While Alan mulls over big red dots in NOAA’s graph where no data exists, Frank Lanser finds that GISS global temperature trend is warmed up by weighting land data more.
Guest post by Frank Lansner http://hidethedecline.eu/
The simple task of combining Land surface temperature with Sea surface temperatures has become an odd complex algorithm for GISS. It seems that they weight land data more and more during the 20th century leading to extra heat added to the GISS global temperatures.
Thus in 1900-1920 the GISS LST+SST graph is mostly spot on the used SST graph (the HADISST/Reynoldsv2). This means that the GISS global temperature around 1900-1920 appears to weight land data zero %
The land fraction is increasing during the 20th century, especially after 1980:
Fig2
The real land fraction of the Earth is of course 30%, but around 1980 GISS uses 40%, in 1988 55% – and in 1995 no less than 73%. (The high land % weighting around 1995 leads to a reduction in temperature decline due to Pinatubo volcanic cooling.)
GISS ends up in 2007 using a land weighting of 67%.
In general GISS defends use of larger land fraction due to their 1200km zones around land stations reaching some Ocean areas. But this does obviously not explain a land fraction that appears to go from near zero to around 70% globally during the 20th century.
Besides, the land temperatures for all stations including the coastal stations appear to show significantly different behaviour compared to the SST´s:
(Taken from PART3 of my new article, http://hidethedecline.eu/pages/posts/the-perplexing-temperature-data-published-1974-84-and-recent-temperature-data-180.php)
And therefore, just shifting SST´s out with land temperatures are questionable.
The topic has been discussed somewhat at Joanne Nova’s site:
http://joannenova.com.au/2010/07/did-giss-discover-30-more-land-in-the-northern-hemisphere/
What is the impact of the still greater land fraction in GISS data? If the 30% land fraction from the real world was used, GISS 2007 would be 0,55K warmer than GISS 1900.
With the still increasing GISS land fraction actually used, we have GISS 2007 0,72K warmer than GISS 1900.
The difference is 0,17K added by not using 30% land constantly. But this calculation could be done in many ways.
We know for a fact that the oceans cover 70% of the planet. So why not use 70% of data from SST?
data sources:
GISS global Land + SST
http://data.giss.nasa.gov/gistemp/tabledata/GLB.Ts+dSST.txt
GISS land temp
http://data.giss.nasa.gov/gistemp/tabledata/GLB.Ts.txt
Both HADISST and Reynolds can be seen using:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere
Frank Lansner: Since you’re having trouble with the filter at Lucia’s and my reply to you at Jo Nova’s is being held by her filter because it has so many links, we could have our discussion here. Just reply to my comment to you above…
http://wattsupwiththat.com/2010/07/17/tipping-point-at-giss-land-and-sea-out-of-balance/#comment-434547
…to set the ball rolling. Then I’ll let Zeke know where we are.
I’m going to bed in a half hour but I’ll be back at it about six hours from now. I’ll look around. I’m sure I’ll find you.
Frank,
This post might be instructive as to why the GISTemp land series differs from a true land-only reconstruction: http://rankexploits.com/musing…..p-mystery/
To quote the reply that Dr. Reto Ruedy sent me:
“The curve NCDC and most likely you are computing shows the mean temperature over the land area (which covers about 1/3 of the globe, a large part of it located in the Northern hemisphere).
None of our graphs represents that quantity. We could obtain it by
creating a series of maps, then averaging just over the land areas
(similar to what we do to get the US graph).
Since our interest is in the total energy contained in the atmosphere which correlates well with the global mean surface temperature, all our graphs display estimates for the global mean, the ones based on station data only as well as the ones based on a combination of station and ship and satellite data. Obviously, the latter is the more realistic estimate and we keep the first one mostly for the following historical reason:
When we started out in the 1980s analyzing available temperature data, historic ocean temperature data were not yet available and we did the best we could with station data. As soon as ocean data compilations became available, we used them to refine our estimates (calling it LOTI). But we kept the earlier estimates also, mostly for sentimental reasons; they are rarely if ever mentioned in our discussions (see also the “note” in the “Table” section of our main web site).
To get back to your question: The mean over the land area is heavily weighted towards the Northern hemisphere and that hemisphere experienced a larger warming than the Southern hemisphere. Hence our estimate which gives equal weight to both hemispheres exhibits a smaller trend, as you noticed, but it still somewhat overestimates the true global mean trend.”
