Archive for the ‘temperature’ Category

Adjustments Grossly Exaggerate Monthly and Seasonal Warming

October 4, 2014

The Bureau of Meteorology has reportedly claimed “an extensive study has found homogeneity adjustments have little impact on national trends and changes in temperature extremes.”  (Weekend Australian, August 23-24).

I have always said that the true test of the homogenisation process is its effect on national trends.  Problems at individual stations like Rutherglen are merely symptoms of a system wide malady.

If the adjustments really do have “little impact on national trends” then the Acorn dataset is a reliable indicator of broad temperature change in Australia.

If not, the Bureau has a problem.

So, how do we define “little impact”?

The Bureau has known since March 2012 that mean annual temperature increase from 1911 to 2010 in adjusted data (+0.94C) is 36% greater than in unadjusted data (+0.69C).  This information is publicly available in Table 1 on page 14 of On the sensitivity of Australian temperature trends and variability to analysis methods and observation networks  (CAWCR Technical Report No. 050), R.J.B. Fawcett, B.C. Trewin, K. Braganza, R.J Smalley, B. Jovanovic and D.A. Jones , March 2012 (hereafter CTR-050).  In this paper the authors claim that the rise in unadjusted data is “somewhat smaller”.  If this is so, then what increase in trend over unadjusted data may be considered to be beyond small or “little impact”? 50%? More than 50%?

What about 200%?

The Bureau has this graphic on their new Adjustments tab, which presumably is meant to support the claim of “little impact”:

Fig. 1: Official comparison (click graphics to enlarge)

BOM graphic

How big is that increase?  The devil is in the detail- monthly and seasonal trends, which the Bureau is yet to analyse.

According to the Bureau, AWAP (Australian Water Availability Project) represents unadjusted data. (It’s not, CTR-050 even calls it “partially homogenised”, and there are major issues with it, but that’s another story to be discussed later.  For now, let’s play along with calling it “unadjusted”).  Using this same “unadjusted” data, and the same method as the Bureau, here are results for the 1911 – 2013 period.  (See the Appendix below for full details.)

These tables summarize the results.  Highlighted cells show large ( > 50%) difference.

Fig. 2:  Summary Table: Percentage Increases to Unadjusted Data- Seasons

summary table seasons

The major effect is on summer trend:  increase in Mean trend 64%, Maxima 200%.

Fig .3:  Summary Table: Percentage Increases to Unadjusted Data- Months

summary table months

In Maxima trends, of the hot months, November, December and January have had large increases, and February and March have had cooling trends reversed.

June and November Mean, Minima, and Maxima trends have been massively increased.

One month (August) has had a warming trend reduced.

May, July, August, and September are largely unchanged.


Compared with ‘unadjusted’ data, for the period 1911 – 2013 Acorn shows obvious changes in monthly and seasonal data.  Exploration of the reasons for this needs to be included in the terms of reference of the forthcoming “independent review”.

The difference between AWAP and Acorn, especially in summer maxima, is of particular concern for anyone wishing to analyse national data.  For example: What was the national summer maximum in 1926?  AWAP says 35.87C.  Acorn says 33.53C.  Which dataset is to be believed?

The Bureau has a problem.

The Acorn dataset is NOT a reliable indicator of broad temperature change in Australia.

Appendix: Background, Charts, Methods, and Analysis

CTR-050 analyses data for the 1911-2010 period, comparing Acorn with several other datasets, including AWAP.  All trends are determined by quadratic fit, rather than linear, to better show the temperature trends across the period: cooling then warming.  The authors also use anomalies from 1981-2010 means.

This table shows the change in temperature over the period, which represents trend per 100 years, (and I am annoyed at myself for not reading this more closely two years ago.)

Fig.4:  Table 1 from CTR 050:

BOM table 1 comps

The authors explain (pp. 41-46) that the difference between AWAP and Acorn is mainly between 1911 and 1955 and is largely due to the large impact on national temperature of very few remote sites in the earlier years of last century, and station moves to cooler sites around 1930 and the 1940s.  That may certainly be true, but the large discrepancy calls for closer analysis.

My methods

Monthly and annual AWAP data (minima, maxima, and mean) 1911 – 2013 obtained from the Bureau allows analysis of the impact the adjustments.  I use 1961 – 1990 as the reference period for anomalies.  I also use quadratic trends and calculate temperature change per 100 years by (last quadratic trendline point – first point) X 100/103.  (These first and last points are accurately determined to 0.01C by zooming in on Excel charts- see Figures 22 and 23 below.)  I calculate percentage change in 100 year trend as {(Acorn trend – AWAP trend)/AWAP trend} x 100.

