Guest post by Bob Tisdale
There are numerous blog posts and discussions about how the GISS global temperature anomaly product GISTEMP differs from the Hadley Centre and NCDC datasets.
The repeated reasons presented for this are, GISS uses 1200km radius smoothing to fill in the areas of the globe with sparse surface temperature readings, and the area this has the greatest impact is the Arctic. Typically, a map or comparison of global temperature anomaly maps is included, similar to Figure 1. The top two maps were cropped from Figure 3 in the Real Climate post “2009 temperatures by Jim Hansen”. I added the third.
The bottom map was created at the GISS Global Maps webpage. It’s a map of the GISTEMP Global Temperature Anomaly product with 250km radius smoothing for the calendar year 2005, the same year as the top two maps. I did not include a temperature scale because the bottom map was provided to allow a visual comparison of the spatial coverage of the HadCRUT product and the GISTEMP product with 250km radius smoothing. Examine the Arctic and the ocean surrounding Antarctica, the Southern Ocean. Notice a difference? In 2005, the HadCRUT data had better coverage of the Arctic and Southern Oceans than the GISTEMP dataset with 250km radius smoothing. What’s missing in the GISTEMP product? There’s no sea surface temperature data.
http://i45.tinypic.com/htsgeq.jpg
Figure 1
GISS DELETES POLAR SEA SURFACE TEMPERATURE DATA
The general regions where GISS deletes Sea Surface Temperature data are shown in Figure 2. Three areas are highlighted: two cover the Arctic Ocean, and a third surrounds Antarctica. The specific locations are clarified in the following. GISS then uses their 1200km radius smoothing to replace the sea surface data with land data.
http://i48.tinypic.com/33adj86.jpg
Figure 2
Tilo Reber in his recent “Diverging views” post at Watts Up With That? noted that the GISS Current Analysis webpage includes the following statement:
“Areas covered occasionally by sea ice are masked using a time-independent mask.”
This means that vast regions of Sea Surface Temperature (SST) anomaly data in the Arctic Ocean and Southern Ocean are deleted from the GISTEMP record. GISS does not delete all of the Arctic and Southern Ocean SST anomaly data, just the data from the areas where the annual sea ice melt occurs, and those are good portions of them.
I have looked for but have not found an explanation for this exclusion of Sea Surface Temperature data in the papers provided on the GISTEMP references page.
THE AREA OF THE ARCTIC OCEAN WHERE GISS DELETES SST DATA
Figure 3 shows four Arctic (North Pole Stereographic, 65N-90N) maps prepared using the map-making feature of the KNMI Climate Explorer. The maps illustrate temperature anomalies and sea ice cover for the month of September, 2005. The calendar year 2005 was chosen because it was used in the RealClimate post by Jim Hansen, and September is shown because the minimum Arctic sea ice coverage occurs then. The contour levels on the temperature maps were established to reveal the Sea Surface Temperature anomalies. Cell (a) shows the Sea Ice Cover using the Reynolds (OI.v2) Sea Ice Concentration data.
The data for the Sea Ice Cover map has been scaled so that zero sea ice is represented by grey. In the other cells, areas with no data are represented by white. Cell (b) illustrates the SST anomalies presented by the Reynolds (OI.v2) Sea Surface Temperature anomaly data. GISS has used the Reynolds (OI.v2) SST data since December 1981. It’s easy to see that SST anomaly data covers the vast majority of Arctic Ocean basin, wherever the drop in sea ice permits. Most of the data in these areas, however, are excluded by GISS in its GISTEMP product. This can be seen in Cell (c), which shows the GISTEMP surface temperature anomalies with 250km radius smoothing. The only SST anomaly data used by GISS exists north of the North Atlantic and north of Scandinavia.
The rest of the SST data has been deleted.
The colored cells that appear over oceans (for example, north of Siberia and west of northwestern Greenland) in Cell (c) are land surface data extending over the Arctic Ocean by the GISS 250km radius smoothing. And provided as a reference, Cell (d) presents the GISTEMP “combined” land plus sea surface temperature anomalies with 1200km radius smoothing, which is the standard global temperature anomaly product from GISS. Much of the Arctic Ocean in Cell (d) is colored red, indicating temperature anomalies greater than 1 deg C, while Cell (b) show considerably less area with elevated Sea Surface Temperature anomalies.
http://i46.tinypic.com/dpygcj.jpg
Figure 3
Basically, GISS excludes Arctic Ocean SST data from 65N to 90N and, for round numbers, from 40E to 40W. This is a good portion of the Arctic Ocean. Of course, the impact would be seasonal and would depend on the seasonal drop in sea ice extent or cover. The sea ice extent or cover has to decrease annually in order for sea surface temperature to be measured.
I’ll use the above-listed coordinates for the examples that follow, but keep in mind that they do not include areas of sea ice in the Northern Hemisphere south of 65N where sea surface temperature data are also deleted by GISS. These additional areas are highlighted in Figure 4. They include the Bering Sea, Hudson Bay, Baffin Bay and the Davis Strait between Greenland and Canada, and the Sea of Okhotsk to the southwest of the Kamchatka Peninsula.
http://i50.tinypic.com/28j9u6u.jpg
Figure 4
Note: GISS uses Hadley Centre HADISST data as its source of Sea Surface Temperature (SST) data from January 1880 to November 1981 and NCDC Reynolds (OI.v2) data from December 1981 to present. To eliminate the need to switch between or merge SST datasets, this post only examines the period from 1982 to present. And to assure the graphics presented in Figures 3 and 6 are not biased by differences in base years of the GISTEMP data and the Reynolds (OI.v2) SST data, the latter of which has only been available since November 1981, I’ve used the period of 1982 to 2009 as base years for all anomaly data.
WHY WOULD DELETING SEA SURFACE TEMPEATURE DATA AND REPLACING IT WITH LAND SURFACE DATA BE IMPORTANT?
Land Surface Temperature variations are much greater than Sea Surface Temperature variations. Refer to Figure 5. Since January 1982, the trend in GISTEMP Arctic Land Surface Temperature Anomalies (65N-90N, 40E-40W) with 250km radius smoothing is approximately 8 times higher than the Sea Surface Temperature anomaly trend for the same area.
The Arctic Ocean SST anomaly linear trend is 0.082 deg C/ decade, while the linear trend for the land surface temperature anomalies is 0.68 deg C/decade. And as a reference, the “combined” GISTEMP Arctic temperature anomaly trend for that area is 9 times the SST anomaly trend.
http://i46.tinypic.com/1zpheme.jpg
Figure 5
By deleting the Sea Surface Temperature anomaly data, GISS relies on the dataset with the greater month-to-month variation and the much higher temperature anomaly trend for its depictions of Arctic temperature anomalies. This obviously biases the Arctic “combined” temperature anomalies in this area.
GISS DELETES SEA SURFACE TEMPERATURE DATA IN THE SOUTHERN HEMISPHERE, TOO
Figure 6 shows four maps of Antarctica and the Southern Ocean (South Pole Stereographic, 90S-60S). It is similar to Figure 8. Cell (b) illustrates the SST anomalies presented by the Reynolds (OI.v2) Sea Surface Temperature anomaly data. SST anomaly data covers most of the Southern Ocean, but GISS deletes a substantial portion of it, as shown in Cell (c). The only SST anomaly data exists toward some northern portions of the Southern Ocean. These are areas not “covered occasionally by sea ice”.
http://i50.tinypic.com/aensly.jpg
Figure 6
Figure 7 illustrates the following temperature anomalies for the latitude band from 75S-60S:
-Sea Surface Temperature, and
-Land Surface temperature of the GISTEMP product with 250km radius smoothing, and
-Combined Land and Sea Surface of the GISTEMP product with 1200km radius smoothing, the GISTEMP standard product.
The variability of the Antarctic land surface temperature anomaly data is much greater than the Southern Ocean sea surface temperature data. The linear trend of the sea surface temperature anomalies are negative while the land surface temperature data has a significant positive trend, so deleting the major portions of the Southern Ocean sea surface temperature data as shown in Cell (c) of Figure 6 and replacing it with land surface temperature data raises temperature anomalies for the region during periods of sea ice melt.
Note that the combined GISTEMP product has a lower trend than the land only data. Part of this decrease in trend results because the latitude band used in this comparison still includes portions of sea surface temperature data that is not excluded by GISS (because it doesn’t change to sea ice in those areas).
http://i45.tinypic.com/im6q29.jpg
Figure 7
ZONAL MEAN GRAPHS REINFORCE THE REASON FOR THE GISS DIVERGENCE
When you create a map at the GISS Global Maps webpage, two graphics appear. The top one is the map, examples of which are illustrated in Figure 1, and the bottom is a Zonal Mean graph. The Zonal Mean graph presents the average temperature anomalies for latitudes, starting near the South Pole at 89S and ending near the North Pole at 89N. Figure 8 is a sample. It illustrates the changes (rises and falls) in Zonal Mean temperature anomalies from 1982 to 2009 of the GISTEMP combined land and sea surface temperature product with 1200km radius smoothing. The greatest change in the zonal mean temperature anomalies occurs at the North Pole, the Arctic. This is caused by a phenomenon called Polar Amplification.
http://i48.tinypic.com/spd4li.jpg
Figure 8
To produce a graph similar to the GISS plot of the changes in Zonal Mean Temperature Anomalies, I determined the linear trends of the GISTEMP combined product (1200km radius smoothing) in 5 degree latitude increments from 90S-90N, for the years 1982 to 2009, then multiplied the decadal trends by 2.8 decades. I repeated the process for HADCRUT data. Refer to Figure 9.