Oops, that link should have been http://rankexploits.com/musings/2010/the-great-gistemp-mystery/
While I support what Frank Lansner is getting at I believe I understand why he got “zero” for the 1900 to 1920 time frame. The graph below is a quick and dirty. I used the 250 km smoothing on the 2×2 grid and EXCLUDED the grid points where land and ocean overlap. (I used 250 km to minimize the number of grid points that I would need to exclude). I then weighted land at 10%, 33%, 50% and 60% to see how each compared to GISS Global (black dashed line)
As you can see, all the lines converge between 1900 and 1920. So it makes little difference over that time period how land is weighted. From 1880 to 1900 33% land weight is closest, but a clear divergence to a larger land weighting in the last few decades.
I can’t do Fortran, just not in my skill set but would be nice to have the exact SST data used on the overlapping grid points in order to do this more accurately. I did some guestimates a couple of different ways on the overlap grid points and didn’t get as big a change as Frank L did, but it seem clear to me that the land weighting changed as more land stations were added.
http://knowledgedrift.files.wordpress.com/2010/07/giss-land-weighting.png
Ok i posted some more at the blackboard, but again, the spam filter.
Everybody: please have some patience, i will look into this tonight danish time.
Me and Jo Nova has used in all 6 reviewers before posting who had no complaints, and im still far (!!) from convinced that Hansens approach is indeed defendable.
Please have patience i have to go work 🙂
K.R. Frank Lansner
Davidmhoffer: Ahhh wonderful with some comon sence, thankyou 🙂
1) Exactly: Whatever GISS claims that their Land/city/airport data covers, in 1900-20 they where weighted ZERO, and recently they are weighted A LOT.
other notes:
2) Another logical problem:
Since SST without any doubt covers 70% of the globe, the resuting global temperatures should not differ that much fro SST.
But they do:
http://hidethedecline.eu/media/GISSglobal/fig1b.jpg
above: If the land graphs where truly for example 50%land and 50% ocean, the the real “land-only” temperatures should be exploding (!) to explain how te big difference to SST can occur.
3) a third logical problem:
Why is the difference land vs SST graph much much bigger in recent years than during the warming 1920-1940?
4) checkout how in the period 1970-74 Land graphs warm 0,1K while SST graphs cools 0,2K:
http://hidethedecline.eu/media/PERPLEX/fig51.jpg
If land graph actually covered 50% ocean, its funny how the land trend then can go its quite own ways?
Frank Lansner: First question from me for this morning: Why does the relationship of the GISTEMP temperature anomaly and HADISST/Reynolds SST anomaly in the following graph indicate that GISS uses zero land data during the period of 1900 to 1910?
http://hidethedecline.eu/media/GISSglobal/fig1b.jpg
Frank Lansner wrote: “2) Another logical problem: Since SST without any doubt covers 70% of the globe, the resuting global temperatures should not differ that much fro SST. But they do:
http://hidethedecline.eu/media/GISSglobal/fig1b.jpg ”
The difference will be caused by the variations in global land surface temperatures. Land surface temperatures vary much more than SST on a decadal basis and on an annual basis. Refer to the following graph for those annual variations in Land and Sea Surface Temperatures (Not Anomalies):
http://i32.tinypic.com/2yvkva0.jpg
You wrote: “3) a third logical problem: Why is the difference land vs SST graph much much bigger in recent years than during the warming 1920-1940?”
Not in any particular order: First, polar amplification, as the globe warms, the Arctic and high latitudes of the Northern Hemisphere exaggerate the warming. Second, land use changes. Third, Urban Heat Island Effect. Fourth, poor surface station siting. Fourth, Etc.
You wrote, “4) checkout how in the period 1970-74 Land graphs warm 0,1K while SST graphs cools 0,2K:
http://hidethedecline.eu/media/PERPLEX/fig51.jpg ”
First, you’re examining a 4-year period of a graph with 5-year smoothing. Is it possible the smoothing is effecting your interpretation? Second, the Global datasets in question as part of the discussion (GISTEMP combined with 1200km radius smoothing and HADISST) are not included in your graph. Third, using unsmoothed Global GISTEMP and HADISST data from January 1971 to December 1974, they both show increasing trends, and the curves are remarkably similar over that 4-year period, as one would expect:
http://i27.tinypic.com/el75g5.jpg
Same thing with Global GISTEMP LST (1200km radius smoothing) and HADISST:
http://i31.tinypic.com/egqy4k.jpg
And the same thing holds true for the Northern Hemisphere GISTEMP LST (1200km radius smoothing) and HADISST:
http://i31.tinypic.com/1zxv0js.jpg
The point being, please stick to the datasets being discussed (GISTEMP and HADISST/Reynolds OI.v2). And if you would, please answer the question presented to you above:
http://wattsupwiththat.com/2010/07/17/tipping-point-at-giss-land-and-sea-out-of-balance/#comment-434815
Regards
Frank Lansner: Let’s take a look at the data and graph upon which you are basing your calculations of percentages of GISTEMP land surface area.
http://hidethedecline.eu/media/GISSglobal/fig1b.jpg
I’ve plotted dozens of global temperature anomaly comparison graphs and I don’t recall a divergence of land surface temperatures that you’re showing in the early part of the 20th century. So I attempted to recreate your graph. Unfortunately, I couldn’t duplicate that divergence from 1900 to 1920 or so.