For example:  Annual means.

Quadratic first point (1911)   Quadratic last point (2013)    Change

AWAP:   -0.13                          +0.56                            +0.69

Acorn:   -0.34                           +0.58                            +0.92

AWAP Quadratic trend per 100 years =  0.69 X 100/103 = 0.67

Acorn Quadratic trend per 100 years =   0.92 X 100/103 = 0.89

Percentage change in trend = {(0.89 – 0.67) / 0.67} X 100 = 32.8%.

While my analysis largely confirms the figures in the Figure 4 above, the devil is in the detail.

Firstly, here are charts for comparison of mean temperatures, showing linear and quadratic trends to 2013:

Fig. 5: Linear

mean linear

Fig. 6: Quadratic

mean quadratic

Linear analysis produces a trend value of 31%, a little less than quadratic .  Acorn adjustments produce a quadratic trend about 32.8% greater than AWAP- not as great as 1911-2010, but still substantial.  Quadratic trend lines produce a better fit than linear and clearly show the earlier cooling.

Fig.7:  Annual Minima

min quadratic

Over 25% increase.

Fig. 8: Annual Maxima

max quadratic

36.7% increase.

Seasonal and Monthly Means:

Fig. 9:  Table of Seasonal Differences for Means.

mean table seasons

Note summer mean trend has been increased by 64%.  Graphs may make the comparison starker.

Fig. 10:  Comparison of 100 year trends in unadjusted and adjusted seasonal data.

mean trends diff seasons

Fig. 11: Percentage Difference in Trends

mean trends diff % seasons

Fig. 12: Comparison of 100 year trends in unadjusted and adjusted monthly data.

mean trends comp

Fig. 13:  Percentage Difference in Trends

mean trends diff % months

February trend doubled, March, June, and November are increased by about 80%.


Fig. 14:  Table of Seasonal Differences for Minima.

min table seasons

Fig. 15:  Comparison of 100 year trends in unadjusted and adjusted seasonal data.

min trends comp seasons

Fig. 16:  Percentage Difference in Trends

min trends diff % seasons

Fig. 17: Comparison of 100 year trends in unadjusted and adjusted monthly data.

min trends comp

Fig. 18:  Percentage Difference in Trends

min trends diff %

Note the doubling of the June minima trend, and October and November increased by 50%.


Fig. 19:  Table of Seasonal Differences for Maxima.

tmax table seasons

Fig. 20:  Comparison of 100 year trends in unadjusted and adjusted seasonal data.

max  trends seasons

Fig. 21:  Percentage Difference in Trends- we need to rescale the y-axis!

max trends diff % seasons

Don’t believe the 200% figure?  Here are close ups of the graph.

Fig. 22:  Summer maxima detail

max summer quadratic bottom

Fig. 23:

max summer quadratic top

Fig. 24: Comparison of 100 year trends in unadjusted and adjusted monthly data.

max trends comp

Note cooling trends in February and March reversed., August reduced.

Fig. 25:  Percentage Difference in Trends

max trends diff % months

Strong August warming slightly reduced.  No calculation for February and March.  January, June, December greatly warmed.  November massively warmed.

Why the huge discrepancies between unadjusted and adjusted data?

Acorn data freely available at

AWAP data available at a cost on request from

A Check on ACORN-SAT Adjustments: Part 1

September 18, 2014

I have commenced the long and tedious task of checking the Acorn adjustments of minimum temperatures at various stations by comparing with the lists of “highly correlated” neighbouring stations that the Bureau of Meteorology has kindly but so belatedly provided.   Up to 10 stations are listed for each adjustment date, and presumably are the sites used in the Percentile Matching process.

It is assumed by the Bureau that any climate shifts will show up in all stations in the same (though undefined) region.  Therefore, by finding the differences between the target or candidate station’s data and its neighbours, we can test for ‘inhomogeneities’ in the candidate site’s data, as explained in CTR-049, pp. 44-47.  Any inhomogeneities will show up as breakpoints when data appears to suddenly rise or fall compared with neighbours.  Importantly, we can use this method to test both the raw and adjusted data.

Ideally, a perfect station with perfect neighbours will show zero differences: the average of their differences will be a straight line at zero.  Importantly, even if the differences fluctuate, there should be zero trend.  Any trend indicates past temperatures appear to be either relatively too warm or too cool at the station being studied.  It is not my purpose here to evaluate whether or not individual adjustments are justified, but to check whether the adjusted Acorn dataset compares with neighbours more closely.   If so, the trend in differences should be close to zero.

