It is odd that I should be blogging about Climategate and my hometown's climate recently. I think I have found another connection.
I was looking at an article The Smoking Gun at Darwin Zero which has a lot of charts of the temperature in Darwin that don't seem to make sense. The article is by Willis Eschenbach who has lead the FOIA challenge to get CRU's data. It was picked up by several media here too.
To give some background, here are some graphics I constructed from the Australian BOM websites handy graph pages. This is their 'Australian high-quality climate site data' which means it has been adjusted for various factors: there is a little page explaining the kinds of things they do to get rid of extraneous factors, with things such as:
A change in the type of thermometer shelter used at many Australian observation sites in the early 20th century resulted in a sudden drop in recorded temperatures which is entirely spurious. It is for this reason that these early data are currently not used for monitoring climate change.
It is this earlier data that Eschenbach includes, and to some extent relies on, in order then gets his very different trend lines. But for a different source of information, in the name of sanity checking perhaps, I thought I'd give readers some other information on Darwin, to complement the Eschenbach article.
First up, here a graph of the maximum temperatures from the Australia site for Darwin: I don't think there is much of interest there.
So, as a layman, it seems credible from these that Darwin is indeed a few degrees hotter in the most recent fifty years compared to the fifty before then. Not hotter in the sense that the maximums are increasing, but hotter in the sense that it does not cool down as much. (Is that the way those numbers work: reader's advice welcome.) I think it is useful to have this in mind when looking at Eschenbach's article.
But the thing that surprised me was in Eschenbach's article, which I don't really understand much (not being a climate scientist in any way), in the first diagram he generates.
He tries to prove something by taking three stations in North Australia and these stations, according to the bottom left of the graphic, are Darwin, Alice Springs, and Yamba. Yamba? But Yamba is a little town near my hometown, famous for its prawns, surely he has not used that?
So I have put the following little note as a comment on their website, and I will let readers know what my mistake was or what Eschenbach says.
I am intrigued that you include YAMBA in your readings, in the second diagram, and call it a Northern Station. It would be surprising if it fits the bill for a neighbouring station. Is that what the CRU people did, or just what you did?
YAMBA lat -29,43 long 153.35 is about 3,000km from Darwin. It sits on a different ocean (Pacific) next to my home town, and has very equitable climate. It is the YAMBA you get when dialing up on the AIS site.
3000km, to give some context, is a little more than the distance between London England and Ankara Turkey. More than the distance from Detroit Michigan and Kingston Jamaica.
Now maybe there is another Yamba (there is another town near Adelaide on the opposite side of the continent in South Australia) or perhaps there is a station called Yamba (I could not find such a thing.) But if not, averaging data from a monsoonal tropical coastal town near the Timor and Arafura Seas, a town in the middle of one of the driest deserts, and a fishing village 3,000km away on the Pacific with a famously equitable climate (or perhaps the South Australian grape-growing town) and then saying it looks nothing like the IPCC graph for Darwin leaves me a little baffled about what point is being made.
Eschenbach first responded
I included everything bounded by the UN IPCC box shown as Fig. 1., from 110E to 155E, and from 30S to 11S. Yamba is within that box.
So let me get this right. The HadCRUT3 paper shows hundreds of stations that it says are used, which correspond to the Aust BOM stations. See figure 1 http://hadobs.metoffice.com/hadcrut3/HadCRUT3_accepted.pdf
But you quote Professor Karlen that NASA only has three stations. You pick three stations, one monsoonal (Darwin), one desert (Alice Springs), one temperate coastal (Yamba), and add them, and then you get surprised that the result does not look like anything the the IPCC graphic? What is the point of that?
Then you do all sorts of elaborate reverse engineerings, to discover that there has been some kind of a data adjustment, when the owners of the data (the Aust. BOM -- I don't think NASA had any stations in Australia in the early 1900s!) warn in their page on the Australian station figures that the early numbers are unreliable without an adjustment.
It seems to me that your figure 4 is the only one of much interest. Where is this smoking gun?
And then, finally, after several Australians made comments about how bogglingly stupid it would be to call Yamba a neighbouring station (some were worried that it was what the BoM or CRU had done, not Eschenbach):
Willis Eschenbach (04:01:23) :
Mike (19:22:41) :
If you want to know how mad it would be to adjust Darwin using Yamba data check out google maps Please tell me that Yamba was not used to adjust Darwin! please, please!
OK. Yamba was not used to adjust Darwin. It's just in the area indicated in Fig. 1, is all, so it's used in that average.
Good. (And even if I think Escehnbach's sensationalizing is regrettable, I do commend his politeness there.)
Following through links, there is a good posting from an Australian scientist whose work actually is temperature ajustments. And the example he gives is just down the road from my home town (Coffs Harbour) again: Port Macquarie.
Here it is
Developing techniques for temperature data adjustment is a large part of my scientific existence, so I'll give a simple example of how it is done.
At Port Macquarie, the observing site moved from a location in town, a kilometre or so south of the town centre and a few hundred metres from the ocean, to the airport which is several kilometres inland. The two sites operated in parallel from 1995 (when the airport site opened) to 2003 (when the town site closed).
During that period of parallel observations, on average maximum temperatures at the airport were 0.6 degrees warmer than those in town, and minimum temperatures 1.5 degrees cooler. It follows that, if you wanted to make the pre-1995 town observations consistent with the post-1995 airport one, at the most basic level you could adjust the town maxima up by 0.6 degrees, and the minima down by 1.5 degrees.
If you wanted to go to a more sophisticated level you would calculate adjustments separately for each month; as it happens the maximum temperature difference between the two sites is largest in summer and near zero in winter (because the sea breezes that cool the coast on summer days aren't much of a factor in winter), so to take the seasonal variation into account you would need to make a larger adjustment in summer than winter.
At a more advanced level again, in some places the effect a site move has is greater (or less) on hot days than it is on cool days (on a hot day, the coastal strip can be 10 or more degrees cooler than a few kilometres inland, whereas on a cold, wet day the difference is often minimal), so different adjustments may be needed for hot days and cold days. I'm currently working on developing a data set which implements this technique across the Australian network. (We aren't always as fortunate to have the amount of overlapping data that we do in the Port Macquarie case, so some of the adjustments have a fairly high level of uncertainty).
There seems to be some surprise that changes in location, even of a few miles, could result in several degrees of measurable difference. I think anyone who has lived in coastal Australia know how much an effect there can be.
For example, as I write this in Sydney, the Observatory Hill readings (I can see it from my window) are 24 degrees (3.00pm 2009-12-12) C [75.2 degrees F]. Canterbury which is 15 minutes drive away, is 26.3 degrees C [79.3 degrees F]. Olympic Park, which I can see from my window about 12km away is 28.1 degrees C. Liverpool on the edge of town is 30 degrees C. Camden is 32.6 degrees [90.7 degrees F]. That is 6.6 degrees C [15.4 degrees F] difference within the same plane! The practical result: houses near the coast are more in-demand than those inland, and people flock to the beach in Summer for the cool.