What to make of THIS bizarre anomaly map?
UPDATE: 2 Feb 2009:
Well, if you can’t fix it, at least you can hide it. If you now try making an anomaly map of baseline vs. self you now get this message:
Surface Temperature Analysis: Maps
Anomalies for any period with respect to itself is 0 by definition – no map needed
Guess it is easier to sweep dirt under the rug rather than actually clean up the code…
What Have I Done?
I was exploring another example of The Bolivia Effect where an empty area became quite “hot” when the data were missing (Panama, posting soon) and that led to another couple of changed baselines that led to more ‘interesting red’ (1980 vs 1951-1980 baseline). I’m doing these examinations with a 250 km ‘spread’ as that tells me more about where the thermometers are located. The above graph, if done instead with a 1200 km spread or smoothing, has the white spread out to sea 1200 km with smaller infinite red blobs in the middles of the oceans.
I thought it would be ‘interesting’ to step through parts of the baseline bit by bit to find out where it was “hot” and “cold”. (Thinking of breaking it into decades…. still to be tried…) When I thought:
Well, you always need a baseline benchmark, even if you are ‘benchmarking the baseline’, so why not start with the “NULL” case of baseline equal to report period? It ought to be a simple all white land area with grey oceans for missing data.
Well, I was “A bit surprised” when I got a blood red ocean everywhere on the planet.
You can try it yourself at the NASA / GISS web site map making page.
In all fairness, the land does stay white (no anomaly against itself) and that’s a very good thing. But that Ocean!
ALL the ocean area with no data goes blood red and the scale shows it to be up to ‘9999’ degrees C of anomaly.
“Houston, I think you have a problem”…
Why Don’t I Look In The Code
Well, the code NASA GISS publishes and says is what they run, is not this code that they are running.
Yes, they are not publishing the real code. In the real code running on the GISS web page to make these anomaly maps, you can change the baseline and you can change the “spread” of each cell. (Thus the web page that lets you make these “what if” anomaly maps). In the code they publish, the “reach” of that spread is hard coded at 1200 km and the baseline period is hard coded at 1951-1980.
So I simply can not do any debugging on this issue, because the code that produces these maps is not available.
But what I can say is pretty simple:
If a map with no areas of unusual warmth (by definition with the baseline = report period) has this happen; something is wrong.
I’d further speculate that that something could easily be what causes The Bolivia Effect where areas that are lacking in current data get rosy red blobs. Just done on a spectacular scale.
Further, I’d speculate that this might go a long way toward explaining the perpetual bright red in the Arctic (where there are no thermometers so no thermometer data). This “anomaly map” includes the HadCRUT SST anomaly map for ocean temperatures. The striking thing about this one is that those two bands of red at each pole sure look a lot like the ‘persistent polar warming’ we’ve been told to be so worried about. One can only wonder if there is some “bleed through” of these hypothetical warm spots when the ‘null data’ cells are averaged in with the ‘real data cells’ when making non-edge case maps. But without the code, it can only be a wonder:
The default 1200 km present date map for comparison:
I’m surprised nobody ever tried this particular ‘limit case’ before. Then again, experienced software developers know to test the ‘limit cases’ even if they do seem bizarre, since that’s where the most bugs live. And this sure looks like a bug to me.
A very hot bug…
UPDATE: 1 Feb 2009: Added non-zero biased maps
Over on WUWT, where this thread has been picked up, there was a discussion in comments where the question was raised “does this only show up in NULL maps?”
The assertion was made that anything beyond about 5C was very unlikely to be a valid anomaly. So I’ve made a couple of more maps that I think show these effects bleeding into the non-NULL cases. These are both December 2009 (the default) vs 1998 baseline (a selected value) anomaly maps. One has a 250 km smoothing, the other 1200 km. I think these may show that “this bug has legs”:
So maybe someone up in Alaska can tell us if this year, after a 12 years of cooling from the 1998 peak, is really 12 C hotter than then…
And the more smoothed so muted 1200 km map:
Where the temperature range is reduced a bit, but the coverage is expanded greatly. Still, an 8 C ‘anomaly’ ought to have been noticed…
In looking at the 1998 ‘warmest year’ map it does look like it has a cool Arctic, so who knows. Maybe the ‘warmest year’ wasn’t as hot as the ‘warmers’ were claiming?
