evidence’ as more proof of impending doom.
As a 20yr in the field meteorologist I find this entire exercise of
attempting to measure a global temperature quite amusing; 3000 micro climates maketh not a global representation in my opinion and that is all surface temp measurement is. Of course temp sensors are placed such that they are to represent the general area as best they can, but after looking at A Watts’ site and my understanding of the UHI and micro climes, I just can’t buy these data sets as a representation of vast areas, including planetary.
In any event, it doesn’t matter if it’s warming or cooling in Antarctica, they have both bases covered already, as we are all aware!
Re: tom (#17), What they’re creating is a statistical index that is meant to be representative of the larger reality. Rather like the DOW30 or the FTSE100
Steve Mc:
I get an error with the script
Looks like a syntactic error
Steve: looks like wordpress putting in an extra space
Re: John A (#19),
Yep. If you reformat the statement to if(x <= max(breaks0))… the error message is: Error: unexpected symbol in: “h=function(x) {…}.
I read some comments by Steig over at RC and experienced another BP spike. He is another over the top AGW guy. I bet big money there will be some magic in this math too. Anyone have any good information on RegEM?
GISS’ construction of a 1950s-2000s Peninsula temperature trend must have been quite a chore. Here’s a plot of the years in which each Peninsula station listed an annual temperature (each row is one station and the stations are ordered north to south):
There are (arguably) only three stations along the 800km Peninsula which span the 1950s to the 2000s. The large majority have to be weaved, spliced etc to contribute.
Having seen so much microclimate variability in continuous US station records I admire anyone who tackles the job of estimating a 50-year trend from these discontinuous ones.
Re: David Smith (#22), Re: David Smith (#23), and Re: deadwood (#24), It would seem to me that the data could be of insufficient quality and quantity to make conclusions of any significance in terms of trends. Does that mean that we are left only with the satellite data? Would anyone with expertise in this area, care to comment?
Here’s a partial Figure from the link provided by Francois in #13. It shows the trend in surface temperature, 1958-2002:
The two stations which appear to establish the hot spot are Faraday (5’th from bottom in #23) and Rothera Point (3’rd from bottom). Per Figure 9 in the same paper, the reported warming is concentrated in June thru August (mid-winter). Interesting.
Re: David Smith (#23),
So what happens if the analysis is done without these two stations? Are these two stations the Strip Bark of the Antarctic?
Re: deadwood (#24),
I wonder how much influence 22 people have on temps in terms of waste heat and condensation and moisture.
Rothera Point: “The station is open throughout the year, in the summer the population will peak at just over 100 people. In the winter months, April to mid October, a compliment of around 22 will be continuing the science work and looking after the station infrastructure.”
http://www.antarctica.ac.uk/living_and_working/research_stations/rothera/
Re: David Smith (#23),
Yes, that is interesting. It may correspond with some reported changes in upper air circulation. See:
Espy, P. J. et al. Mesospheric planetary waves over Antarctica during 2002. Received 22 June 2005; revised 21 September 2005; accepted 26 September 2005; published 4 November 2005. GEOPHYSICAL RESEARCH LETTERS, VOL. 32, L21804, doi:10.1029/2005GL023886, 2005.
Can anyone see a significant relationship between the surface temperatures and the upper air activities in the period?
Re: David Smith (#23),
Faraday and Rothera are the two stations with a long continuos record..uhmmm…
David, look at what BAS says of these two stations.
Faraday
Rothera
Really no movement and construction there…..
Anyway, the warming is in winter because west peninsula temperature is influenced by the sea ice cover…no ice, warming…a lot of ice, cooling. In summer the change is less because the sea is always ice free.
The direct effect on temperature of circulation changes are greatly enhanced by the presence or not of sea ice. I bet that what dominates temperature is sea ice.
Re: David Smith (#23),
Re: Paolo M. (#29),
From Anthropogenic heat island at Barrow, Alaska, during winter: 2001–2005 (text from paragraph 33 on page 8 of 12, Figure 6d on page 7 of 12):
(emphasis added)
Re: Phil (#37), Some of the tags didn’t work.
MUHID is actually MUHI(subscript D)
m s-1 is actually m s(superscript -1) or meters per second
Re #24 I don’t know the answer, deadwood. All may be well, or not. I’m going to read some of the earlier papers cited in Chapman et al to see if they discuss the issue.
That “fig 1” looks a little strange to me. The “grids” offshore appear coarser than the “grids” onshore.
If the land area has been smoothed, why didn’t they smooth it all? Is it possible that smoothing the offshore area would have shown a hotspot off the coast?
I refuse to believe that up until now that we’ve been accounting Antarctic tempertures in our head and on the back of envelopes….
rothera…
The Rothera weather station has indeed had some “issues” lately, if this blog describes events accurately:
http://antarctic-diary.blogspot.com/2007_11_01_archive.html (Scroll down to headline “More movement”, below the penguin pictures). First the whole station is completely covered in snow, and then it is mounted on a roof. WUWT should investigate the adjustments;-)
Re: Matti Virtanen (#33),
that blog deals with Halley and not Rothera.
Re: Paolo M. (#35), Sure, my bad. But anyway, it looks like being covered in snow seems to be a problem for all man-made structures in the Antarctic. How do you keep a thermometer at 2m above ground when it snows as much?