In all cases I used differences in annual minima anomalies from the 1961-1990 mean, or if the overlap was shorter than this period, anomalies from the actual period of overlap.  Where I am unable to calculate differences for an Acorn merge or recent adjustment due to absence of suitable overlapping data (e.g. Amberley 1997 and Bourke 1999, 1994), as a further test I have assumed these adjustments are correct and applied them to the raw data.

I have completed analyses for Rutherglen, Amberley, Bourke, Deniliquin, and Williamtown.

The results are startling.

In every case, the average difference between the Acorn adjusted data and the neighbouring comparison stations shows a strongly positive trend, indicating Acorn does not accurately reflect regional climate.

Even when later adjustments are assumed to be correct the same effect is seen.

Interim Conclusion:

Based on differencing Raw and Adjusted data from listed comparison stations at five of the sites that have been discussed by Jennifer Marohasy, Jo Nova, or myself recently, Acorn adjustments to minima have a distinct warming bias.  It remains to be seen whether this is a widespread phenomenon.

I will continue analysing using this method for other Acorn sites, including those that are strongly cooled.  At those sites I expect to find the opposite: that the differences show a negative trend.

Scroll down for graphs showing the results.



(Note the Rutherglen raw minus neighbours trend is flat, indicating good regional comparison.  Adjustments for discontinuities should maintain this relationship.)

Amberley (a)


(Note that the 1980 discontinuity is plainly obvious but may have been over-corrected.)

Amberley (b): 1997 merge (-0.44) assumed correct

 amberley inc 1997

Treating the 1997 adjustment as correct has no effect on the trend in differences.

Bourke (a)


Bourke (b):  1999 and 1994 merges assumed correct.

bourke inc merges

No change in trend of differences.



(Note the adjusted differences still show a strong positive trend, but less than the other examples.)



(Applying an adjustment to all years before 1969 produces a strong positive trend in differences.)

Better Late Than Never- BOM Releases Adjustment Details

September 11, 2014

On Monday, quietly and without any announcement, a new tab appeared on the Bureau’s ACORN-SAT webpage.

adj tab

This “Adjustments” tab opens to a page explaining why homogenisation is necessary, supposedly showing how the adjustments don’t make much difference to the mean temperatures, and how Australia really is warming because everyone agrees.  More on this later.  So how do we get to see the actual adjustments for each site?  Tucked away under the first graph is a tiny link:

adj tab link

Click on that and a 27 page PDF file opens, listing every Acorn station, dates and reasons for adjustments, and most importantly, a list of reference stations used for comparison.  (You have to go to Climate Data Online to find the station names, their distance away, site details, and their raw data.)

Finally it will be possible to check the methods and results using the correct comparison stations- until now we could only guess.

Back in September, 2011 the Independent Peer Review Panel made a series of recommendations, including that

“C1. A list of adjustments made as a result of the process of homogenisation should be assembled, maintained and made publicly available, along with the adjusted temperature series. Such a list will need to include the rationale for each adjustment.”

The Bureau responded on 15 February 2012, just before the release of Acorn:

“Agreed. The Bureau will provide information for all station adjustments (as transfer functions in tabular format), cumulative adjustments at the station level, the date of detected inhomogeneities and all supporting metadata that is practical. This will be provided in digital form. Summaries of the adjustments will be prepared and made available to the public.”

That was two and a half years ago.  What took so long?  Why was it not publicly available from the start?  Perhaps it is just a co-incidence that the long awaited information was released shortly after a series of articles by Graham Lloyd appeared in The Australian, pointing out some of the apparent discrepancies between raw and adjusted data.  Graham Lloyd deserves our heartfelt thanks.

The Bureau of Meteorology has been dragged kicking and screaming into the 21st Century.  The Bureau is having trouble coming to terms with this new era of transparency and accountability, an era in which decisions are held up to public scrutiny and need to be defensible.

I trust we won’t have to wait another two and a half years for the other information promised, such as “sufficient station metadata to allow independent replication of homogeneity analyses” and “computer codes… algorithms… and protocols”,  “the statistical uncertainty values associated with calculating Australian national temperature trends” and “error bounds or confidence intervals along the time series”

The final recommendation of the Review Panel, and undertaking by the Bureau:

“E6. The Review Panel recommends that the Bureau assembles and maintains for publication a thorough list of initiatives it has taken to improve transparency, public accessibility and comprehensibility of the ACORN-SAT data-set.