The two datasets are similar between the latitudes of 50S-50N, but then diverge toward the poles. As noted numerous times in this post, GISS deletes sea surface temperature data at higher latitudes (poleward of approximately 50S and 50N), and replaces it with land surface data.
http://i47.tinypic.com/2uzfc6r.jpg
Figure 9
Figure 10 shows the differences between the changes in GISTEMP and HADCRUT Zonal Mean Temperature Anomalies. This better illustrates the divergence at latitudes where GISS deletes Sea Surface Temperature data and replaces it with land surface temperature anomaly data, that latter of which naturally has higher linear trends during this period.
http://i45.tinypic.com/xnsp40.jpg
Figure 10
SOURCE
Maps and data of sea ice cover and temperature anomalies are available through the KNMI Climate Explorer:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere
What tangled data webs they weave, with data crafted to deceive…
No one has the right to delete ANY data. If you need to look at the past or see mistakes then the data would be useful to show.
But this is deliberate tampering with data and deleting the evidence!
Another excellent article by Bob Tisdale. Much appreciated.
And I agree wholeheartedly with Joe Lalonde:
“No one has the right to delete ANY data.”
The cost of data storage is so minuscule today that there is no excuse whatever for deleting taxpayer-funded data. What gives them the right?!
The data was deleted because GISS has something to hide.
Things really have gotten to the point where nothing any official body cobbles up regarding SST or any other climate indicator seems even worth discussing. Hansen’s latest exercise in GISS/NASA subterfuge reminds one of J.B. Rhine’s parapsychology “experiments” in the 1930s, which Martin Gardner revealed to have omitted negative results because they “skewed the data” (!).
[snip]
Too bad they can not delete Greenland Ice Temperature Data:
http://www.globalwarmingskeptics.info/forums/thread-188-post-5233.html#pid5233
A cooling trend of about 3300 years.
Thanks, Anthony. The Golden Three Shell Game lead-in photo was apropos.
How odd. I started playing arounf with the Global Maps from the link in the article. I tried to run NH cold season from the unadjusted data and was surprised to get an error saying that the unadjusted data is only available to 1999. Interesting. Unadjusted data which takes zero work is unavailable after 1999 but adjusted data which takes considerable work is.
I ran it from 1881 (despite what it says, 1880 doesn’t work) and was surprised to see how tiny the anomalies are that result. 0.2 degrees from top of graph to bottom. And yes, a clear downward trend from 15S to 75N after which….rocket ship.
Good Golly, Miss Molly.
Do you find this jolly?
Each rock o’erturned reveals a snake
And there’s a forest of trees to shake.
=========================
Meanwhilst, the forest fires in Quebec are casting a hazy shade over New England of a type that Britain is likely enjoying from Iceland. Today, our Channel 9 weatherman said that temps across the state were about 4 degrees lower today than they were predicted to be because of this smoke overcast.
Of course, you couldn’t get him to acknowledge that this is a factor in climate change predictions, that the Clean Air Act can easily explain any actual warming we’ve experienced since 1972.
“The data was deleted because GISS has something to hide.”
The data is NOT deleted. The other day I spent a few hours writing some R code to download the data that GISS uses from the source. As Bob notes GISS uses two sources. I will discuss the first source 1880-2003, HADISST. The data gives you monthly figures for the entire globe from before 1880 up to 2003. The data is in 1 degree bins. If you look at the source data, you will find that the product contains one of 3 numbers: a value for those bins that are land, or a value for SST, or a value for ICE when there is ICE. The same is true I believe for the reynolds product.
The question is How does GISS use this information in the calculation of their final product. Given a grid or sub grid that is sometimes ICE ( a value of -1000 in the source data) and sometimes a valid SST, how do GISS process this? That is an answerable question. Personally, I would like to see the GISS approach explained in more detail. They very well may disregard grids that are occassionaly covered by ICE or may use land figures in interpolating over these positions. But they didnt delete the data.
For example: Taking the data from SOURCE, you have the following figures for
grid box 4546 (around lat 82) for the months jan through december: -1000 means it is covered by ICE, -180 is -1.8C.
jan feb mar apr may june
1881 -180 -1000 -180 -180 -1000 -180 52 189 87 123 64 -180
Well, the obvious reason for not plotting SST data in areas covered by ice is that SST is not then measured. I presume your complaint is about areas that have SST measurements in summer only, but are iced in winter. How do you think they should deal with regions with only summer SST readings when compiling an annual average plot?
Can somebody do a sanity check on something I’ve stumbled across? I believe that GISS is based on adjusted NOAA data. Plot 12-month running mean of GISS versus 12-month running mean of NOAA anomalies. The respective URLS for the data are…
Notice that GISS is almost always lower than NOAA. However, since November 2009, the GISS 12-month running mean has pulled ahead of NOAA, and is now leaving a lot of daylight. There might be an explanation for the GISS/Hadley delta, and easily one for GISS versus UAH and RSS. But I can’t see the justification for what’s happening now versus NOAA.
I’ve also plotted the 12-month running means for Hadley/GISS/RSS/UAH/NOAA. GISS is THE ONLY ONE of the 5 whose 12-month mean to April exceeds the 1998 super El Nino peak. In other words, GISS is the outlier, against Hadley/RSS/UAH/NOAA.
In case you’re wonder why I’m pounding away on the 12-month mean, it’s because the 12-month mean to the end of December 2010 will constitute the 2010 annual mean temperature.
First we had a questionable procedural problem with CRU, GISS, et.al. :
1) deleting non-conforming weather stations by the thousands
Then we had ethical problems:
2) adding data (interpolating) into huge 1000 sq km voids, while pretending otherwise.
Now, we have a moral problem:
3) deleting data that does not follow the scripts.
How low can they go?
How about:
4) measuring CO2 levels on top of an active volcano system.
#4 is ignorant and more.
It does look damning, but I can see why water which is sometimes covered in ice is difficult to measure.
I think the best thing to do in these circumstances is to NOT graph the data. To (presumably intentionally) spread temperature measurements that are pretty obviously not representative over this area, especially when those spots are so very few and so very far apart, is disingenuous at best.
We need proper data. The current measurements, and indeed, their sources, are obviously invalid and deeply (politically) tainted.
Bob Tisdale…. You ROCK!!!!
Someone needs to go over to GISS and break all the red crayons…
Mike
Here’s an idea. If you don’t have data over a large area, don’t include that area in your less-than-global temperature analysis. The crime here is one that I have heard too often in AGW science: we didn’t have reliable data so we just assumed what ever data we wanted. GISS does it at the poles. They did it to ‘prove’ that hurricanes are getting stronger, the Amazon is really susceptible to drought, the Himalayan glaciers are all melting, diseases are spreading due to warming, animals are increasingly going extinct because of AGW and there was no Medieval Warm Period. All of these false claims have been ‘supported’ by data manipulation that is nearly criminal.
If you want good science these days, you have to go to your local elementary school science fair. The professionals are becoming a joke.
One thing that strikes me about the ice extent curves is that the shoulder values, May and Nov, are remarkably consistent year on year.
I took the daily differences from the average over the whole data period, then extracted the four month averages for the winter and summer extremes for each year. From these, I created the following plots:
a) summer diff versus preceding winter diff;
b) winter diff versus preceding summer diff;
c) summer diff versus preceding summer diff;
These gave me R2 values of .17, .01, .15 respectively. (Indeed, (b) was a slight negative correlation.) This may make physical sense. Each winter, the arctic ocean freezes right over, and the freezing beyond that may be far more a function of what’s going on in the Nth Atlantic etc. then what’s going on around the pole. So very low correlation with preceding summer minimum. OTOH, the summer melt will depend both on ice thickness and on any buffer ice hanging around the channels to the other oceans.
So finally I plotted summer versus sum of preceding winter and the summer before that. Sure enough, R2 went up to 0.32.
Examining figure 10, it appears to me that the over temperature anomalies are in the areas not actually covered by real recordings. Therefor if you “0” all areas that have no thermometer recordings there would be no warming. Actually there would be cooling.
Is this due to stupidity or is this due to fraud? One more piece of evidence to go into a very large basket.
Look at the south pole! no warming! because they have a weather station recording actual temperatures. WUWT
Nick Stokes:
Either way it’s problematic, especially if Bob is correct about them using data interpolated from land.
Sounds to me like it needs a rethink.
Thanks Bob for this outstanding post and expertise! Giss not only deletes SST that is mostly covered with sea ice but (your quote)
Nick is right. After downloading the source data and looking at it, the approach one should use is not immediately clear. take the example line of data I produced from the source. Suppose we have three adjacent grid squares: land:sea:sea.
suppose further that the land is at 0C and the SST is -1.8C and -1.8C. when the water is open. Now in some months this water is frozen. The SST file does not report a temperature for this. It reports a dummy value of -1000. So, using the example above, suppose the values are then 0c; ICE; -1.8C. Does anyone have a clear idea of how to average over these? As Nick suggests this is a interesting problem. CRU appears to handle the problem one way. GISS uses a different approach. The DIFFERENCE between GISS and CRU is not some nefarious plot. The difference is caused by a analytical choice on how to handle these cases. The difference in results is an uncertainty in the estimate of global temperature. two different ways of handling the same problem. That’s the sum total of this.