I used GISTEMP combined (LST&SST) and GISTEMP LST data, both with 1200km radius smoothing. I used HADISST, not a combination of HADISST and Reynolds OI.v2 SST data as you had, since I was only interested in the early part of the data, and merging the two datasets adds another couple of steps. All data use 1951 to 1980 as base years for anomalies and are smoothed with a 60-month running-mean filter, centered on month 30. The data cover the entire term of the GISTEMP data, January 1880 to May 2010. And all of the data is from the same source, the KNMI Climate Explorer. Here’s my graph, using the same color coding as yours:
http://i27.tinypic.com/33e3kth.jpg
As you will note, the LST data does not diverge as greatly from the SST and combined data in the early 20th century. This could mean any number of things, but I did double-check the data I used in this graph. A question, what type of smoothing did you use, since it apparently wasn’t a centered 60-month running-mean filter? Otherwise, your data would not have extended for the full term of the data.
I also created a second graph of that data, but in this graph I plotted the difference between the SST data and the two GISTEMP datasets. Note how the two residual curves “overlap” during the base years of 1951 to 1980. This can also be seen in your graph above and my attempt to duplicate it. This means your analysis is skewed by your election to use the standard GISTEMP base years of 1951-1980.
http://i28.tinypic.com/33kuhwl.jpg
So what happens if we use different base years, 1900 to 2000 for example? The next graph is a comparison of GISTEMP Combined (LST&SST), GISTEMP LST, and HADISST, with the base years of 1900 to 2000. Note that there is better agreement during the extended base period. The divergence between the SST data and the two GISTEMP datasets after the mid-1990s is influenced by my not using Reynolds OI.v2 SST data. Reynolds OI.v2 SST anomalies have a higher trend than HADISST. (And my assumption is that GISS uses it because of availability and not the higher trend. Monthly updates for Reynolds OI.v2 are available early in the month, while HADISST updates lag by at least one month.)
http://i30.tinypic.com/212bwxs.jpg
And the last graph is the comparison of the differences between the HADISST and the two GISTEMP datasets, using 1900 to 2000 as base years for anomalies. This graph should better represent the relationships you are expecting from the data.
http://i32.tinypic.com/2m3i7oj.jpg
Look reasonable?
In summary:
1.Please check your data, especially the land surface data.
2.Please check your smoothing to determine if it’s imposing a bias on the ends of your data, or simply use a standard centered x-month running-mean filter. So you lose 36 months of data on the ends–who cares?
3.Please try base years that represent the entire term of the data you’re examining, and:
4.From an earlier comment at Jo Nova’s, you need to mask the land surface data.
Regards
Bob,
As I mentioned over at Lucia’s place, its not just masking. The zonal weighting in GISTemp LST has a similarly large effect in lowering the trend vis-a-vis a true land-only temp record that applies weights proportional to land area.
Zeke Hausfather (July 20, 2010 at 1:32 pm): Thanks for the reminder, but unless Frank accounts for the land and ship station data that’s extended out over the oceans in the basic GISTEMP land surface data with 1200km radius smoothing that he uses, his results will continue to be skewed.
Hi Bob!
As I wrote at the Jo nova blog, your efforts here are highly appreciated, and i will considder your points.
In the mean time I wrote:
http://hidethedecline.eu/modules/news/modify_post.php?page_id=27§ion_id=27&post_id=187
K.R. Frank
Wops heres the correct link:
http://hidethedecline.eu/pages/posts/are-giss-pure-land-data-important-to-understand-the-differences-between-giss-data-and-other-datasets-187.php
K.R. Frank
Hi bob!
I wrote a general answer to you at Joanne novas site.
Thanks Again!
Dear Bob.
regardng your article:
I offered you a bottle of wine if you would go through my new article:
http://hidethedecline.eu/pages/posts/the-perplexing-temperature-data-published-1974-84-and-recent-temperature-data-180.php
If you had done this, you would certainly not have written as you do.
One more time. Please look at PART 2 of my article, chapter 3.4.
This is where I explain that GISS inlcudes ocean in their station data series, and where I show a graphic of the ocean included.
So if anyone is aware of this, its me. The fact that you and others keep writing that you think im not aware of ocean data in GISS station “land” data might be my fault due to bad communication.