Agreed. The Bureau will provide such information on the Bureau website by March 2012.”

I must have missed that.




Homogenisation: A Test for Validity

September 8, 2014

This follows on from my last post where I showed a quick comparison of Rutherglen raw data and adjusted data, from 1951 to 1980, with the 17 stations listed by the Bureau as the ones they used for comparison when detecting discontinuities. 

Here is an alternate and relatively painless way to check the validity of the Bureau’s homogenisation methods at Rutherglen, based on their own discontinuity checks.  According to the “Manual” (CAWCR Technical Report No. 49), they performed pair-wise comparisons with each of the 17 neighbours to detect discontinuities.  An abbreviated version of this can be used for before and after comparisons.  For each of the 17 stations, I calculated annual anomalies from the 1961-1990 means for both Rutherglen and the comparison site, then subtracted the comparison data from Rutherglen’s.  I did the same with Rutherglen’s adjusted Acorn data.

A discontinuity is indicated by a sudden jump or drop in the output.  The ideal, if all sites were measuring accurately and there are no discontinuities, would be a steady line at zero: a zero value indicates temperatures are rising or falling at the same rate as neighbours.  In practice no two sites will ever have the same responses to weather and climate events, however, timing and sign should be the same.  Therefore pairwise differencing will indicate whether and when discontinuities should be investigated for possible adjustment.

Similarly, pairwise differencing is a valid test of the success of the homogenisation process.  Successful homogenisation will result in differences closer to zero, with zero trend in the differences over time.  The Bureau has told the media that adjustments are justified by discontinuities in 1966 and 1974.  Let’s see.

Fig. 1:  Rutherglen Raw minus each of 17 neighbours

pairwise diffs Rutherglen Raw

Note: there is a discernible drop in 1974, to 1977.  There is a very pronounced downwards spike in 1967 (ALL differences below zero, indicating Rutherglen data were definitely too low.)  There also a step up in the 1950s, and another spike upwards in 1920.  Rutherglen is also lower than most neighbours in the early 1930s.  Also note several difference lines are obviously much higher or lower than the others, needing further investigation, but the great majority cluster together.  Their differences from Rutherglen are fairly consistent, in the range +/- 1 degree Celsius.

Now let’s look at the differences AFTER homogenisation adjustments:

Fig. 2:  Rutherglen Acorn minus the neighbours: The Test

pairwise diffs Rutherglen Acorn

The contrast is obvious.  The 1920 and 1967 spikes remain.  Differences from adjusted data are NOT closer to zero, most of the differences before 1958 are now between 0 and -2 degrees Celsius, and there is now an apparent large and artificial discontinuity in the late 1950s.  This would indicate the need for Rutherglen Acorn data to be homogenised!

Compare the before and after average of the differences:

Fig. 3:

pairwise diffs Rutherglen Raw v Acorn average

There is now a large positive trend in the differences when the trend should be close to zero.

There are only two possible explanations for this:

(A)  The Bureau used a different set of comparison stations.  If so, the Bureau released false and misleading information. 

(B)   As this surely can’t be true, then if these 17 stations were the ones used, this is direct and clear evidence that the Bureau’s Percentile Matching algorithm for making homogenisation adjustments did not produce correct, successful, or useful results, and further, that no meaningful quality assurance occurred.

If homogenising did not work for Rutherglen minima, it may not have worked at the other 111 stations. 

While I am sure to be accused of “cherry picking”, this analysis is of 100% of the sites for which the identities of comparison stations have been released.  When the Bureau releases the lists of comparison stations for the other 111 sites we can continue the process.

A complete audit of the whole network is urgently needed.

Rutherglen: Spot the Outlier

September 2, 2014

In today’s Australian there was another article by Graham Lloyd, “Climate scientists defend data changes”. The Bureau of Meteorology is quoted as claiming that “statistical analysis of minimum temperatures at Rutherglen indicated jumps in the data in 1966 and 1974….. These changes were determined through comparison with 17 nearby sites”.

 Two and a half years after being asked to explain the reasons for the myriads of changes to the data, the Bureau has finally given up some of the information it should have released in 2012.  I have been given the names of these 17 sites.  They are:

74034 Corowa, 82053 Wangaratta, 82002 Benalla, 72097 Albury Pumping Station, 82100 Bonegilla

74106 Tocumwal, 81049 Tatura, 81084 Lemnos, 72023 Hume Reservoir, 82001 Beechworth

72150 Wagga Wagga, 74114 Wagga Research Centre, 80015 Echuca, 74039 Deniliquin (Falkiner Memorial)

74062 Leeton, 74128 Deniliquin, and 75032 Hillston.

 This at last allows me to understand how they went about turning a cooling trend of -0.33C per 100 years into a warming trend of +1.74C. 

 Fig. 1: Rutherglen unadjusted data vs adjusted, 1913 – 2013

rutherglen tmin

  I checked the monthly unadjusted minimum data for Rutherglen, the adjusted data for Rutherglen, and the unadjusted data at all 17 of the listed neighbours, in the period 1951 – 1980, which according to the Bureau is the critical period containing the 1966 and 1974 break points.  30 years is a suitably long period for analysis.  For the technically minded, I calculated monthly anomalies from the 1951-1980 means for each record, then 12 month averages.  This should allow us to see the problems around 1966 and 1974.

 Here is a chart of the results.  Can you spot the outlier?

 Fig. 2:  Rutherglen raw (unadjusted), the 17 neighbours’ raw data, and Rutherglen Acorn (adjusted)

 rutherglen v Acorn v neighbours all

You won’t be able to pick out the light blue line of Rutherglen raw data in the spaghetti lines of the neighbours, but you should be able to see the dark red of the adjusted data peeping above and below the others.

 For a clearer picture, here is the same information, but with the 17 neighbours averaged to a single orange line.

 Fig. 3: Rutherglen unadjusted (blue), average of the 17 neighbours (orange), and Acorn- the homogenised version of Rutherglen (dark red).

 rutherglen v Acorn v neighbours avg

Forgive me, but I thought the idea of “homogenising” was to adjust the data so that it is not so different from the neighbours.  That happens in1966.  They got that right, but not in 1974, where the adjustments have increased the difference, and have produced warming.  Odd things also happen in 1952, 1954, 1957, 1969, and 1975-80.

 It is clear that the changes to the temperatures at Rutherglen do not “homogenise” them.  They make the differences from the neighbours greater, and change a cooling trend into a warming one.

 This is not unique to Rutherglen- adjustments warm the temperature trends at 66 of the 104 Australian sites, and warm the national mean temperature trend by around 47%.

 But what would I know- I’m just an amateur according to Professor Karoly.

The Australian Temperature Record- A Quick Update

August 23, 2014

This morning (Saturday 23 August) the Weekend Australian published articles by Graham Lloyd, their Environment Editor, on homogenisation practices at the Bureau of Meteorology as questioned by Jennifer Marohasy.  As I had a small part to play in bringing this to public light, here is a brief post to bring readers up to date.

The last paragraph in the second article reads:

“And the bureau says an extensive study has found homogeneity adjustments have little impact on national trends and changes in temperature extremes.”

This is laughable.  Here is a graph of the national means of Raw and Homogenised minima data from 83 sites (out of 104) that I was able to compare directly. (I also analysed the remaining sites, finding 47% bias, but because large slabs of data had to be left out this is not reliable.)

Fig. 1: Australian mean minimum temperature anomalies 1910-2012

Tmin comp

 The ‘raw’ trend is +0.63C per 100 years.  The adjusted trend is +1.05C.  The effect of the homogenisation adjustments is an increase in the national trend of +0.42C or 66.6%.  So much for “little impact.”

The article referred mainly to adjustments at Amberley and Rutherglen.

Fig. 2:  Amberley minima

amberley tmin

According to the BOM, the major adjustment was due to a pronounced discontinuity around 1980, that is, Amberley’s drop in temperature is not reflected in those of neighbouring sites, as is evidently correct.

Fig. 3:  Amberley compared with the mean of 5 Acorn neighbours

amb raw v reg mean raw

 However, the nearest Acorn site only 50km away, Brisbane Aero, also has a pronounced cooling trend, and a local cooling cannot be discounted.

An adjustment to the raw data before 1980 may be warranted, however, the size of the adjustment is questionable to say the least.  The resulting trend at Amberley has now become greater than the trend of adjusted data at every one of the Acorn neighbours, and more than +0.86C greater than their mean.

Fig. 4: Amberley’s adjusted (Acorn) data vs mean of adjusted data at 5 closest Acorn sitesamb acorn v reg acorn

Rutherglen in Victoria again shows cooling turned into warming.