Terms such as ‘extrapolate’ and ‘interpolate’ have completely different meanings so should be used in the proper context. It appears in this instance GISS is extrapolating temperatures, not interpolating.
As far as parsing the word ‘delete’, as Bob’s chart clearly shows, GISS most definitely “deleted” data, which is a legitimate usage of the word in the strictest definition. If the data was there before, but not after, it was in fact deleted. “Delete” may seem accusatory, but data “manipulation” is another term many view as a dirty word, which again is nothing of the sort.
Apparently GISS is making assumptions about polar amplification due to “global warming”, but where is the justification for such speculation? Ah yes, climate models.
Thanks Bob for this outstanding post and expertise! Giss not only deletes SST that is mostly covered with sea ice but (your quote)
If there is anything I could add, it’s how Giss gloated in their recent paper
. Now, it is worse than (I) expected. “Rapidly” means +0.17°C global trend since 1997 vs. no trend at all at HadCRUT3!
Thank you Bob, fantastic analysis. A copy or email link of this one article would awaken any scientist in the physical sciences realm save climatologists. You don’t have to be a climatologist to see exactly what is going on and what they are doing to manipulate the data. Got an update global list handy of every physical science scientist on this globe? 🙂 Wish I had one to give you.
Kind of sad how physics, chemistry, and most other sciences are getting values closer and closer to reality, more and more digits of accuracy, while one single man in climatology, Jim Hansen, has deceived science to accept that accurate numbers should not be sought, and in fact, after coerced a paper through publication that says it’s OK to throw most of the real world data away. His NASA division will just assume all of these values at will by picking where the surviving stations reside. If that’s not fraud I don’t know what is.
Why o’ why is not science at an uproar over that one single deception? I’m ashamed of science lately! Not an honest bone left or you would hear the shouting from the rafters everywhere, not just in the few honest blogs. Enough, my blood pressure’s rising.
It would be enlightening if for each blank 250km area in the GISS map you took the HadCRUT value at that cell, if it exists, and fill it in, HadCRUT data is better than no data, right. Then do the Hansen trick of 1200 kilometer extrapolation. You just know the two maps would always look like night and day.
Am I running this Global Map thing right? I ran adjusted NH cold season land only for a bunch of ranges. 1910 to 1920 showed an arctic anomaly of -5 degrees. No, not -0.5, MINUS FIVE DEGREES. Ran it with NOAA ocean added in, same result. I’ve played around with zonal data a lot and that didn’t look right so I started running individual years. There are a few odd years in there but the whopper is 1917 which has annual temp anomalies ranging to -6 in the arctic. You can run warm season and cold season and I got -3 and -11 (!!) which averages to about -6. I then interpolated from the annual graph and got:
~ 60N -0.5
~ 85N -6.0
average ~ -2.75
Zoom over to gisstemp:
http://data.giss.nasa.gov/gistemp/tabledata/ZonAnn.Ts.txt
They show 64N to 90N at -1.19.
They show 44N to 64N at about -0.5 which is a pretty good match for Global Map.
But 60+ on Global Map just looks totaly out of whack. Its late, I’m interpolating, maybe I see what I want to see but this just looks wrong to me. I even looked at seasonal which unfortunately is only broken down by hemisphere, but it doesn’t show anything super unusual that year though winter was a bit cooler. 1912 also appears to have some huge anomalies in the arctic.
Carrick, From memory ( dont trust but verify) I think the way the GISS program is going to work is as follows. The 1 degree SST bins are going to be mapped into the 8000 equal area subgrids. if there is an SST present you will get a SST value. if ICE is present you get NA. Then the entire subgrid structure is averaged. So its not entirely clear that only land values will contribute to the value. IF your grid goes: land:ice:land
then you will average a land value over that grid that is ice. Plenty of issues but there is not a clear analytic choice in my mind. Unless somebody wants to estimate an air temp over ice. suggestions? I know Zeke is working on this as chad got him the data.
2010-Jun-01 3:00 Water temp at N:88°44′, W:042°43′ is -2.0
Air temp is -28.2
The info is there if you look hard enough. Giss just doesn’t look at all.
Try drifting buoy 25593
Maybe GIZZ needs to hook up with these guys-
http://iabp.apl.washington.edu/
25593 says air temp is -28.2
25595 says water temp is -2.0 and it is located at N:88°43′, W:42°58′.
2010-Jun-01 3:00
It would appear that data has not been “deleted” from the record and lost for all eternity, but rather that data has been substituted in the polar regions. See
Steven mosher says:
May 31, 2010 at 8:11 pm
Still, one has to wonder why the polar regions are uniformly a large positive anomaly.
Today’s high at Point Barrow was a whopping 34F, low 30F.
Nick Stokes and Steve Mosher,
Yes, I agree that nothing nefarious is going on but this problem is inherent in any “global mean” temperature measurement regimen.
Absent a nearly ubiquitous network of thermometers placed at equidistant arbitrarily high concentrations in all three spacial dimensions in the entire atmosphere and ocean, and monitored at an arbitrarily high temporal frequency, the idea of a “global mean” temperature is of very limited utility.
This is like the UHI effect, change the data to show colder long ago, because they can’t increase the current data because everyone knows if it’s sweltering now.
There’s been simply too many changes with no or little explanation why, and as they say perception is everything. So if you act like a crook, everyone thinks you’re a crook.
Thanks Bob for yet another excellent example of why global temperature anomalies have little value in determining what is happing to climate. Without good data no conclusions can be reached and it’s time climate scientists went back to the drawing board with the CAGW hypothesis.
In this situation the null hypothesis of natural climate change has to take precedence.
carrick
GISS builds one SST record from two sources. The first source HADISST has NO SST where there is ice. The cell is either
A: land, in which case there is no temperature value in the SST dataset
B: Open sea, in which case there is a temp value
C: ICE, in which case there is NO temperature value.
Reynolds, appears to be different> here is there format:
lon = 180.5 lat = 89.5 sst = -1.8 ice = 100 tagls= 1.
lon = 180.5 lat = 88.5 sst = -1.8 ice = 96 tagls= 1.
lon = 180.5 lat = 87.5 sst = -1.7 ice = 96 tagls= 1.
lon = 180.5 lat = 86.5 sst = -1.7 ice = 94 tagls= 1.
The HADISST has no comparable ice concentration value. Its binary. Further,
HADISST has no SST value where there is ICE. GISS merge these two datasets. The first question is how do GISS decide to do this. Arguably, one can adopt the hadisst approach. where there is ice, there is no SST. Merging hadisst and reynolds thus would have the effect of not using the reynolds SSTs where there is ice or partial ice.
It appears that is what they do. what that means of course is that when you average the temperatures at the south pole you are using the land stations that ring anarctica to provided estimates of the air temp over that ice.
lets take a look at panel C. You will see the land stations. march 2005
lets pick scott base
http://www.antarctica.ac.uk/met/READER/surface/Scott_Base.All.temperature.html
The temperature of the land station (-21C) is used to extrapolate over those areas covered by Ice.. areas where the SST is arguably -1.8C
So, what value would you suggest that people use when an ocean area is covered by ice? The value of the SST under the ice? -1.8C. no value? or the value of a land station 100s of km away ( like -21C)
Those are your choices, can u think of others. neat problem.
Steven mosher says:
May 31, 2010 at 9:37 pm
“The DIFFERENCE between GISS and CRU is not some nefarious plot. The difference is caused by a analytical choice on how to handle these cases. The difference in results is an uncertainty in the estimate of global temperature.”
The difference is not an “uncertainty”, but a GISS warming bias. They use data with a 8-fold higher warming trend to interpolate sea surface data.
This is an important posting by Bob Tisdale, as it proofs very easily and understandable a signifant warming bias in GISS data, particularly the prominent polar regions.
Sera says:
May 31, 2010 at 10:47 pm
2010-Jun-01 3:00 Water temp at N:88°44′, W:042°43′ is -2.0
Air temp is -28.2
The info is there if you look hard enough. Giss just doesn’t look at all.”
GISS is well aware of this data. here is the problem. For one of your datasets you have no historical value for the SST or the air temp where there is ice. For more current data you have ice concentration and STT for every cell, but no air temp above Ice or above water. And finally you have the latest data ( the bouys) where you have the SST value and the air temp. Having a consistent treatment over all three datasets is historically problematic. it requires the dreaded modeling word.
Carrick. put this one in your thought locker
The statistics of surface air temperature observations obtained from buoys, manned drifting stations, and meteorological land stations in the Arctic during 1979–1997 are analyzed. Although the basic statistics agree with what has been published in various climatologies, the seasonal correlation length scales between the observations are shorter than the annual correlation length scales, especially during summer when the inhomogeneity between the ice-covered ocean and the land is most apparent. During autumn, winter, and spring, the monthly mean correlation length scales are approximately constant at about 1000 km; during summer, the length scales are much shorter, i.e. as low as 300 km. These revised scales are particularly important in the optimal interpolation of data on surface air temperature (SAT) and are used in the analysis of an improved SAT dataset called IABP/POLES. Compared to observations from land stations and the Russian North Pole drift stations, the IABP/POLES dataset has higher correlations and lower rms errors than previous SAT fields and provides better temperature estimates, especially during summer in the marginal ice zones. In addition, the revised correlation length scales allow data taken at interior land stations to be included in the optimal interpretation analysis without introducing land biases to grid points over the ocean. The new analysis provides12-hour fields of air temperatures on a 100-km rectangular grid for all land and ocean areas of the Arctic region for the years 1979–1997
Clearly the minds at GISS think the large amount of ice melt that goes on in these areas creates a kind of climate buffer. Melting a large amount of ice will actually remove a lot of heat energy from the local climate and work to minimize any observed increases in temperature. If the world actually was warming, the presence of large bodies of ice would actually maintain a stable local climate, and the best indicator of warming would be in increased rates of ice melt rather than increased temperatures.