Re: Matti Virtanen (#46),
By lifting and replanting the legs as needed to remain on the surface. After so many years of combatting the burial of the South Pole researh stations, they finally placed the facility on legs to remain above the surface. Think of it like a spider who can grip the surface when needed, and lift the legs as needed to reposition them on the surface. Future AWS designs may look more like an interplanetary lander with spider like legs. By digital camera and remote command, they would operate the legs to maintain position on the surface.
The h-function in the script in #1 is being rendered incorrectly by WordPress; I’ll post up ASCII version]
So the recent trend, since 1969, has been cooling in eastern Antarctica. This report is actually their discovery that 1957 was colder than 1969. So it helps confirm the 1945-1975 cooling period. Did they examine whether their 1957 temperature matches what was previously known about the cooling period (and how models imagine it happened)?
Phil:
An interesting article. Thanks for the link. Is there anything more recent?
Re: bernie (#39), I am not aware of anything more recent, but that article was published in the JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, D06118, doi:10.1029/2006JD007837, 2007, which I trust you would consider to be pretty recent.
First David Smith comment: Thanks for your sleuthing and the resultant graph. Interesting and informative–much appreciated. Maybe it’s how one defines “span”, but my eyes count two, not three, stations that span the period of the 50’s through the 90’s. At any rate, any surface peninsula temperature constructions seem not to be made of whole cloth. Again, thank you.
A HOT SPOT means invariably contamination of the data. Any study of worth should flag up that possibility and either account for it’s impact on results or discount it all together.
When reading and attempting to analyze a paper the nature of Steig et al. (2009), I find it efficient to look for the potential weaknesses and particularly those opportunities where the authors can cherry pick data and methods and in turn do not provide either reasonable or detailed rationalizations for their selections. In the case of Steig after reading and rereading the paper I am not all that confident that I understand what they have done. I’ll give my understanding here with the hope that my misinterpretations will be pointed out and corrected.
In the methods section it is apparent that the sparse data for the Antarctica was in-filled using the RegEM spatial algorithm using iterations over time of the covariance of the distant weather station temperatures. It is does not say directly how the gaps in the weather station temperature reports are handled but I suspect that that is handled by RegEM also. The paper refers to RegEM being used on the reconstructed fields and separately of the temperature field.
I think the authors used the satellite IR derived and AWS weather station temperatures with the more extensive coverage that they present in order to build two reconstructions using RegEM algorithms and those sparser but longer reporting stations. These reconstructions are calibrated and verified in the 1982-2006 time period. Evidently that reconstruction model is used along with the temperature data from the sparser and longer lived stations to reconstruct the whole of the Antarctica going back to 1957 (which is the when many of the sparser stations started reporting).
The method of the reconstruction in in-filling the interior temperatures of Antarctica depends critically on the correlation of temperatures over long distances. What frustrated me here was that I did not see in the paper how this was demonstrated or how well the correlations held up over time. Also not covered in detail was the infilling of temperature gaps over time. In my mind demonstrating that the correlations holding up over time within rather narrow limits is critical in extending the reconstruction back from 1982 to 1957.
In order to use the IR data, a cloud cover masking technique had to be used in an attempt to remove the effects of clouds interfering with the satellite view of the ground. I would suspect that this technique presents the greatest source of uncertainty in the reconstruction. The authors appear to have the some reservation judging by their comments. Also they selected a threshold value of retaining data as not being cloud contaminated of 10 degrees C and because it “…produces the best validation statistics in the reconstruction procedure.” That to me would appear to be a clear cut case of data snooping. The original attempts (Comiso 2000) at using these masking techniques resulted in a colder Antarctica, but one that the Steig paper authors claim would not be consistent with their reconstruction.
Finally I find problematic the Steig authors reference to a warming Antarctica from 1957-2006 with little or no references that I could find in the paper to the more recent years trends (1970-2006) and what a trend might be anticipated to look like when the estimated warm period between 1935-1945 was included in the discussion. They say nothing to forewarn of a 50 year temperature trend being critical to start dates when considering decadal natural climate cycles.
snip – off topic
Re: Jon (#48),
snip –
Steve– look, Jon, this has nothing to do with financial examples which were what you objected to. Please limit yourself here to financial examples. Otherwise you’re re-hashing something that you argued on a prior thread. Please take it to that thread. And, if you want to quote me against myself, fine. But please don’t quote your own prior comments using inflammatory words that I didn’t use as supposed evidence.
I hope that this has not already been posted.
It might be important to improve Antarctic temperature
history, because Svensmark’s group has said the following…
“Evidence that the Earth’s climate really responds to variations in cloud cover comes from Antarctica. When the rest of the world warms up, the southern continent tends to cool down, and vice versa (diagram 2). This contrary behaviour is predictable, because clouds have an unusual warming effect over Antarctica.”
http://www.space.dtu.dk/English/Research/Research_divisions/Sun_Climate.aspx
3 Trackbacks
[…] Smith provided an interesting graph on a thread at Climate Audit link HERE). This plot is the dates corresponding to the actual record as collected from individual […]
[…] had been admiring Steve McIntyre’s demolition of the Steig et al 2009 claims (much promoted in the Australian media) that Antarctic was “warming after all”. I think […]
[…] had been admiring Steve McIntyre’s demolition of the Steig et al 2009 claims (much promoted in the Australian media) that Antarctic was “warming after all”. I think this is […]