I wrote in my article PART 4:
“I am sure that the algorithm or specific method used by GISS to combine Land temperature and SST explains some of these apparently odd findings. But whatever the “algorithm” used by GISS is, can it be justified that GISS gradually weights the warm NH-Land graph more and more? And ends up with around 67% NH land fraction in 2007 although NH only has 40% land? Maybe, this algorithm or method deserves some attention?
”
And in the WUWT article i write: http://wattsupwiththat.com/2010/07/17/tipping-point-at-giss-land-and-sea-out-of-balance/#more-22126
I write:
“In general GISS defends use of larger land fraction due to their 1200km zones around land stations reaching some Ocean areas. But this does obviously not explain a land fraction that appears to go from near zero to around 70% globally during the 20th century.
”
Now, Your article, Bob:
You focus on the similarities between CRU and GISS – i suppose to say that the resulting GISS is ok?
The thing is, CRU and GISS ends up rather alike. But in CRU data i find much more direct land data adjustment than for GISS. On the contrary for GISS, the direct land data adjustments are not so big at all (to my surprice) but in stead the GISS warming trend thats similar to CRU comes when combining the SST and “land”.
SOmething thats messy in al this is, that you seem to trust that CRU land is not ocean while GISS is… Yes yes, GISS has ship and island data included, but a BIG part of the GISS ocean area in their “land” data is obviousy from coastal stations. These stations are exactly the same as for CRU. So its nonsense to say “CRU is just land data”.
Just becasue CRU says that their coastal stations are land while GISS (the same) coastal stations covers huge ocean areas, you cant just treat the same data as if completely different.
I have raised some serious problems in data, and I know you disagree strongly, but i have not seen conving arguments from you, its not bad will.
K.R. Frank
Frank Lansner: Why are you including your comment above…
http://wattsupwiththat.com/2010/07/17/tipping-point-at-giss-land-and-sea-out-of-balance/#comment-438046
..on this thread. The post I wrote yesterday is not linked anywhere in this thread. So you’re confusing those who are attempting to follow this.
Also, I replied to your comment at the post at my blog:
http://bobtisdale.blogspot.com/2010/07/land-surface-temperature-contribution.html
There I wrote:
Frank Lansner: You wrote, “I offered you a bottle of wine if you would go through my new article:
http://hidethedecline.eu/pages/posts/the-perplexing-temperature-data-published-1974-84-and-recent-temperature-data-180.php
“If you had done this, you would certainly not have written as you do.”
It was my intent to eventually address your request, in a few days. I have other priorities. Also, I don’t drink alcohol. I gave it up.
You wrote, “One more time. Please look at PART 2 of my article, chapter 3.4.
This is where I explain that GISS inlcudes ocean in their station data series, and where I show a graphic of the ocean included.”
But your analysis where you assume that GISS increases land surface area does not address this.
You wrote, “I am sure that the algorithm or specific method used by GISS to combine Land temperature and SST explains some of these apparently odd findings. But whatever the ‘algorithm’ used by GISS is, can it be justified that GISS gradually weights the warm NH-Land graph more and more? And ends up with around 67% NH land fraction in 2007 although NH only has 40% land? Maybe, this algorithm or method deserves some attention?”
But GISS does NOT weight “the warm NH-Land graph more and more.” You need to mask the areas where land surface data extends out over the oceans in your analysis.
You wrote, “In general GISS defends use of larger land fraction due to their 1200km zones around land stations reaching some Ocean areas. But this does obviously not explain a land fraction that appears to go from near zero to around 70% globally during the 20th century.”
Please provide a link to an article written by a member of GISS where “GISS defends use of larger land fraction due to their 1200km zones around land stations reaching some Ocean areas.”
You wrote, “You focus on the similarities between CRU and GISS – i suppose to say that the resulting GISS is ok?”
I also included NCDC data. Or did you miss that? This post was not about the accuracy of land surface temperatures; it was about the contribution of land surface temperature readings to combined land and sea surface temperature data. And there is basically no difference between those of GISS, Hadley Centre, and NCDC.
Frank Lansner,
I think I follow your logic. You seem to be saying that there is an apparent, very important, apparent, increase in the GISS records of land/ocean area represented in the temp records due to the 1200 km interpolation of land temperatures over ocean areas, not in GISS’s computer code itself. This has manifested itself through the years as stations have been strategically removed.
Is that right? Is that where Bob Tisdale has misjudged your work in http://wattsupwiththat.com/2010/07/23/bob-tisdale-on-giss-landsea-ratios/? He keeps saying you need to mask the ocean areas that are actually land temps but that is exactly what you are pointing out. Correct?