That being said, we are talking about a closed system here. The GISS needn’t offset for climate buffers like melting sea ice. We are interested in the workings of the climate as a whole, and therefore I think offsetting a localized temperature buffer is disingenuous at best. Any buffer that does exist should be understood, not covered up.
Nick Stokes wrote, “Well, the obvious reason for not plotting SST data in areas covered by ice is that SST is not then measured. I presume your complaint is about areas that have SST measurements in summer only, but are iced in winter.”
I believe I note in the post that the impact is seasonal.
You asked, “How do you think they should deal with regions with only summer SST readings when compiling an annual average plot?”
How do I think GISS should deal with areas with seasonal sea ice? Not sure what you feel would be so difficult. Use SST data when it’s available. They could extend land surface data out onto ice when there is no SST data. But if the land surface data can’t extend to its full 1200km because there’s open ocean in the way, they could then use the Reynolds SST climatology or another climatology for the remaining sea ice areas.
Steven Mosher;
So, what value would you suggest that people use when an ocean area is covered by ice? The value of the SST under the ice? -1.8C. no value? or the value of a land station 100s of km away ( like -21C)>>
So what value to we put on surface temps on land that are covered with snow?
Para-climatology?
Steve Mosher: You wrote, “The DIFFERENCE between GISS and CRU is not some nefarious plot.”
I never said it was a nefarious plot. I noted that SST data is apparently not used when it is available in areas with seasonal sea ice. GISS chooses to use the dataset with the higher trend (land) and that choice biases their version of global temperature. If the SST data is available, why not use it?
Regarding your earlier comment, you wrote, “Taking the data from SOURCE, you have the following figures for grid box 4546 (around lat 82) for the months jan through december…”
82N is far enough north to be considered permanent sea ice. What does the source reveal for a cell from “Areas covered occasionally by sea ice…” that “…are masked using a time-independent mask”, for example, for the cell with the coordinates 75N, 45E?
My interpretation (assumption) of the two sentences in their Current Analysis webpage…
http://data.giss.nasa.gov/gistemp/sources/gistemp.html
“Areas covered occasionally by sea ice are masked using a time-independent mask. The Reynolds climatology is included, since it also may be used to find that mask.”
…is that they monitor SST and the climatology in areas with what has been permanent sea ice and if the two differ, it means they have an SST reading. If there’s a SST reading, the cell gets put into the permanent “time-independent mask”. But again, that’s an assumption.
Regardless, SST data in areas with occasional sea ice appear to be permanently masked. That is, they are not used in (deleted from) the analysis of global temperature. And as noted above, GISS chooses to use the dataset with the higher trend (land) and that choice biases their version of global temperature.
All right, we already know that GISS uses less instead of more information for assembling its global temperature dataset. But how could they explain the deletion of available Arctic SST data??
They try to explain their ‘divergence problem’ with differences in Arctic coverage, by stating that they have more data than all the others, including CRU and NCDC. In fact, they have less data and they are even deleting a significant part of it – maybe in order to get the desired results.
I’ve rarely seen such hipocrisy from a high-profile institution. Hansen is going to get some tough questions.
Steven mosher says:
So, what value would you suggest that people use when an ocean area is covered by ice? The value of the SST under the ice? -1.8C. no value? or the value of a land station 100s of km away ( like -21C)
Those are your choices, can u think of others. neat problem.
Yes. Acknowledge that temperature anomalies have a meaning only locally and cannot be integrated into a global map from which a global anomaly is made and then used to define whether and how much the planet is heating.
This was displayed to me graphically by a recent weather here in Athens, Greece: a hot wind came, 27C, with a lot of clouds, which means no sun to heat the building. There is very little energy content in the air so the house remained a cool 23C just by shutting the windows. Next day with the sun full force hitting the walls the inside and outside temperature balanced, no matter the closed windows.
It is the heat content that is important. The poles can show huge anomalies in the winter, even without extending land temperatures. The winds do that anyway. Even if one had thermometers every 100m^2 there would be little meaning in the value for the energy in the area. It is the skin surface that radiates as T^4 and that is either the ice temperature or the water temperature.
Interpolating coastal land temperature anomalies over the Arctic basin should be regarded as a serious flaw. The variance of anomalies are at least 3 times larger over land surface than over open water or sea ice.
Furthermore, in case of a reduction in sea ice coverage the coastal areas can experience very high temperature anomalies because of the missing cooling factor. The local anomalies on the coastline could be abnormally high compared to the Central Arctic basin and the Siberian or North American interior. These skyrocketing temperatures are measured by numerous Arctic stations, including Barrow (Alaska), Vrangel Island (Northern Siberia), etc. while nobody measures the actual air temperature in the Central Arctic Ocean.
Steve Mosher, You wrote, “Merging hadisst and reynolds thus would have the effect of not using the reynolds SSTs where there is ice or partial ice. It appears that is what they do.”
What they appear to do based on my review of the maps of their outputs through KNMI is not use Reynolds SST where there has been ice or has been partial ice. That is, Reynolds SST data is masked (deleted) if there was ever ice in the grid.
You asked carrick, “So, what value would you suggest that people use when an ocean area is covered by ice? The value of the SST under the ice? -1.8C. no value? or the value of a land station 100s of km away ( like -21C).”
The GISS 1200km smoothing seems logical for infilling temperatures for areas with sparse surface readings, though it’s a subject of much debate. And in areas with sea ice, extending land data out over sea ice, when it is sea ice, also seems logical. But, during periods when SST data is available, it should be used. Masking SST data and extending land surface data in its place biases the data in those areas, and that’s what this post was about.
This is an artificially created problem.
On land they don’t stop measuring the surface temperature in the Winter. Doesn’t matter if it’s a bare-land Winter, a snow-covered Winter, or an ice-covered Winter. They just measure the temperature of the air at the surface, business as usual. They don’t stick thermometers into the ground under your driveway. If I go ice-fishing on a lake in Manitoba or Wyoming, I know exactly what the water temperature will be at the bottom of the hole I drill. It will be 0°C. But is that what gets listed in the temperature records? No. What gets listed is the air temperature. What do they do in the Summer? Same thing. For these purposes, nobody cares that the water surface is at 0°C, or the dirt in the back yard is at 10°C, or that the core of the planet might be 6000°C, nor should they.
On the ocean, however, they get schizoid for some reason. When there’s no ice, they measure the Sea Surface Temperature. When there is ice, they appear to be flummoxed. Why? Beats me. If they really want to know the SST, i.e. the temperature of the liquid surface under the ice, there’s no mystery about that. It’s pretty much the same value anywhere on the planet where sea ice forms, namely -1.8°C. It could be a little higher than that if the ice formed very slowly and squeezed out a bit of salt while freezing. Unless it is abnormally salty, however, and froze very quickly, it will not be colder than -1.8°C, nor warmer than 0°C, no matter how much salt was squeezed out. That’s a pretty tight window, and it could be tightened a lot more with a few measurements.
OK then, so what happens if I go ice-fishing offshore in Nunavut or Alaska? Do I care that the water at the bottom of my fishing hole is always -1.8°C? Not really, especially if I’m freezing my buns off because the air temperature is -50°C. And if they’re interested in getting a set of values for the temperature of the surface of the Earth, they should be using values for the surface of the Earth, regardless of whether it’s covered with grass, ice, or peanut butter.
There’s no real conceptual problem here. If you want to know the temperature of the liquid surface of the sea, it’s either whatever it is, or it’s -1.8°C. If you want to know the temperature of the air above the surface of the sea, it’s simply whatever it is.
One last point about this 1200km thing: I live at 45°N. One entire time zone here is 1180km across, or 20km shy of GISS’s ‘reach out and touch me distance’. The temperature across town can be several degrees up or down from where I am sitting. It’s pretty random. At highway speeds, i would have to drive non-stop for 12 hours to visit my friends who live one time zone away. What are the odds that their temperature right now is the same as mine, or anywhere close to it? Abject nonsense.
/dr.bill
Manfred says:
June 1, 2010 at 12:26 am
Steven mosher says:
May 31, 2010 at 9:37 pm
“The DIFFERENCE between GISS and CRU is not some nefarious plot. The difference is caused by a analytical choice on how to handle these cases. The difference in results is an uncertainty in the estimate of global temperature.”
The difference is not an “uncertainty”, but a GISS warming bias. They use data with a 8-fold higher warming trend to interpolate sea surface data.
This is an important posting by Bob Tisdale, as it proofs very easily and understandable a signifant warming bias in GISS data, particularly the prominent polar regions.
