Queensland in Jones et al 1990

A couple of interesting references and comments on the Cardwell site were sine int. Let’s focus on Cardwell and its neighbours a little.

Cardwell is on the coast of Queensland. Carl Smith sent in this link showing a small town on a coastal highway – see also his comment here. In Jones et al 1990, there are 4 Queensland coastal sites (BOM number shown): Cooktown 31016, Innisfail 32025, Cardwell 32004, Bowen 33007. with Charters 34002 directly S and Georgetown 30018 to the W, as shown in the map below.


Innisfail is very close to Cardwell and should be a very good comparandum. Two of the sites (Bowen, Cooktown) were not in the Torok network; all 4 sites could be located in GHCN. I’ve plotted versions for all 4 sites together with the HadCRU3 gridcell. One of the first striking things is that none of these sites has been updated in GHCN since the early 1990s – what do they do over at GHCN? Jones says that his sites are at GHCN so I presume that Jones hasn’t updated these sites either (but Jones isn’t telling). In the original CRU edition (for which stations can be identified from information at CDIAC), there are 2 stations in this gridcell: Cairns and Townsville – both classic Jones-type sites.

Here’s a plot of the 4 sites plus HadCRU3. Obviously there are major differences between the site histories. It is obviously inconceivable that the differences between Innisfail and Cardwell are due to differing “regional” climate – the usual Team excuse for inconsistency. So one(or both of them) is contaminated in some way by non-climatic effects. Which one? I don’t see how you can make a decision merely by looking at the plots. How can one find firm ground in this type of morass? And these examples are coming from Australia, where there is presumably pretty good data.


Dale has plotted up max-mins from the Jones versions for 3 of these sites). The Torok versions should also be examined for a complete story – here are Dale’s 3 graphics, which distinguish max and min records for these 3 sites. THe change in diurnal temperature range (DTR) for these 3 sites in the Jones data is very striking – being up to 4 deg C. There’s quite a bit of discussion of DTR in AR4, which says that there was subtantial change in DTR up to 1979, but little since then (on a global basis) and which discusses some studies attributing DTR changes to increased cloudiness. In any event, one would expect to see a consistent pattern of climate-induced DTR in these closely related sites: can you see any?


  1. Posted Apr 29, 2007 at 10:39 AM | Permalink

    It is obviously inconceivable that the differences between Innisfail and Cardwell are due to differing “regional” climate

    It is a very variable area weather wise with a huge rainfall gradient. See this rainfall map, where Innisfail has over 3 meters and Cardwell only 1 meter for 2005. And the wettest town in Australia, Tully is between them with an average of 4 meters and a record annual rainfall of 5 meters.

    Click to access aws2005_map2.pdf

  2. Steve McIntyre
    Posted Apr 29, 2007 at 10:50 AM | Permalink

    Interesting gradient. But Cardwell and Cooktown have similar sorts of precipitation. David, do you think that these sorts of differences could be climatic as opposed to nonclimatic? In any event, shouldn’t these folks spend a little time showing the validity of taking an average (one of the underlying problems in this entire area is that this is more suited to technical reports than little articles in academic climate journals.)

  3. Posted Apr 29, 2007 at 11:19 AM | Permalink

    There are just so many issues. The changing land-use from rainforest to dairy farming to cane farms, all changing with economics of the time. Changing landuse on the tablelands around the area as orographic effects are very important there.

    Another non-human cause could be the shifting boundary of climate regimes. With such a large gradient, if the sub-tropical dry region shifts slightly north I imagine it could affect Cardwell but not Innisfail which is probably dominated by the orographic effects of the ranges.

    I would also be concerned about the range of skill and training of the recorders of these instruments.

  4. Posted Apr 29, 2007 at 11:26 AM | Permalink

    For example, just an hypothesis, if the sub-tropical climate were to advance north, resulting in a decrease in afternoon thunderstorms that typically reduce the maximum daily temperatures, it could explain Cardwell increasing in annual temperatures while Innisfail showing no trend. You could explain such an advance by regional forest clearing and conversion to grassland and eucalypt.

  5. Al
    Posted Apr 29, 2007 at 12:19 PM | Permalink

    Are there any satellites with a sufficiently wide focus angle and frequency sensitivity to measure ‘outgoing heat flux of planet Earth’ _directly_?

    Because the more time spent wandering through piles of data from weather stations show only a couple of things conclusively.

    1) Cherry-picking by the hockey team. (By picking start dates, end dates, included stations, excluded stations, applied adjustments etc.)
    2) Complete chaos in methodology.

    Even an extremely rigorous reevaluation of _all_ of the weather station data is unlikely to be sufficient to fully rebut the hockey team at this point. And there’s no way to go back a hundred years with concrete ideas on how to monitor local temperature systematically.

  6. Posted Apr 29, 2007 at 1:12 PM | Permalink

    Pertinent to the revealing of disparate trends in North Queensland temperature histories is the flat trend 1939-2006 from Willis Island (16.3 S 150.0 E ) in the Coral Sea. I have a graphic here on a blog post last year which was in response to a Govt proposal to protect the Great Barrier Reef with shadecloth.
    Willis Island mean T trend
    If you use GISS to find the data, you see that 2006 was a bit cooler than 2005.

  7. Peter Hartley
    Posted Apr 29, 2007 at 2:02 PM | Permalink

    Following up on Steve’s comment about how ridiculous it is that many of these Queensland records have not been kept up to date, if one goes to the GISS site linked by Warwick one finds that the only stations in NE Queensland with records complete up to 2006 are

    Mackay, Rockhampton, Townsville, Gladstone, Cairns


    Willis Island.

    This leads one to ask — how did they let Willis Island slip into the up-to-date data set? It spoils the Team story line! Fire the person responsible for that mistake! The rest of these places are all rapidly developing regional cities with nicely growing UHI effects that can be passed off as “global warming”, no sweat.

  8. Posted Apr 29, 2007 at 10:37 PM | Permalink


    data is unlikely to be sufficient to fully rebut the hockey team at this point.

    Just bring the data for the rural stations up to date and recalculate HADCRU including them up to the present. If the result shifts temperatures since 1990 downward then the result is biased by UHI.

%d bloggers like this: