How reliable are Coral Sea SST (Sea Surface Temperature) data ?

There is a discussion at the Jennifer Marohasy blog, “Sea-surface Temperatures along the Great Barrier Reef” Posted by John McLean, January 5th, 2009, re recent research from AIM in Townsville that global warming is harming the Great Barrier Reef (GBR). The AIM scientists use the UK Hadley Centre SST data to show global warming is affecting coral.
This got me to take another look at a post of mine on Willis Island in the Coral Sea, a site that refuses to warm.
[Note here how politicians were running with the pro-warming conclusions, long before the published paper by De’ath et al is available.]
At the time I compiled Hadley and Reynolds SST data along with lower troposphere satellite temperature trends for the grid cell 15 to 20 degrees South and 145 to 155 East, which neatly has Willis Is. fairly central and extends west to the GBR coast.
This graphic shows that the Hadley SST data warms by ~0.75 degrees C while Willis Island land data actually cools slightly.
Willis Is composite trends
Now a good photo of the Willis Is. weather instruments can be found on the “Australia’s Reference Climate Station Network” web page.

www.bom.gov.au/climate/change/reference.shtml#rcsmap

by clicking on Willis Island on their map.
[Let me know where the actual BoM RCS temperature data can be downloaded from please]
Willis Is weather station
Now take in this idyllic scene and ask yourself, could the sea surface temperatures warm without warming air above them, which must then be reflected by the thermometers inside the Stevenson Screen which looks to be only 100-200 metres away ? That is what the Hadley Centre and the US based NOAA/NCDC are asking us to believe with their over-adjusted SST data. That the sea can warm without affecting air so close over the island. Note both the satellite data and Willis Is. cool slightly 1979-2005 while the Hadley SST’s warm. For the period 1982-2005 both the satellite data and Willis Is. cool slightly while both sets of SST’s warm.
Note: I downloaded the SST and satellite data from the KNMI website www.climexp.knmi.nl/

Now to a related subject, I recently wrote on the deterioration over the last ~10 years in Australian rainfall data.
I notice that the BoM station Willis Island 200283 is missing many months of temperature data in recent years. You have to ask how seriously the BoM view their task to collect these “world heritage” data when they perform as badly as this. Monthly mean T data from GISS.
Willis Island missing monthly T data
I think the job needs to be taken away from them or the BoM split in some way so our vital weather data can be properly collected, sufficiently funded and properly archived then disseminated freely. Australia deserves better than the dogs breakfast we now have.
In view of the evangelical certainly of the Rudd Govt and its proponents like Garnuat when they talk about global warming, hands up those who are surprised that key remote area weather records such as this have as recently as 2006-2007 been so degraded. There is no near neighbour for Willis Island but lets hope the BoM can recover these missing month data somehow.
Finally, for those who are not familiar with SST data. These temperature reading have been taken for centuries now from ships at sea, by varying methods, earlier buckets thrown over the side then more recently by instruments at engine water intakes. The data is so variable that large adjustments are required to filter out usable data series, adjustments far larger than the IPCC compliant trend itself. Here is a comment I made yesterday at the Jennifer Marohasy blog with links to some of my earlier articles.
I see some say that Lough and or De’ath used the HadSST2 data whereas John McLean has graphed the Reynolds data which I understand incorporates some satellite input. Re SST data in general, there have been so many large adjustments/corrections mainly around WWII to get SST trends to fit passably with IPCC GW that any reliance on century long and decadal trends to fit this or that theory could be unwise. And of course years ago CRU adjusted coastal SST warmer to fit land data.
The next page illustrates the huge adjustments needed to make SST fit land trends and another point is that the older the paper dealing with SST, the less warming trend and the more prominent is warmth in the late 19C.
I did some comparisons of Reynolds and Hadley SST data around New Zealand
Page pointing out large variations in temperature trends for NZ region.
Graphic showing differences between Reynolds and Hadley V3 data
Graphic showing differences between Reynolds and Hadley V2 data
Blog article April 08 pointing out huge global discontinuity between Hadley and Reynolds late 1990’s

Clearly all these datasets are in a state of flux despite all the pressures on the various teams to please the IPCC and agree closely; any policy-makers who think the science is settled need counselling.
I would like to have the computing facilities to quickly interrogate grid cell differences in these datasets.

3 thoughts on “How reliable are Coral Sea SST (Sea Surface Temperature) data ?”

  1. Thanks so much for putting this together. It’s fascinating.
    Regarding the missing values (shown as 999.9) – the problem has appeared post 1992 and it seems was particularly bad during 2006. Given all the extra funding for ‘climate change’ how can it be that basic data collection has slipped so badly?

  2. Warwick: Your table shows Willis Island data from GISS, which gets its data from the GHCN. Is the data recorded by the BoM and not retrieved by the GHCN? Or is it recorded by BoM, retrieved by the GHCN, but not retrieved by GISS? In other words, is the problem with BoM or with the other links in the chain?

  3. Bob, I have BoM daily data purchased in April 2007 and the gap Jul-Sep 2004 is there, then my BoM dailies have 17 days in Oct 2004, then a huge gap to 23 Oct 2006, then pretty good data to end April 2007.
    So assuming that the station was not manned and gaps are due to instrument malfunctions, post Apr-07 the BoM may have recovered data for GHCN/GISS for the period Nov-Dec 2004, then through 2005 plus Jan and Mar 2006. Why GISS has gaps Jan-Feb-Mar 2007, is a puzzle.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>