If there is a warming bias it will only occur when teh arctic is warm. If, as everyone seems to think, we are in a cycle (solar, PDO whatever) which has or is about to change then the GISS extrapolkation will introduce a cooling bias and we should see the large drop in arctic temperatures similar to what happened between 1940 and 1970. It’s a case of swings and roundabouts.
However, I still think GISS is closer to the mark than Hadcrut. Hadcrut anomalies have been about the same as the satellite anomalies recently which is ridiculous when you consider the different base periods used. I can understand the attraction of Hadcrut at the moment. UAH NoPol anomalies are ~2 deg above the 1979-98 mean and RSS clearly shows the arctic region is ‘warmer’ than normal.
http://www.remss.com/msu/msu_data_monthly.html?channel=tlt
Figure 8 http://i48.tinypic.com/spd4li.jpg
GISTEMP shows a temperature increase of about 2.6°c between 1982 and 2009, which is equal to a trend of 1.0°c per decade. No one measured this trend directly, because as I mentioned it before, there are no permanent temperature recording sites within about 1000km of the North Pole.
The northernmost meteorological stations are located at a latitude of about 80N – Alert (Nunavut), Nord Ads (Greenland), Ostrov Vize (a Russian island). The 1982-2009 trend is 0.7°c per decade in this latitude band which is probably amplified by sea ice reduction.
It is mathematically impossible to get a warming trend of 1.0°c per decade (or anything higher than 0.7) in the Central Arctic basin with spatial interpolation. It is only possible by extrapolating the zonal mean values shown on Figure 8.
Steve Mosher: To illustrate my investigations of the subject at hand mentioned in my earlier reply, here’s a sample graph of Reynolds SST anomalies and the anomalies of the two GISTEMP combined LST+SST (250km and 1200km) datasets for the grid 75N-77N, 45E-47E. Note that there are only two datasets plotted. Reason: There is no data for the GISTEMP combined LST+SST (250km) dataset The output from the KNMI Climate Explorer read, “No valid data. Please check your choices on the previous page.”
To me that means that the GISTEMP combined LST+SST (1200km) data for that cell is only land surface temperature data that has been extended out over the ocean.
http://i48.tinypic.com/30c60ww.jpg
And here’s a plot of the Reynolds sea ice cover (presented as percentage). There has been sea ice in that cell, so GISS masked (deleted) the SST data.
http://i49.tinypic.com/11bja1l.jpg
There is no need to merge water temp with air temp in the arctic. We have over 30 years of of air temp records and that should suffice. From the above link:
“The statistics of surface air temperature observations obtained from buoys, manned drifting stations, and meteorological land stations in the Arctic during 1979–1997 are analyzed. Although the basic statistics agree with what has been published in various climatologies, the seasonal correlation length scales between the observations are shorter than the annual correlation length scales, especially during summer when the inhomogeneity between the ice-covered ocean and the land is most apparent. During autumn, winter, and spring, the monthly mean correlation length scales are approximately constant at about 1000 km; during summer, the length scales are much shorter, i.e. as low as 300 km. These revised scales are particularly important in the optimal interpolation of data on surface air temperature (SAT) and are used in the analysis of an improved SAT dataset called IABP/POLES. Compared to observations from land stations and the Russian North Pole drift stations, the IABP/POLES dataset has higher correlations and lower rms errors than previous SAT fields and provides better temperature estimates, especially during summer in the marginal ice zones. In addition, the revised correlation length scales allow data taken at interior land stations to be included in the optimal interpretation analysis without introducing land biases to grid points over the ocean. The new analysis provides12-hour fields of air temperatures on a 100-km rectangular grid for all land and ocean areas of the Arctic region for the years 1979–1997. “
Hi Anna V-
“It is the heat content that is important. The poles can show huge anomalies in the winter, even without extending land temperatures. The winds do that anyway. Even if one had thermometers every 100m^2 there would be little meaning in the value for the energy in the area. It is the skin surface that radiates as T^4 and that is either the ice temperature or the water temperature.”
Good point!
John Finn wrote, “If there is a warming bias it will only occur when teh arctic is warm. ”
It will occur anytime sea ice declines and there is SST data available to be included in the GISTEMP product.
@davidmhoffer says:
May 31, 2010 at 9:54 pm
“There are a few odd years in there but the whopper is 1917 which has annual temp anomalies ranging to -6 in the arctic.”
How does that compare with the other cold winters last century; 1940, 1947 and 1963.
http://climexp.knmi.nl/data/tcet.dat
Whether or not there was a SSW will make a big difference;
http://www.appmath.columbia.edu/ssws/index.php
DR wrote, “Apparently GISS is making assumptions about polar amplification due to ‘global warming’, but where is the justification for such speculation? Ah yes, climate models.”
Polar amplification is a natural effect and occurs whenever climate changes. If the Northern Hemisphere is warming, the higher latitudes warm more than lower latitudes, and when the Northern Hemisphere is cooling, higher latitudes cool more. No need for climate models on that one.
Regards
Nick Stokes says:
May 31, 2010 at 8:11 pm
Well, the obvious reason for not plotting SST data in areas covered by ice is that SST is not then measured. I presume your complaint is about areas that have SST measurements in summer only, but are iced in winter. How do you think they should deal with regions with only summer SST readings when compiling an annual average plot?
___________________________________________________________________________
The way any true scientist would. You use the actual real life measured data and you leave it BLANK when you can not measure it with an explanatory note. Also you do not go ,b.removing,/b. gobs of data (weather stations) and substitute a 1200km interpolation function for the real available data. This causesThe Bolivia Effect
As far as I am concerned this type of messing with the data makes the entire data set completely useless and that is before you get into the temperature data error Also note as you read that article standard error is a cop-out which deflates the error! SE is the SD divided by the square root of the number of measurements, so it is much smaller than sd when the significant of the measurements are judged.
Friends! Romans! Countrymen!
Lend me your ears…
There is Art and there is Science!
and there is the Science of Art
and the Art of Science
Here we have the Art of Science!
Gaze upon the pretty pictures
Breath in the artist’s message
Let it transform your soul
We are here for but a little while
And then, poof, we are gone!
Do not disect the work
Chew it, taste it, swallow it whole!
Devour it and let it become part of you!
Do you feel it?
Soon it will churn your guts
It will transform you
Make you a new person
Let you see all that is seeable
Let you feel all that is feelable
Let you smell all that is smellable
etc.
etc.
etc.
Ours is not to wonder why
Ours is but to swallow and die..
Of course, nobody should take any notice of these kinds of blog posts. No self-respecting climate scientist would do so. After all, they are written by amateurs and they are not in the peer-reviewed literature. Science is a closed shop, Bob – surely you realize that by now!
I added an update to the version at my blog that clarifies what is being illustrated:
http://bobtisdale.blogspot.com/2010/05/giss-deletes-arctic-and-southern-ocean.html
It reads:
There appears to be some confusion on the WattsUpWithThat version of my post GISS Deletes Arctic And Southern Ocean Sea Surface Temperature Data about what this post illustrates. I prepared a graph for this post but chose not to use it since it appeared redundant to me. It should clarify what is being presented. It is a comparison graph of GISTEMP Arctic Surface Temperature anomalies for the grid 65N-90N, 40E-40W, which is a major portion of Arctic Ocean as shown above in Figure 4. One dataset is the combined land plus sea surface data; the other is the land-only data. The two datasets are identical. If you subtract one from the other, the difference is 0.0 (zero) for all months. This indicates that there is no Sea Surface Temperature data in the combined product in this grid.
http://i48.tinypic.com/34o3hjq.jpg
Update Figure 1
Does the Sea Surface Temperature data exist for this area? Yes. It is illustrated above as the green curve in Figure 5.
To me, this indicates that GISS deleted the sea surface temperature data for this portion of the Arctic.
END UPDATE
REPLY: Thanks Bob, but doesn’t that “WattsUpWithThat version of my post” sort of imply that WUWT made a mistake? It is your post verbatim at your invitation to repost it. I made no changes to it. – Anthony
“Crafted story” data is better than real data, when you are trying to prove a hoax.
Why not let everyone have the actual data to make up their own story with? With a open plot — Supply the actual software, open the procedures, the analysis, no more guessing games, just the facts. Then we can at least argue from the same known point …
What happened with searching for the truth? And is science ever settled?
Nick Stokes says:{May 31, 2010 at 8:11 pm}
” How do you think they should deal with regions with only summer SST readings when compiling an annual average plot?”
Perhaps a good reason to view “annual averages” as meaningless.
Anthony: Sorry. I missed a prepositional phrase in there. Thanks for picking it up. The confusion is in the comments, not in the WUWT version of the post. I’ve changed the initial sentence to read: There appears to be some confusion in the comments in the WattsUpWithThat thread of my post…
Better?
There is another way of looking at this. Here is a map of the Arctic sea surface temperatures at the warmest they have ever gotten in the modern era, September 15th, 2007.
http://www7320.nrlssc.navy.mil/GLBhycom1-12/navo/arcticsst/nowcast/sst2007091918_2007091500_903_arcticsst.001.gif
In the 75N circle, the sea surface temperatures are mostly -2.0C and slightly lower, the area which is still covered in ice .
In some areas in the Arctic basin, however, sea surface temperatures got as high +6.0C.
Generally, this map points to a physical limit to how high GISS can make the anomalies, at least in September. The 1200 km smoothing algorithm should show ZERO anomaly for the areas which are still ice covered in September because the sea surface temperatures should be close to -2.0C and slightly lower.
If you look at the GISS monthly zonal anomaly map – from 2006 to today – station temperatures only – 1200 km smoothing -, they have temperatures in the far north averaging about +3C (and as high as +9.2C) above normal in September which is entirely unphysical. ALL the ice would have melted in September over the last 4 years.
http://data.giss.nasa.gov/work/modelEt/time_series/work/tmp.4_observedTs_1_2006_2010_1951_1980-0/map.gif
The same map using SST numbers from HADISST and Reynolds have much lower anomalies (closer to the reality of the ice melt) but it is cut-off at 80N and 65S.
http://data.giss.nasa.gov/work/modelEt/time_series/work/tmp.4_observdSST_1_2006_2010_1951_1980-0/map.gif
There is a problem here.
Anto wrote, “Of course, nobody should take any notice of these kinds of blog posts. No self-respecting climate scientist would do so. After all, they are written by amateurs and they are not in the peer-reviewed literature. Science is a closed shop, Bob – surely you realize that by now!”
But there are those who have read this post and understand the significance of the bias created by deleting SST data in areas of seasonal sea ice. It now offers them a source during discussions with other bloggers who claim that what GISS does, does not create a bias, when in fact it does create a bias.
Regards
Bob Tisdale says:
June 1, 2010 at 3:50 am
John Finn wrote, “If there is a warming bias it will only occur when teh arctic is warm. ”
It will occur anytime sea ice declines and there is SST data available to be included in the GISTEMP product.
Which like I said will have the reverse effect when the arctic cools and sea ice increases . If there are cycles ati npaly as everyone on this blog seems to think then it will all come out in the wash. GISS exhibits a big increase (~2 deg) in arctic temperatures between 1910 and 1940 and a big drop ( >1 deg) between 1940 and 1970. The global decadal trends between 1982 and 2009 for GISS, HadCrut and UAH respectively are ~0.18, ~0.17 and ~0.16. RSS is ~0.18. While your observations and analysis are interesting I think they need to be put in perspective.
climatepatrol: But if you create a weighted average of GISTEMP global land surface temperature anomalies with 1200km radius smoothing (30%) and Reynolds (OI.v2) global SST anomalies (70%) including the SST data where they delete it, the weighted average of the two datasets used in the combined GISTEMP product also flattens over the past decade:
http://i49.tinypic.com/11j22h0.jpg
In other words, the “weighted average” of the sources does not equal the end product with GISTEMP but it does with HADCRUT and NCDC combined data.
@davidmhoffer says:
May 31, 2010 at 9:54 pm
“There are a few odd years in there but the whopper is 1917 which has annual temp anomalies ranging to -6 in the arctic.”
How does that compare with the other cold winters last century; 1940, 1947 and 1963.>>
Even more odd. land/ocean(noaa)
1940 +1.0
1947 +1.8
1963 -1.1
Part of the confusion I think is from what they are measuring. If you start in 1881 the top 5 degrees N latitude is mostly grey meaning no data. The first data appears in 1911 with a stripe from about 45w to 80e. In 1933 it turns into 45w to 160e. in 1939 it is 45w to 180e and in 1947 it is 170w to 180e.
When I run 1881 to 1946, 45w to 180 w is entirely grey, no data
When I run 1947, 45w to 180w shows +1 to + 2 in that section
When I run 1881 to 1947, it shows EXACTLY the same thing in that section.
1947 alone blends nicely into the areas to the south. 1881-1947 stands out like a sore thumb. So it seems clear that it is taking the 1947 data for that section and using it for the entire 1881 to 1947 time period.
Same thing with the block from 80e to 160e. It is grey until 1933. If you run 1881 to 1932 it is grey. If you run 1882 to 1933 it is filled in, my assumption again being that they have used a single year of data to cover the 1882 to 1933 period.
How all this relates to the gisstemp zonal mean for 64n to 90n is beyond me.
Mike Bryant says: “Someone needs to go over to GISS and break all the red crayons…”
They’d still need some red crayons, but not as many for coloring in the Arctic and Southern Oceans during seasonal sea ice decline. They could always use what they’ve got left over and mix them in with some yellow crayons.
Steven mosher says: “Carrick, From memory ( dont trust but verify) I think the way the GISS program is going to work is as follows. The 1 degree SST bins are going to be mapped into the 8000 equal area subgrids. if there is an SST present you will get a SST value. if ICE is present you get NA. Then the entire subgrid structure is averaged. So its not entirely clear that only land values will contribute to the value. IF your grid goes: land:ice:land”
From the update that I referred to above, here’s a comparison graph of GISTEMP Arctic Surface Temperature anomalies for the grid 65N-90N, 40E-40W. One dataset is the combined land plus sea surface data; the other is the land-only data. The two datasets are identical. If you subtract one from the other, the difference is 0.0 (zero) for all months. This indicates that there is no Sea Surface Temperature data in the combined product in this grid, and that’s a chunk of the Arctic Ocean.
http://i48.tinypic.com/34o3hjq.jpg
The standard GISTEMP combined product is not land plus sea surface data for the coordinates 65N-90N, 40E-40W; it is land surface data only.
HadSST Arctic data: no net warming since 40ties
http://climexp.knmi.nl/data/ihadsst2_0-360E_66-90N_na.png
Antarctic data: cooling since 80ties:
http://climexp.knmi.nl/data/ihadsst2_0-360E_-70–90N_na.png
By not accounting for polar regions, GISS diverges from HadCRUT more and more.
http://www.woodfortrees.org/plot/gistemp/from:1998/plot/hadcrut3vgl/from:1998
“Man made” warming accounts for 0.3 deg C since 1998. J. Hansen is a movie archetype of a mad scientist.
Bob,
WRT 65-90N 40e-40w. cool. I can probably ( give me a few days ) go pull all the historical data for the SSTs in that zone and the land temps in that zone. Can we work with a narrower slice in long? thats a big chuck. Currently it will have to be HADISST only till I find a good source of asscii for the reynolds product.
Bob, check out the thread on LUCIA WRT GISS mystery. Contact Zeke. He’s a got a GISS ocean data set
John Finn wrote: “Which like I said will have the reverse effect when the arctic cools and sea ice increases . If there are cycles ati npaly as everyone on this blog seems to think then it will all come out in the wash.”
But as long as global temperatures rise and polar amplification causes the Arctic to warm at a higher rate, GISTEMP will exaggerate the rise in global temperature by deleting sea surface temperature in areas of seasonal sea ice decline and extending land surface data over the ocean.
You concluded you comment with, “The global decadal trends between 1982 and 2009 for GISS, HadCrut and UAH respectively are ~0.18, ~0.17 and ~0.16. RSS is ~0.18. While your observations and analysis are interesting I think they need to be put in perspective.”
They are in perspective. This post was about biases created by GISS in their GISTEMP product, not about HADCRUT or the two TLT datasets.
Also, if you’re not aware, the Hadley Centre created an upward bias in their HADSST2 dataset in 1998, when they spliced two SST datasets together. The HADSST2 data represents about 70% of the HADCRUT3 product. I covered it in this post about the Met Office’s prediction about 2010:
http://bobtisdale.blogspot.com/2009/12/met-office-prediction-climate-could.html
Regards
Bill Illis: I can’t get your GISS links to work.
“Generally, this map points to a physical limit to how high GISS can make the anomalies, at least in September. The 1200 km smoothing algorithm should show ZERO anomaly for the areas which are still ice covered in September because the sea surface temperatures should be close to -2.0C and slightly lower.”
As I’ve said before in the GISS SOURCE DATA for 1880-1992 THERE IS NO SST
where there is ICE. there is a missing value. do you
1. fill in this missing value with a SST from under the ICE OR
2. estimate the airtemp OVER THE ICE by using land station data from miles away?
I suppose if they used the SST under the ICE (-1.8C to 2.0C) as opposed to the airtemp over the ice people would argue that some months the air over the ice is Colder and that they were biasing the record by using unobserved SST.
using land temps from nearby stations is one solution. I know 1200km freaks people out. hmm. if I’m stuck in the arctic, stranded at 0.0 with 1200km to walk which way would I walk? hehe. thats a joke, but also folks need to consider correlation scales in that part of the world.
During autumn, winter, and spring, the monthly mean correlation length scales are approximately constant at about 1000 km; during summer, the length scales are much shorter, i.e. as low as 300 km.
Does 1,000 mean between 1,010 and 990; between 1,100 and 900 or even between 1,500 and 500?
Similarly what is the what is the accuracy of the 300 number?
The new analysis provides12-hour fields of air temperatures on a 100-km rectangular grid for all land and ocean areas of the Arctic region for the years 1979–1997
Is this new data or has the old data been somehow “upscaled” to a higher “resolution” than it was actually collected at?
Bob,
Judy Curry complained the other day that the SST data was a bit of a challenge ( err mess ) At some point the folks doing there own GTIs are bring in SSTs. Your knowledge of the datasets would be really helpful..
John Blake says:
May 31, 2010 at 6:47 pm
Things really have gotten to the point where nothing any official body cobbles up regarding SST or any other climate indicator seems even worth discussing.
I do agree. Why to get worried about technical details where the motivations are not technical. We should discuss how to smooth the smoothers. Do they belong to a Union of Warmists as nobody can remove them from their jobs?…kind of “Warmists United will never be defeated”?