In thinking about the likely nature of this bug, one idea that came to mind is that it might be in the ‘display’ part of the code. Perhaps when there are no ‘anomalies’ to display, the graph drawing code does not bother to map some 9999 missing data flags to the normal default of ‘grey’? It’s all speculation. Well, other than the fact that the map as drawn full of red oceans at 9999 C is clearly a bug.
head over to the airvent for a discussion of the anomaly method and removing stations
REPLY: [ You mean like at http://noconsensus.wordpress.com/ -E.M.Smith ]
Yes, I was poking around yesterday and trying some odd cases like check the anomaly of one year against itself and I got a lot of brown. But it didn’t occur to me what a big wrong thing that was. Duh…
And it is not just the oceans that are brown-red; what are all of those dang brown dots doing in the middle of Asia and Amazonia? (250KM ‘reach’ parameter).
I’m referring to 2008 vs itself, Jan-Dec Annuals.
Poor old Madagascar, indeed also a substantial part of Africa seem to be quite hot;
9999F would seem quite incandescant…
Unfortunately, the 250KM reach param is only usable on GISS ‘data’, not for the GHCN historic data. Would be nice to see how noisey everything is before the big wet blanket of spatial ‘mohoginization’ with such a long reach as 1200 KM).
The image linked at my name above says it all…
One other clue; it seems that the ‘historic’ data a.k.a ‘unadjusted’ can only be mapped with the 1200KM spread parameter. Choosing 250KM elicits this message;
“Internal Server Error
The server encountered an internal error or misconfiguration and was unable to complete your request.
Please contact the server administrator, firstname.lastname@example.org and inform them of the time the error occurred, and anything you might have done that may have caused the error.
More information about this error may be available in the server error log.
Apache/2.2.11 (Unix) Server at data.giss.nasa.gov Port 80″
entering 1997 for all years, choose jan-dec, 250KM overreach, ‘unadjusted’, turn on the oceans;
and it comes back with an empty file instead of the ‘server error’ above.
THese commments are all with reference to Cheif’s link above about the Nasa Mapper.
Not useful but interesting nonetheless. Quirky quirky quirky…
“I got a blood red ocean everywhere on the planet.” Discovery #1 – something’s rotten in the state of GISS.
“the code NASA GISS publishes and says is what they run, is not this code that they are running.”: Discovery #2 – something else is rotten in the state of GISS.
I see how your discovery #1 may explain Arctic Ocean ‘warming’ but I don’t see how it could relate to ‘warming’ on the Antarctic continent. Any thoughts?
REPLY: [ There are a lot of “short records” and major gaps in the Antarctic record. LOTS of missing data, so lots of opportunities for a “missing data effect” to have an impact. BTW, I’ve seen one reference that says the code does let you choose RCRIT to be 256 km, and the code I have does not, so I am wondering if their 15 Nov 2009 ‘update’ included variable range (or 2 step range). So much to check, so little time ;-) -E.M.Smith ]
Pingback: Is The NULL Default Infinite Hot? « The Excellence In Weather Network
I guess you didn’t do a Polar view map yet. I did them and I’ll put the links up to the maps and trust me after you see the South Pole view with 1200km you will be ROTFL.
First here is with SST data at 250km:
Here is without SST data at 250km:
Now let the fun really begin
Here is with SST data at 1200km:
Finally here is without SST data at 1200km:
REPLY: OMG! That “let the fun really begin” is Just Astounding! I’m rushed right now, but those will go in the main article when I get a chance. Until then: Anyone reading this article = click those links!. (I’m going to have to get in the habit of doing polar views… ) -E.M. Smith ]
Its even more interesting when I go in the recent times and put there the interval and reference 2009-2009 and the Had/Reyn source for oceans – then I see how much of the land and sea returns the error – all around antarctic, almost whole arctic, whole middle of Africa and South America, big chunks of Canada and Greenland etc. – and in comparison with the interval and reference 1951-1980 this bug perhaps can be a hint, how the real GISS coverage shrinked recently.
