Archive for November, 2014

Rain, clouds, and temperature

November 19, 2014

Looking at the continent of Australia as a whole, and using 12 month running means to smooth the very noisy data, we can see some intriguing patterns.

Firstly, here is a comparison of tropospheric temperatures above Australia from the University of Alabama- Huntsville (UAH), with surface air temperatures from the Bureau of Meteorology’s ACORN-SAT database.   To be comparable, both datasets are in anomalies from their 1981 – 2010 means.  The data are monthly since December 1978, with a 12 month running mean.

Fig. 1

uah v mean

Both datasets show concurrent rises and falls and are very similar (though not always).  Note how Acorn means were very much cooler in 2011 -2012 and much hotter in 2013.  Note also that 2014 has Buckley’s of being the hottest year on record.

Mean equals the average of maximum and minimum, so let’s look at maxima and minima.

Fig. 2

uah v max & min

Note that UAH usually tracks Acorn maxima, except when it doesn’t- shown above by the Xes.

Perhaps it has something to do with rainfall, or lack of it.  In the next plot, rainfall is inverted, so dry is at the top, wet at the bottom.

Fig. 3

uah v rain inv

Incidentally, the Bureau also has 9 a.m. and 3 p.m. cloud data available.  Note how closely both cloud datasets match, and how rainfall largely corresponds.

Fig. 4

rain v cloud

And the Southern Oscillation Index runs in close partnership with rainfall- sometimes SOI leads rain, sometimes rain leads SOI.

Fig. 5

rain v soi

Which is why I don’t take a lot of notice of predictions based on SOI.

Now see what happens when we plot inverted rainfall (dry at the top, wet at the bottom) and maxima.

Fig. 6

rain v max

Only once does 12 month mean maximum temperature precede 12 month rainfall (1991-1992).  At all other times, rainfall peaks or troughs occur before maxima (or at most, simultaneously).

With minima, the lead is even more obvious, however there are apparent exceptions in 1982 and 1994-1995, although these may be further examples of rain leading minima by more than a year (marked with “?”).

Fig. 7

rain v min

When we compare maxima with minima, the pattern is clear.

Fig. 8

max v min

Only in the summer of 1994-1995 do the records diverge.

Generalisations (and farmers have known about these rules of thumb for years):

  1. Climate is cyclical.  Rain and temperature rise and fall in roughly two or three year cycles.
  2. It always rains after a drought.
  3. Dry years are followed by spikes in maximum and minimum temperatures, from one to several months later.
  4. Wet years, with heavy cloud and rain, cause sharp drops in minimum and maximum temperatures, from one to several months later.
  5. Maximum temperatures lead minimum temperatures by several months in wet years, and by a shorter period in dry years.
  6. There are exceptions to all of the above.

Next step: Australia is a large continent with several distinct climatic regions.  I will next look at smaller regions to see if the above generalisations hold true and indeed may be modified or enhanced.

More Bizarre Adjustments

November 5, 2014

In September, the Bureau of Meteorology added two extra tabs to its ACORN-SAT webpage, in response to media and public pressure.  The first tab (“Adjustments”) included a link to a list of temperature adjustments for each of its 112 Acorn stations.  (This had been promised two and a half years earlier.)

Soon after, and probably in response to continued interest in adjustments at Amberley, Rutherglen, and Deniliquin (amongst others), six links to PDF files were added at the bottom of the adjustment page, which gave further explanations and summaries of adjustments at six individual sites- Amberley, Deniliquin, Mackay, Orbost, Rutherglen, and Thargomindah. (Click to enlarge.)

station summaries

Two days ago I posted about the bizarre case of Mackay 33119, listing differing adjustments from the two sources, extra neighbours found, and finding that the set of adjustments in the individual summary did not match the end result (the Acorn record for Mackay).

I thought this must be just a freak problem with Mackay.  Surely the other examples couldn’t all be wrong.

Not so.

Here is a table summarising the adjustments listed by the Bureau in the 28  page Station adjustment summary list, compared with the individual station summaries (click to enlarge).

adj comp table

Only one station (Deniliquin) has matching pairs of adjustments- but none are the same.

Out of 25 pairs of matching adjustments, only one pair has the same adjustment.

Most of the adjustments differ by only a few hundredths of a degree, but some are hugely different (over 1 degree in the case of Mackay).

There are a total of 60 adjustments, but 10 of these do not have a matching adjustment.  Seven of the extras are in the individual station summaries, three are from those in the original 28 page list.

Note that these station summaries are “indicative of the sorts of adjustments made across the 112 ACORN-SAT sites”. As a result, we can have no confidence in the accuracy of the Bureau’s adjustments, and we are left wondering what the Bureau would have us believe are the real temperatures at any site.

An old school teacher’s response to such sloppy work?

Fail.  Check your work and repeat.  Stay in at lunch time until you get it right.

How an ABC Poll went badly wrong

November 4, 2014

Yesterday the ABC (and other media) were breathlessly reporting the doom and gloom climate story from the IPCC, including 4 to 5 degrees of warming if we don’t stop using fossil fuel.

ABC News Radio put up a straw poll on their website to gauge public opinion.

After a slow start, the responses climbed.  At first, the vote was predominantly “Yes”, but that changed quickly.  By early afternoon the vote had swung to 52% No 48% Yes, by early evening it was 81% No 19% Yes, and at 5.45 a.m. today (Queensland time- not that Mickey Mouse time) it was as below:-

poll

91% No, 9% Yes.