Bob 7:53 am
Sorry, I hate when the links don’t work.
I’ve saved and uploaded them here.
http://img243.imageshack.us/img243/8504/giss1200kmsmoothingmap.gif
http://img535.imageshack.us/img535/4999/hadsst2006map.gif
Or you can get them from GISS here. (change the Mean period to 1 month and the Time interval to 2006 to 2010 so the resolution is high enough – there are different options for which temperature series to use).
http://data.giss.nasa.gov/gistemp/time_series.html
Enneagram:
there is a technical question on the table. we know the data for SST is missing for those cells under ice ( in hadisst) when construction an average of global temps how would you handle that. infill with some constant or extrapolate from the nearest land station. Observe: the water temp under the ice will likely be around -2.0C. the air temp above the ice can be much colder or warmer. your constructive suggestion is welcomed
Steven mosher says: “Judy Curry complained the other day that the SST data was a bit of a challenge ( err mess ) At some point the folks doing there own GTIs are bring in SSTs. Your knowledge of the datasets would be really helpful.”
Judy Curry is observant. I’d be happy to help. When and where and how? You can leave me your email address in a comment at my blog. I won’t post it. Or you could ask Anthony for mine.
Bob Tisdale,
you did a very good analisys.
I can’t understand what is the point of Steven Mosher.
To estimate the surface temperature in the polar regions is not as easy as elsewhere?
So, that doesn’t mean you have to do it in a stupid way as GISS does.
Anyway, you said that a warm bias “will occur anytime sea ice declines”.
But in the Antarctic ocean there is not a general decline in sea ice, we are told, yet Gistemp is warmer than UKMO-CRU also in the southern ocean.
Why don’t we have a thermometer at the North Pole? Surely an automated station with a radio-thermal power supply reporting by radio would only need to be visited once a year (if that) to reposition it. An Osprey VTOL refueled in midair a couple times by a KC-130 tanker could get in and out whenever needed. Not an inexpensive proposition but compared to big ticket budget items in national climate research like satellites its cost would be like one drop in a bucket.
Steven mosher: You asked, “Can we work with a narrower slice in long?”
The effect in the output should be the same no matter which slice in longitude you pick. 40E to 50E does seem to get a lot of sea ice variation.
http://i48.tinypic.com/14nh6rq.jpg
But don’t put yourself out in an attempt to satisfy me. I’m more interested in what GISS presents to the world as an output, and that’s visible in the comparison of the GISTEMP land-only and combined products.
http://i48.tinypic.com/34o3hjq.jpg
There’s no difference in the data, which again means the SST data has been deleted between the time it is downloaded and the time it is output.
Regards
Bob,
Just ask charles. leave a note to the moderators. he will probably read it tonight.
Bill Illis: Thanks for the links to the plots you were talking about. You wrote earlier, “The same map using SST numbers from HADISST and Reynolds have much lower anomalies (closer to the reality of the ice melt) but it is cut-off at 80N and 65S.”
And for the data north of 65N, that data should only represent the SST data north of the North Atlantic and Scandinavia.
I agree with dr.bill says:
June 1, 2010 at 2:37 am
We do not measure the temperature of the ground over land, we measure the Air Temp.
So why not do the same over the Sea, by using the Sea temp you must bias it a lot warmer in the Winter, but then again the Gridded Global temp is pretty meaningless anyway to average folk.
It is only important to those with a “CAGW” message to try and frighten us with.
I’ve gone through the Global Map images in some detail and it is clear that geographical coverage of the high arctic was expanded in 1933, 1939, and 1947. My question is how would this data have been incorporated into the over all 64N to 90N zonal mean that gisstemp reports?
If the data was simply added as additional anomaly data against the 1951 to 1989 mean and averaged across the geography, then polar amplification would result in the region as a whole showing higher anomaly changes than before those stations were added. If an adjustment was made, on the other hand, to arrive at an average temperature adjustment for the geography with the additional coverage normalized, the colder raw temperatures would suppress temperatures for the period the extra coverage was added in. This time period (1930 to 1950) featured a large rise in global temperatures. I would be curious to understand if adding this additional coverage influenced that period up or down. After 1947 my expectation is that it would make no difference, but prior to 1933 it would be difficult to compare trends for that latitude band versus current coverage.
dr. Bill,
I really enjoy your posts. Thanks and keep them coming! 🙂
I’ve gone through the Global Map images in some detail and it is clear that geographical coverage of the high arctic was expanded in 1933, 1939, and 1947.>>
oops and 1912 as well. That was the beginning of that warming period. Did the additional coverage reduce, increase, or make no difference to the anomaly data for the area as a whole.
Steven mosher says:
June 1, 2010 at 7:56 am
Enneagram:
there is a technical question on the table. we know the data for SST is missing for those cells under ice ( in hadisst) when construction an average of global temps how would you handle that. infill with some constant or extrapolate from the nearest land station. Observe: the water temp under the ice will likely be around -2.0C. the air temp above the ice can be much colder or warmer. your constructive suggestion is welcomed
_________________________________________________________________________
Steve, Would not the temperature just above ice be 0C or colder thanks to the ice acting as a heat sink?
Gail Combs says:
June 1, 2010 at 11:35 am
Steven mosher says:
June 1, 2010 at 7:56 am
Enneagram:
there is a technical question on the table. we know the data for SST is missing for those cells under ice ( in hadisst) when construction an average of global temps how would you handle that. infill with some constant or extrapolate from the nearest land station. Observe: the water temp under the ice will likely be around -2.0C. the air temp above the ice can be much colder or warmer. your constructive suggestion is welcomed
_________________________________________________________________________
Steve, Would not the temperature just above ice be 0C or colder thanks to the ice acting as a heat sink?
_________________________________________________________________________
Out of curiosity I tried a kitchen experiment. I keep gatorade quart jugs 1/3 full of ice in the freezer. I took one out filled it with water and set it in the frig for a couple of hours so it would come to equilibrium at about 0C/32F per thermometer reading of the water above the ice. I then poured out the water and suspended a thermometer 4″ above the ice and let stand for another 2 hours.
The kitchen is 82F and the thermometer above the ice read 43F so the ice does modify the temperature directly above its surface when there is no wind.
I would suggest a better experiment. The historic range of temperatures reported by official sources for arctic summers on land is available. Set a walk-in type fridge to the hi, lo and mid range of those settings and suspend a thermometer over ice at the regulation height. Take a number of reading at the various temperatures. This should give you at least a ballpark figure for the offset in temperature due to the ice. You can also do this for water at the temperature reported for the open water in the arctic ocean.
Compare the lab results to what ever real life results you can get your hands on.
Anyone who sail will tell you the temperature over open water in the summer is a lot cooler than that on land. This can be verified with SST vs land temps in nearby areas. (Note I am talking about summer. In the winter the temps are warmer. This is known as the “lake effect” to those from the great lakes area.)
A C Osborn: Regarding Sea Surface Temperature and Marine Air Temperature:
Sea surface temperature data was determined by the powers that be to be more reliable than the air temperature dataset for the oceans known as Night-time Marine Air Temperature or NMAT, hence the use of sea surface temperature in global temperature products.
Gail Combs, Enneagram, Steven mosher: Regarding polar HADISST data in early years:
Keep in mind that most of the polar ocean SST data in early years are infilled completely. There were very few SST readings in southern portions of the Southern Ocean as late as the 1960s and 70s. Sample of the ICOADS readings in 1975:
http://i49.tinypic.com/2rm71aw.jpg
I set the contour levels low so that the data locations would show up better.
Even the Southeast Pacific was still bad in 1975, and for the datasets that aren’t satellite based (HADSST2 and ERSST.v3b), it’s not much better there now. No ship traffic yields no SST readings.
Also keep in mind that major portions of the SST data in all ocean basins in the early part of the instrument temperature record are also infilled. Here’s a one-month snapshot of the ICOADS SST readings in 1885:
http://i50.tinypic.com/2pt2bkn.jpg
From that, the Hadley Centre fills in the rest for their HADISST dataset, and NOAA does the same for their ERSST.v3b data. The HADSST2 data leaves the gaps but expands the grids to 5deg lat by 5deg long.
Well-said Bob:
Bob Tisdale wrote (in response to Nick Stokes):
“How do I think GISS should deal with areas with seasonal sea ice? Not sure what you feel would be so difficult. Use SST data when it’s available. They could extend land surface data out onto ice when there is no SST data. But if the land surface data can’t extend to its full 1200km because there’s open ocean in the way, they could then use the Reynolds SST climatology or another climatology for the remaining sea ice areas.”
Bob Tisdale wrote:
“If the SST data is available, why not use it?”
You’ve made it simple for the administrators.
I’m with Anna V on the folly of mixing intensive metrics such as temperature with extensive metrics of energy. And this goes double for the folly of mixing air temperatures with SSTs. What you get is an artificial measure of nobody-knows-exactly-what!
dr bill,
Your suggestion that they should use SST= -1.8C in winter misses the point of SST measurement. It’s about climate. When there is a water/air interface, air near the surface is near thermal equilibrium with the water, so SST is a stable accessible proxy for average air temperature. When there is no interface, it’s not. Winter Arctic air temperatures are far lower than -1.8C.
Gail Combs,
I analysed the “Bolivia effect” here. It’s pretty weak stuff.