Fig.5: Rutherglen minima

rutherglen tmin

And again, the Acorn adjustments make Rutherglen’s trend greater than every one of its neighbours’ adjusted trends, as well as their mean:

Fig. 6: Rutherglen Acorn vs neighbours’ Acorn (mean)

rutherglen acorn v reg acorn


The BOM is defending its territory, but this latest media exposure will mean increasing and critical scrutiny.

Click for further examples and background.




Adjustments vs CO2

August 3, 2014

Steven Goddard has posted about the remarkable correlation between USHCN adjustments and atmospheric carbon dioxide concentrations:

goddard co2

Here’s my plot of Australian adjustments to minima, Acorn minus raw vs CO2 data (downloaded from NASA GISS at ):

acorn vs co2

R2= 0.777 not as impressive as 0.988, so not proof of anything except past cooling adjustments which we already knew.  Interesting all the same.

Tarcoola- A Cooling Outlier

July 28, 2014

In a previous post I looked at the warming outliers in the Acorn network- those sites that had homogenisation adjustments that created a difference of more than +2 degrees Celsius between the Acorn trend and the raw trend in minima.  In all of these six examples, the adjustments had created trends that were not just greater than the raw trends at each site, not just the mean of their Acorn neighbours raw data trends, but greater than the Acorn trends of their neighbours, and in all but one, greater than each of the individual Acorn trends of their neighbours.

In this post I consider the opposite scenario.  I look at one cooling outlier, Tarcoola in South Australia, where the cooling adjustments have created a difference in trend of -2.81C per 100 years.  There is one other, Forrest in W.A., with an enormous cooling adjustment of around -2.14C, but I have little faith in the accuracy of the data there.  Greg Geegman suggested in a comment that if a site that is adjusted downwards is cooled relative to the neighbour group, this may indicate the Percentile Matching algorithm operates iteratively, although Technical Report No. 49 does not mention this.  An alternative explanation might be that the algorithm is too sensitive and exaggerates necessary adjustments.

All data may be downloaded from the BOM website: Site networks and Climate Data Online.

Tarcoola is in the centre of South Australia:Tarcoola map

As before, I compare Tarcoola with its neighbours in the Acorn network, using anomalies from the 1961-1990 mean.

Fig. 1:  Tarcoola Acorn vs ‘Raw’ minimaTarcoola tmin

Cooler trend than raw.  Note the spurious data pre-1930.

Fig. 2: Tarcoola raw vs mean of Acorn neighbours’ rawtarcoola raw v reg mean raw

Tarcoola appears to need cooling.

Fig. 3: Tarcoola Acorn vs mean of neighbours’ meantarcoola acorn v reg raw

Cooler trend than neighbours’ raw

Fig. 4: Tarcoola Acorn vs mean of neighbours’ Acorntarcoola acorn v reg acorn

Cooler trend than neighbours Acorn

Tarcoola Acorn trend is also cooler than each of the neighbours’ individual Acorn trends.

So which neighbours were used to make the Tarcoola adjustments?


Both warming and cooling outliers show Acorn adjustments outperforming those of the neighbours. This suggests that the algorithm exaggerates adjustments, both warming and cooling, and needs serious re-examination.

Carnarvon- a closer look

July 24, 2014

A lot of interest was generated by my last post on the Acorn outliers, especially the dependence on very distant sites for homogenisation adjustments.  In this post I will compare Carnarvon’s closer neighbours – excluding Wittenoom, Meekatharra, and Morowa- to show how  a better understanding of Carnarvon’s minimum temperature record may be derived, and how reliance on more distant sites with different climate regimes can distort adjustments.

Carnarvon Airport 6011 has been recording temperatures since 1945.  Before that, Carnarvon Post Office 6062 has data from 1885 to 1950 so there is useful overlap.

Fig. 1: Carnarvon PO and Carnarvon Airport (raw minima)carnarvon raw

These records can be spliced by reducing the Post Office data for 1946-1948 by 0.4C (as the Post Office recorded increasingly warmer minima than the Airport in these years as shown by the monthly temperatures for 1949-1950) but making no changes to PO data before this, to produce a long composite record.  I commence at 1910 to compare with the official BOM figures.

Fig. 2: Carnarvon splice vs Carnarvon Acorncarnarvon tmin

Note how Acorn reduces minima from 1974.  Note the size of these adjustments.

Fig. 3: AdjustmentsCarnarvon adjustments

So at 1910 the Acorn record shows minima 1.8C less than the raw data.