-Looks like the area covered by GISS is shrinking even much faster than the arctic sea ice… ;)
Here I made some pictures:
REPLY: [ Very Interesting! And something I’d not thought of. You do get to see how moth eaten the temperatures have become lately (and how the ‘missing data bug’ makes places “hot hot hot”… -E.M.Smith ]
The all-red-oceans map bug happens in any case where the Baseline and Time Interval match and the Smoothing is set to 250km. If any one of the four numbers is different by even a single year, it seems to work correctly. I smell a divide-by-zero error that is not being properly trapped.
It’s always been a puzzle to me why the heatingist parts are those with fewest thermometers
I readily agree that the hot red oceans (with no data source) is an issue.
But I think it is more interesting to turn on a data source for the oceans, and plot the null anomaly (base period year(s) = ‘Time Interval’ year(s))
Then you can easily see the hot red dots in Asia, Amazonia, and a Juge blob in Africa (2009 in all year entries, Jan-Dec period).
Why is Northern Canada so hot?
Watts Up With That?
REPLY: [ And on this “quality” rests the fate of the worlds economy… -E.M.Smith ]
Pingback: Chiefio asks: why does GISS make us see red? « Watts Up With That?
I sent a pdf to the Pub.
Anytime you want a blinking gif, just let me know.
I did it by importing each image into Photoshop,, make each one a layer,align them, pop over into ImageMaker, and they have a way to make a multilayer image into a blinkie.
Also, the postscript ps version seems to have thicker black lines than the PDF’s, for the continental outlines.
Basic testing would be test the boundaries from mini (negatives) to maximum and one on each side if possible.
Only a moron of a developer would make a null value default to the maximum value.
«In thinking about the likely nature of this bug, one idea that came to mind is that it might be in the ‘display’ part of the code. Perhaps when there are no ‘anomalies’ to display, the graph drawing code does not bother to map some 9999 missing data flags to the normal default of ‘grey’? It’s all speculation. Well, other than the fact that the map as drawn full of red oceans at 9999 C is clearly a bug.»
Hi Michael, it’s seems to me that 9999 is not the null value. I think it is more likely this bug to be prior to the plotting routine – which is correctly displaying grey spots when passed the null flag. Somewhere prior to that some code is ignoring the absence of data flag and using it in some calculation.
Another thing that has been making curious, do you know if the 250 Km and 1200 Km figures refer to the cartographic plane or to the geodetic surface? The projection they use, which I’m not sure which is, can’t be used to measure distances, the distortions are just to large for that.
REPLY: [ Any evaluation of the 9999 has to be speculative until you have the code to inspect. But the “grey” is supposed to be ‘no data’ and 9999 is used as a missing data flag in several places in GIStemp, so there is the possibility that the graphing package has a bug where it looks to draw anomalies and ‘when it goes to draw the first one’ draws the grey, but with no ‘first one’ it never does the grey step. Speculative, yes, but worth checking (if you had access to that code)… From this distance, it’s just not possible to say for certain if it is that, or something bigger. Oh, and they measure real distances, not ‘as on the graph’. You can see this by looking at the top of Canada. That one lone station in Eureka has a very wide rectangle. That’s 250 km high and wide, but prints as a wide stripe. The boxes at the equator show as nearly square. -E.M.Smith }
Pingback: Is something rotten in Alaska? – Moth Eyes
I had a look at the raw temperature data for Alaska in Dec 1998 and 2009: it turns out that, yes, Dec 2009 was, in fact, warmer than 1998. No, the bug doesn’t “have legs”.
REPLY: [ Thanks for the information. It’s shaping up to look more like the graphical routines only turn the missing data flag of 9999 into ‘gray’ when there is at least one non-zero anomaly to process; AND it looks like during the ‘peak hot year’ of 1998, the Arctic was particularly cold. Guess that “global warming” isn’t so Global after all… But in any case, a bug is a bug. The “proof” will be if they find code in the graphics step that, when changed, fixes this. But your information does raise some new interesting things to ponder. In particular, does the Arctic have a negative correlation with the rest of the world? If so, how long are the patterns? And if they are long and strong (such as in sync with the PDO) might the whole 1951-1980 default baseline be a ‘cherry pick’ that makes the Arctic ‘always hot’?… Ah well, those will be for another day. But in an amusing way, it leads me back to where I started this thread. Benchmarking the baseline. -E.M.Smith ]
Hi again Michael,
I went to check your last sentence, I don’t know where Eureka is, but I can see the wider rectangles at the poles.