Thank you, Our ABC!

The Bizarre Case of Mackay 33119

November 3, 2014

What has the Bureau of Meteorology done to Mackay’s temperatures?

Mackay’s temperature records from the old Post Office, Te Kowai, and the Met. Office have been combined into one, and this has been “homogenised” by reference to Mackay’s neighbours.

But in attempting to justify their actions the Bureau has provided TWO lists of neighbour stations and TWO lists of adjustments at their website.

First, the list of “neighbour” stations used at Mackay (from the 28 page Adjustments document at http://www.bom.gov.au/climate/change/acorn-sat/documents/ACORN-SAT-Station-adjustment-summary.pdf ).

33119    Mackay Met. Office (Mt Bassett) 1960-2014 is the official Acorn site.

33046    Mackay Post Office  1910-1949  (4 km away)

33047    Te Kowai  1910-2011  (10 km)

33058    Pine Islet Lighthouse  (70 km)

39023    Cape Capricorn Lighthouse (335 km)

33013    Collinsville Post Office  (154 km)

39083    Rockhampton Aero  (283 km)

32005    Cape Cleveland Lighthouse (290 km)

33077    Pacific Heights (Yeppoon- 273 km)

32078    Ingham  (421 km)

39122    Heron Island  (380 km)

32037    South Johnstone Experiment Station (Tully- 517 km)

34002    Charters Towers PO  (329 km)

33001    Burdekin Shire Council (Ayr- 255 km)

33007    Bowen PO  (160 km)

39069    Walterhall (Mt Morgan- 300 km)

35019    Clermont PO  (246 km)

However, they provide a different list in the explanation for Mackay’s adjustments given at http://www.bom.gov.au/climate/change/acorn-sat/documents/station-adjustment-summary-Mackay.pdf at a link from the site above, and the two explanations are quite different.

Here is the second list, including four extras:

mackay explanation list

You can imagine the reaction of Mackay residents on finding that Mackay’s temperatures have been homogenised using Tully, Heron Island, Townsville, Mount Morgan, Charters Towers, Clermont, and their old rival, Rockhampton.  None of these places has a climate anything like Mackay’s.

Further, the Bureau claims that Townsville and Rockhampton are excluded from climate analyses because they are both affected by Urban Heat Island (UHI) warming, but here they have been included in the climate analysis of Mackay.

Now the lists of adjustments:

mackay adjustments comp

The matching adjustments are completely different, and there are three extra breakpoints detected by statistical means, with adjustments, including two extra for maxima.  So which set of adjustments was actually used?

Here is a chart of Mackay’s annual maximum temperature records. Suffice to say that the Mackay record is a mess, and good luck to anyone trying to homogenise it.

Fig. 1:

mackay max chart

These are the results from applying the two lists of adjustments to the raw Mackay temperatures, to see which matches the Acorn records.

Fig.  2:   Calculated maximum temperatures (raw temperature with listed adjustments applied) minus Acorn temperatures.  Zero difference equals a perfect match.

mky replic

Fig. 3:   Minima:

mky replic min

The original list given in the 28 page list of adjustments appears to be the one used for both maxima and minima.  Mackay Acorn maxima cannot be replicated with the Station temperature adjustment summary list, which has two adjustments clearly not used, and is moreover confusing and does not follow the protocol for 1939-1940.  Similarly, there is an additional adjustment for minima which does not match the Acorn record.

The summary list appears to have been put together in a hurry in an attempt to head off criticism about lack of transparency.  But why the different adjustments?

And were the actual adjustments justified?  A simple test is to find the differences between the station being homogenised and its neighbours.  If Mackay has been properly homogenised, the average difference after homogenisation should have a trend close to zero.  Here are the results:

Fig. 4:  Average differences in anomalies of the 10 listed neighbours for the period around the “statistical” breakpoint at 01/01/1971.

mky 1971 raw adj diff comp

The adjustment of about -0.3C for all years up to 1970 makes the differences worse.  Interestingly, when the two most distant sites to the north and south are excluded (South Johnstone and Heron Island), the trend in raw difference is almost zero.  The raw Mackay MO record is similar to the neighbours, without any adjustment.

Fig. 5: As for Fig. 4, but excluding 2 distant sites:

mky 1971 raw adj diff excl 2 sites

The adjustments to the Post Office for 01/01/1941 and 01/01/1948 cause the following differences:

Fig. 6:

mky PO raw adj diff comp

Once again, there is a major difference between the Acorn record and the average of the neighbours, as shown by the steep trend- not much better than the raw difference.

To conclude,

  1. the Bureau has made an embarrassing mistake in publishing two different lists of adjustments and neighbours for Mackay
  2. the adjustments listed in the Mackay station adjustment summary are not those actually made
  3. adjustments are based on “neighbours” up to 500 km away, including two with UHI effect
  4. very few of these neighbours have climates similar to Mackay’s
  5. differencing shows that homogenising makes Mackay Met Office maxima LESS like the neighbours, and Post Office maxima not much closer.

If the adjustments at Mackay are, as the Bureau claims, “indicative of the sorts of adjustments made across the 112 ACORN-SAT sites”, then we can look forward to finding many more problems.