Now lets look at how Carnarvon’s neighbours compare.  To do this we need to convert to anomalies from the 1961-1990 mean.  Neighbours are listed at Climate Data Online, ranging from 19 km  to 296 km away, but most are too short or have too much missing data to be useful.

The neighbours that I have used for comparison are:  Hamelin Pool (174.4 km away), Winning (211.3 km), Emu Creek (248.5 km), and Learmonth (296.3 km).  Learmonth is almost due north of Carnarvon, and like Carnarvon right on the coast, while the others are inland sites.

Fig. 4: Carnarvon and closest useful neighboursCarnarvon neighbours

We can see that there is close agreement for most of the time.  There are minor periods of disagreement in the 1970s and 1990s, but the major disagreement is 1926-1950.  Which is a more accurate reflection of Carnarvon minima- Carnarvon PO or Hamelin Pool?  To find which is the odd one out, we need to look at other sites with data for this period.

Fig. 5: Carnarvon vs Winning- anomalies from 1910-1939 meanCarnarvon v Winning

Winning, although confirming the approximate agreement 1910-1920,  has very little data for 1926-1950, so we have to look further afield… all the way to Geraldton, which like Carnarvon is on the coast, but 447 km south.

Fig. 6:  All neighbours including GeraldtonCarnarvon inc Geraldton

Hamelin Pool is clearly the outlier, so we can accept Carnarvon raw temperatures as reasonably accurate from 1910 to 1970.  There is a short period in the 1970s of disagreement, but little difference after that… and Acorn does not adjust after 1974 anyway.

How does this compare with Acorn?

Fig. 7: All anomalies including AcornCarnarvon acorn vs all

Can you pick the outlier?

We can only presume that the Acorn homogenisation depends on data fed into the algorithm from much, much further away.

I can see no justification for any major adjustment to the raw record at Carnarvon.


The Australian Temperature Record Revisited: Part 4- Outliers

July 16, 2014

In my previous posts I showed how the Acorn adjustments to the ‘raw’ minimum temperature data have the effect of enormously increasing the apparent trend across the whole network, and very differently in different regions.  In this post I am looking more closely at the six locations where the adjustments cause a change in trend of greater than +2 degrees Celsius.  These are:  Brisbane Airport, Amberley RAAF, Dubbo, Rutherglen, Rabbit Flat, and Carnarvon.

And I am mystified.

The purpose of homogenisation adjustments is to remove discontinuities in data, which show up as differences between the ‘candidate’ site’s record and those of its neighbours, the ‘reference’ sites.  The Acorn method of detecting discontinuities uses pairwise comparison with up to 40 neighbouring sites, and this includes sites many hundreds of kilometres distant.  Adjustments are made with a Percentile Matching algorithm which compares with up to 10 neighbouring sites.

I use my own method to compare sites with neighbours.  When comparing any sites, anomalies from a common base period (1961-1990) are used.  Only sites with data (at least 15 years) in this period can be used.  Sites also need long data records.  While in closely settled areas there will be a selection of observation sites, very few meet these requirements.  Therefore I compare the data of each of these six locations with those of their nearest surrounding Acorn sites’ ‘raw’ data, (adjusted by me only when necessary to create a long combined series), individually and with their mean.

Even with only five neighbours, for Carnarvon and Rabbit Flat these can be over 500km away.

I then repeat this using Acorn (adjusted) data for the neighbours.

The results are surprising.

Here are the six outliers and their surrounding Acorn neighbours:Outliers map

Note the remoteness of Rabbit Flat and Carnarvon.


Brisbane Air

Fig. 1a: Brisbane ‘raw’ spliced vs Acorn minimabris raw v acorn

The neighbours are: Amberley, Cape Moreton Lighthouse, Bundaberg, Gayndah, Miles, and Yamba Pilot Station.

Fig. 1b: Brisbane raw vs mean of neighbours (‘raw’ data)bris raw v reg mean raw

Fig. 1c: Brisbane Acorn vs neighbours’ raw meanbris acorn v reg raw

Note the adjusted trend (+1.95C per 100 years) is greater than the mean of the neighbours (+1.06) by nearly +0.9C.

Fig. 1d:  Brisbane Acorn vs mean of neighbours (Acorn, adjusted data)bris acorn v reg acorn

As you would expect, the data are now very similar, and the trend for Brisbane is thus 0.23C per 100 years less than the trend for the mean of the neighbours’ Acorn data.  (This is the only outlier site where this happens, as you will see.)