Gail Combs: June 1, 2010 at 11:35 am
re “walk-in fridges and blocks of ice”
That is a brilliant suggestion! And I mean that in the North American sense, i.e. genuinely brilliant.
/dr.bill
anna v, you reinforce Bob’s points nicely.
When there is ice: continental.
When there is not: maritime.
Simple alternation of state.
Administrators: Is it too much to ask your model-oriented staff to handle these 2 physically distinct states appropriately?
Hi Nick,
If you have another look at what I wrote, you’ll see that I wasn’t suggesting that they use -1.8°C for anything at all. I’m pretty ‘old school’ about data, and I don’t like proxies a whole lot. They tend to fall into that most dangerous category of ‘things you know that aren’t actually true’. I’m a theoretical physicist, and while we’re adept at whipping up models on short notice, we also require that our models line up with reality, i.e. measurements. There doesn’t seem to be much ‘rubber hitting the road’ in climate affairs, and applying elaborate mathematical techniques to ‘things we know that aren’t true’ won’t make them any truer.
/dr.bill
Gail Combs says:
June 1, 2010 at 1:25 pm
Well, at least during daytime. At night the land is often cooler and the lake breeze is replaced by a land breeze.
In Northeast Ohio we generally used “lake effect” to refer to big snowfalls due to cold air picking up moisture from Lake Erie and dumping it as snow as the terrain lifts the wind.
I used to patronize “Lake Breeze Orchard.” He wouldn’t have done as well had he named it “Lake Effect Orchard.” 🙂
Paul Vaughan says:
June 1, 2010 at 3:42 pm
> Administrators: Is it too much to ask your model-oriented staff to handle these 2 physically distinct states appropriately?
Staff? What staff? Staff implies payroll. There’s a payroll?
Ric Werme,
Did you really misunderstand that I was addressing Anthony & WUWT??
[ :
Clarification:
We’re talking about GISS here (not Anthony Watts & WUWT moderators, who do a fine job).
Anthony/Bob, you should consider having this reprinted over at Biggovernment.com. It would open up some folks who keep an eye out for government funny business that may not know about your site.
Bob, I encourage you to run a “super-succinct version” of this article in a near-future “remake”. There’s some undeniable substance in this article, but it may be lost on some folks who generally “ski” (skim & skip) through “technicalese”. In time the evolving narrative will sharpen into a classic fireside story.
reynolds sst are basically a costant over frozen sea and that’s the reason for the very little trend compared to gistemp, anyway the air above sea ice warmed much more so there no reason to compare sst trend with gistemp….after all uah msu polar ocean trend is +0.51°C/decade much closer to giss and polar amplification is stronger near the surface.
lor82 says: “reynolds sst are basically a costant over frozen sea and that’s the reason for the very little trend compared to gistemp, anyway the air above sea ice warmed much more so there no reason to compare sst trend with gistemp….after all uah msu polar ocean trend is +0.51°C/decade much closer to giss and polar amplification is stronger near the surface.”
You missed the point of this post. GISS deletes the Sea Surface Temperature data when there is open ocean in areas where there is seasonal ice. The data exists but GISS elects to delete it. If there’s ice poleward of the Siberian and Canadian Arctic shoreline, extending the land surface temperatures out over the ice is a logical way to infill missing data, However, when there is sea surface temperature data available it should be used. They claim to track sea ice cover as well. So a simple area weighted average of SST & LST data is all it takes. And if GISS were to do that, then their trend would be more in line with the Arctic TLT trend you quoted:
http://i45.tinypic.com/19b095.jpg
Regards
lor82 says: “…polar amplification is stronger near the surface.”
Please provide a reference for this. Thanks.
@Bob tisdale
I’ve not missed the point, i was commenting this:
“By deleting the Sea Surface Temperature anomaly data, GISS relies on the dataset with the greater month-to-month variation and the much higher temperature anomaly trend for its depictions of Arctic temperature anomalies. This obviously biases the Arctic “combined” temperature anomalies in this area.”
this is not correct, the greater month-to-month variation and the much higher temperature anomaly trend is just due to sst below sea ice being almost constant to -1.8°C….to demonstrate this statement you should replace every gistemp grid pixel with reynolds sst every time the sea is ice free and then you can say if this makes any difference or not….but usually ssta are very high in regions without sea ice..
serreze 2009:
http://www.the-cryosphere.net/3/11/2009/tc-3-11-2009.pdf
Pay no attention to that chap behind the curtain. He is just a diversion to stop you from watching what is really going on.
The Gaianist Language does not contain words like ‘average’, ‘standard deviation’, ‘smoothing’, ‘trendline’. etc
You will find NO statistical Mathematics Text books in Gaia’s Library.
Her mantra is:- “What happens next, starts from here and now !”
No experiment that gaia performs on this planet pays any attention to ‘averages’ ; which are a human aberration. She deals with the earth as it now is; and each and every element reacts now to its own environment; without any reference to what any other element sees.
So Gaia does no 1200 km smoothing; nor any 250 km smoothing; she has a thermometer on every single atom; and it will do what it is supposed to do based on what it knows.
By paying such attention to detail; Gaia always gets it right. There is no peer reviewed literature on Gaia’s mistakes; she makes none, and gets the right answer every time.
So only fools throw away that much information; and replace it with less information; based on some computer smoothing of what started out as perfectly good experimental information.
So my advice to statistical mathematicians is to find a real job; and stop erasing real information in exchange for faux misinformation.
Mother Gaia knows what’s next.
lor82: You replied, “I’ve not missed the point, i was commenting this:
‘By deleting the Sea Surface Temperature anomaly data, GISS relies on the dataset with the greater month-to-month variation and the much higher temperature anomaly trend for its depictions of Arctic temperature anomalies. This obviously biases the Arctic ‘combined’ temperature anomalies in this area.’”
You continued, “this is not correct, the greater month-to-month variation and the much higher temperature anomaly trend is just due to sst below sea ice being almost constant to -1.8°C….to demonstrate this statement you should replace every gistemp grid pixel with reynolds sst every time the sea is ice free and then you can say if this makes any difference or not….but usually ssta are very high in regions without sea ice…” and provided a link to Serreze et al (2009).
First, I qualified the graphs in Figures 5 and 7 a few of sentences earlier. I wrote, “Basically, GISS excludes Arctic Ocean SST data from 65N to 90N and, for round numbers, from 40E to 40W. This is a good portion of the Arctic Ocean. Of course, the impact would be seasonal and would depend on the seasonal drop in sea ice extent or cover. The sea ice extent or cover has to decrease annually in order for sea surface temperature to be measured.”
But you are right that the SST data in Figures 5 and 7 are biased by the seasonal sea ice cover. However, as I showed in my earlier reply to you, we can approximate the impact by weighting the SST anomaly data based on the percentage of sea ice cover for those coordinates. Here’s the link again:
http://i45.tinypic.com/19b095.jpg
The sea ice cover data I used in that graph was the same used by GISS; that is, the Reynolds (OI.v2) Sea Ice Concentration dataset. And here are the percentages of Sea Ice Cover for the coordinates used in the weighted SST approximation above (65N-90N, 40E-40W):
http://i48.tinypic.com/a9lt8j.jpg
And regarding your statement, “this is not correct…”
Actually it is correct, as I showed above in the weighted SST comparison. The GISS LST anomalies have greater year-to-year variations and have a higher linear trend.
And I did research this topic in more depth than what was presented in the post. For example, the Barents Sea for the coordinates of 70N-78N, 40E-55E is fundamentally ice free in September from 1982 to 2009. (September 1989 had a sea ice cover of 0.18%) Here’s a .gif animation showing the September Sea Ice Cover from 1982 to 2009:
http://i45.tinypic.com/2wfs114.jpg
And here’s a comparison graph of the September Barents Sea temperature anomalies for the Reynolds OI.v2 SST for the coordinates of 70N-78N, 40E-55E and GISTEMP LST 1200km data that’s been extended out over. In this example, GISS extends LST anomalies with a positive trend over an SST dataset with a negative trend.
http://i46.tinypic.com/es3if4.jpg
You suggested that I “should replace every gistemp grid pixel with reynolds sst every time the sea is ice free and then you can say if this makes any difference or not.” Feel free to do so if you think that method would provide a more accurate approximation than the SST data that’s been weighted by sea ice cover used in this comparison graph:
http://i45.tinypic.com/19b095.jpg
Regards
Wow how many thousands of words of speculation and nobody bothered to RTFM? If I may be bold and quote from the p.5:
We use ocean temperature change only in regions that are ice-free all year (a map of this area is included in Supplementary Material), because our data set is intended to be temperature change of surface air. Surface air temperature (SAT), measured at heights of 1.25-2 meters at meteorological stations, is of most practical significance to humans and it is usually SAT change that is reported in climate model studies. Change of sea surface temperature (SST) is a good approximation to change of SAT in ice-free ocean areas; climate model simulations [Hansen et al., 2007] suggest that long-term SAT change over ice-free ocean is only slightly larger than SST change. However, ocean water temperature does not go below the freezing point of water, while
surface air temperature over sea ice can be much colder. As a result SST change underestimates SAT change when sea ice cover changes. Indeed, most climate models find that the largest SAT changes with global warming occur in regions of sea ice [IPCC, 2007]. Thus we estimate SAT changes in sea ice regions by extrapolating actual SAT measurements on nearby land or islands; if there are no stations within 1200 km we leave the temperature change undefined.