This makes me even more curious, they seem to be using some sort of regular area grid. This is impossible to attain in a sphere’s surface, much less in an elipsoid’s. Not saying what they do is wrong, but definitly awkward.
Hope you don’t mind me asking if you know something more about this grid.
REPLY: [ Eureka is that lone rectangle at the very top of Canada a bit toward the East. The code makes boxes based on a non-uniform geography (i.e it’s based on LAT and LON of two edges as dividers between boxes) then fills it based on distance from the center point of the box to a station (250 or 1200 km). The display shows it as a rectangle / square. Answering questions between these three things, you must consult the code… It is listed under the right hand ‘categories’ box under GIStemp source code. -E.M.Smith ]
Posted here as the most likely place to get noticed by the folks who would have insight;
I stumbled onto another guy with an apparently unique approach to the question of thermometers;
I’d never heard of CLOJURE; but my lexicon was established in the olden daze, things like BASH, SNOBOL, and YACC…
Perhaps the value is in the different (?) data source he uses?
Anyway, yet another thing that I can make no informed decision about…
Are those datasets more raw than the other ‘raw’ ?
What is the correct word for unmanipulated data?
Pingback: Blood Red Oceans? « Things I Find Interesting
EM I did a little analysis on my site this morning about the this problem and found out that GISS has killed the maps and put up a little reminder notice:
They are apparently following this site, or Anthony’s, as none of the links work anymore. I think that it is notable that they took this down within three days of somebody reporting it, and make no mention that I could find about the fact that they changed it. Worse still, the new page insults you for even trying to make the map. I think you have more influence than you know, E. M.
All told, I hope that they use the same code for making maps on the website and making them for “official” use. That would definitely not be good for them when a possible gaping hole in their code is revealed.
Funny they took it down without explanation. Are they again so ashamed, they can’t say a word?
But anyway, we already have our maps of GISS non-coverage, so it fulfilled some purpose.
Now I wonder how they can project the anomalies to Arctica, central Africa, central South America and all around the Antartica – when they have almost no data for there…
Anybody can make a bug – even NASA apparently. But sometimes the bugs are revealing ones…
Hey, that’s climate “science”. They now post this message to hide the bug :
Anomalies for any period with respect to itself is 0 by definition – no map needed”
More weird things : if you display Jan anomalies for period 1951-1981 versus base 1951-1980, you’ll get a 1°C warming at some areas like Saudi Arabia.
It mathematically means that Jan 1981 has been 1°C x 30 = 30°C warmer in Saudi Arabia.
That’s crazy ! Who’s wrong here ???
( http://data.giss.nasa.gov/cgi-bin/gistemp/do_nmap.py?year_last=2009&month_last=12&sat=4&sst=1&type=anoms&mean_gen=01&year1=1951&year2=1981&base1=1951&base2=1980&radius=250&pol=reg )
And if you display Jan of 1951-1979 versus base period 1951-1980, you’ll also get a warming spot of 1°C at Saudi Arabia.
It means that in order for 1951-1980 to have an anomaly of 0°C at that spot (that’s the least we can expect), 1980 must be 1°C x 30 = 30°C cooler !
I just went to the NASA page, and used 1951-1980 for both ranges, and got the following message, no graph :
Anomalies for any period with respect to itself is 0 by definition – no map needed
“EM I did a little analysis on my site this morning about the this problem and found out that GISS has killed the maps and put up a little reminder notice:
Just another cover up is what I see.
Keep all records and send them a FOIA !
In the FOIA I would be very specific and I would include demand to show all communications (emails etc.) concerning the bug and the subsequent changes in software and of the GISS website. (It would be probably quite funny to see how they coped with it – because the link to the little notice is clearly just a patch, it most probably doesn’t cope with the bug in the software itself.)
But still – we now have the maps of the non-coverage by real measurements for 2009 anyway. The bug made it very simple for us to see where – even they now try to whitewash it and thus make the suspicion even more flagrant :)
Pingback: Weather not climate, remember? « Wolfville watch