Neighbours are the same as Brisbane’s, including Brisbane, 50km away.

Fig. 2a:  Raw vs Acornamberley tmin

Fig. 2b: Amberley and mean of neighbours (raw).amb raw v reg mean raw

Fig.2c: Amberley Acorn vs neighbours mean (raw)amb acorn v reg raw

Note the trend is more than one degree steeper than the trend of the neighbouring Acorn sites’ raw data.

Fig.  2d: Amberley Acorn vs neighbours’ mean (Acorn)amb acorn v reg acorn

Amberley’s adjusted trend is +0.87C greater than that of the mean of its neighbours’ adjusted data.


Neighbours are: Gunnedah, Scone, Bathurst, Cobar, Wyalong

Fig. 3a:  Raw vs Acorndubbo tmin

Fig. 3b: Dubbo and mean of neighbours.dubbo raw v reg mean raw

Fig.3c: Dubbo Acorn vs neighbours mean (raw)dubbo acorn v reg raw

+1.47C difference.

Fig.  3d: Dubbo Acorn vs neighbours’ mean (Acorn)

dubbo acorn v reg acorn

Now only +1.29C per 100 years greater than the neighbours.


Neighbours are: Deniliquin, Wagga Wagga, Sale, Kerang, Cabramurra

Fig. 4a:  Raw vs Acornrutherglen tmin

Fig. 4b: Rutherglen raw and mean of neighbours (raw).rutherglen raw v reg mean raw

Note that Rutherglen’s cooling trend is only 0.3C different from that of its neighbours.

Fig.4c: Rutherglen Acorn vs neighbours mean (raw)rutherglen acorn v reg raw

Fig.  4d: Rutherglen Acorn vs neighbours’ mean (Acorn)rutherglen acorn v reg acorn

+0.51C per 100 years greater than the neighbours.

Rabbit Flat

Rabbit Flat is a roadhouse in the Tanami Desert on the track between Alice Springs and Halls Creek.  Climate Data Online shows the current Rabbit Flat site 015666 as being 71km away from the old closed site 015548, though the Acorn Station Catalogue says it’s only 200 metres.  This in itself is peculiar.

The nearest non-Acorn site is Balgo Hills 211 km  away.

Acorn neighbours are:  Giles (567km), Halls Creek (328km), Victoria River Downs (433km), Tennant Creek (440km), and Alice Springs (568km).

Fig. 5a:  Raw vs Acornrabbit flat tmin

Fig. 5b: Rabbit Flat and mean of neighbours (raw).rbt flt raw v reg mean raw

Fig.5c: Rabbit Flat Acorn vs neighbours mean (raw)rbt flt acorn v reg raw

+1.17C more warming than neighbours.

Fig.  5d: Rabbit Flat Acorn vs neighbours’ mean (Acorn)rbt flt acorn v reg acorn

Rabbit Flat adjustments give it a trend +0.75C more than the neighbours’.


Carnarvon’s Acorn neighbours are Learmonth (298km), Wittenoom (560km) , Meekatharra (524km), Geraldton (447km), and Morawa (538km).  The only non-Acorn site with continuous data for the early part of last century is Hamelin Pool 6025 (174km away).

Fig. 6a:  Raw vs Acorncarnarvon tmin

Fig. 6b: Carnarvon and mean of neighbours (raw).carnarvon raw v reg mean raw inc morawa

Now note the effect of just one of the neighbours- Morawa.

Fig. 6c:   Carnarvon raw vs neighbours’ mean excluding Morawacarnarvon raw v reg mean raw excl morawa

Note the much closer comparison.

Fig.6d: Carnarvon Acorn vs neighbours mean (raw) (including Morawa)carnarvon acorn v reg raw

Note the trend is +1.58C per 100 years more.

Fig.  6e: Carnarvon Acorn vs neighbours’ mean (Acorn)carnarvon acorn v reg acorn

The difference is +1.49C.

The Acorn trend at Carnarvon is also greater than the Acorn trends at each of the neighbours separately.

Conclusion: -

The Acorn adjustment algorithm creates homogenised data by comparing with up to 10 neighbouring sites.  I have shown that the adjustments have made the Acorn trends greater than, not only the raw data trends for each site, not only the raw data trends of the closest neighbours in the Acorn dataset, but in every case but one, greater even than the trends of Acorn homogenised data from the same neighbouring locations.   The adjustments created thus appear to be spurious and the algorithm faulty.



Get every new post delivered to your Inbox.

Join 42 other followers