Sunday
Nov212010
by 
Bishop Hill
Bishop Hill Climate models hopelessly simplistic
Nov 21, 2010 Climate: Models Climate: Sceptics P Gosselin has an interesting story about an Austrian meteorologist who is completely underwhelmed by the reliability of climate models. As Karsten Brandt apparently puts it:
It is simply nonsense. These prognoses are not worth the paper they’re printed on. The Gulf Stream has an impact on European weather that is 100 times larger than CO2.”
Reader Comments (62)
ZT
The advanced course text is:
10 PRINT "GLOBAL WARMING IS CAUSED BY CO2"
20 GOTO 10
OMG!!!! They are using a GOTO statement!!!!!! How could they!!!
I would flunk anyone using a GOTO statement. Utterly terrible coding!.
Of course, I suspect ZT was making a wry joke. ☺
BBD and Phillip - I just had a quick look at UKCIP (Phillip, thanks for the ref.) and it does make seasonal gridded predictions/projections. The UKCIP02 data is apparently available to researchers but, if I've understood correctly, the predictions/projections do not start until next year. This model has been superceded by UKCIP09 which covers the same time period so it looks as if next winter would be the one to watch. Having said that perhaps one could start tracking from Jan11 - I'll try and clarify, it would be interesting to know how 2002 output performs compared to what the weather/climate actually does and it looks as if UKCIP02 data will be removed from the mainstream site before the end of the year:
http://www.ukcip.org.uk/index.php?option=com_content&task=view&id=161&Itemid=287
19 Nov: NATO: Official Text: Active Engagement, Modern Defence
Strategic Concept for the Defence and Security of The Members of the North Atlantic Treaty Organisation adopted by Heads of State and Government in Lisbon
15.Key environmental and resource constraints, including health risks, climate change, water scarcity and increasing energy needs will further shape the future security environment in areas of concern to NATO and have the potential to significantly affect NATO planning and operations...
http://www.nato.int/cps/en/natolive/official_texts_68580.htm
Climate models are completely reliable, - it's just that they have the sign wrong... :)
http://www.tititudorancea.com/z/volcanic_eruptions_affect_rainfall_over_asian_monsoon_region.htm
"Tree-ring researchers at Columbia University's Lamont-Doherty Earth Observatory (LDEO) showed that big eruptions tend to dry up much of Central Asia, but bring more rain to southeast Asian countries including Vietnam, Laos, Cambodia, Thailand and Myanmar--the opposite of what many climate models predict."
Fearing the august world of climatology might be unfairly maligned by the 'GOTO' comments here I conducted the following short check, using the whistle blower supplied code: yamal/reform.for.
$ wc ./yamal/reform.for -l
244 ./yamal/reform.for (i.e. the program is 244 lines long)
$ grep -i goto ./yamal/reform.for | wc -l
11 (i.e. that same program contains 11 GOTO statements)
I conclude that this particular program is 4.5% GOTO statement. For those unaware, GOTO statements are unconditional transfers of program flow, a programing construction noted for producing hard to maintain software).
I wonder if Julia Slingo would care to let us know exactly how many 'GOTO' statements are in the MET office code?
Where do you fit this in then?
Buffy the Vampire Slayer Studies
8:06 PM | Don Pablo de la Sierra
Nothing particularly wrong with a few Gotos in the code if it is clear what they are doing. Definitely had a few scattered in the Reading SGCM (Simplified Model) that the majority of the Phd students used for their studies. I don't remember there being as many comments as shown here though, and there aren't many in this version...
http://www.met.reading.ac.uk/~mike/dyn_models/igcm/
Good Lord, Shub. Not a hint of tongue in cheek.
I suggest that any course with the suffix "studies", and for that matter the suffix "science" is not one I want my taxes to subsidise.
@ not banned
Sorry - been off dealing with the day job. Thanks for the link - interesting.
@ not banned
As you say - the UKCIP02 model is being retired and UKCP09 is replacing it. Data for both projections only from 2011 on. Rather broad strokes - eg winter 2011 - 2040 +0.5K over 1961-90 climatology... +1.5k 2041 - 70 (Midlands and South) etc.
Standard stuff really. It's always very difficult to pin anything down. And you will always be told to look at the trend - don't mistake weather for climate etc.
Like fighting smoke really.
Rob B
Back in the mid 1970's I worked for DEC at the Bell Laboratories in New Jersey, and so knew both Thompson and Ritchie. I learned C from Ritchie and Unix from Thompson. And there is a GOTO in C for one reason alone, which I cannot remember, but Thompson was of the opinion that it was not a valid reason, claiming anyone using it was demonstrating sloppy thinking and probably did not think about what they were coding adequately.
I agree with Thompson.
And I think the totally sloppy code I saw in Harry-Read-Me demonstrates that. True the code Harry hacked up is in "Fortran", but Fortran has changed over the years into something akin to C, so I hold that the use of GOTO is a sign of sloppy thinking. I have yet to see a chunk of code with a GOTO in it where it was truly necessary. And I have seen far too many examples of "spaghetti" code stitched together with "GOTO" in a manner to make it incomprehensible.
Just my opinion, but then again I spent a couple years as a programming consultant at the Cornell computing center explaining to academics pretending to be programmers why their programs didn't work.
As for these wonderful modeling programs that graduate students wrote and are so perfect that they correctly predict the climate 100 years from now, I can only double up in laughter.
Interesting: A cursory glance at the code that Rob B.mentioned:
shows GOTO code like this:
IF (ABS(DAY-BEGDAY) .GT. 1.0E-02) THEN
IF (ABS(DAY-BEGDAY) .LT. ABS(DAYNEAR-BEGDAY)) THEN
DAYNEAR = DAY
ENDIF
GOTO 180
ELSE
GOTO 200
ENDIF
...and plenty of code cloning in e.g. subroutines HANAL, HANAL1, and HANALV.
It is, as they say, worse than we thought.
see http://www.met.rdg.ac.uk/~mike/dyn_models/igcm/igcm1_src