Acorn or A Con: Acorn’s Daily Adjustments

Ken Stewart, June 2012

Comments on Jo Nova’s site, especially from Richard of NZ, indicate a need for the public to see samples of the daily adjustments made by the authors of ACORN-SAT. So here they are. All are for minima.

Please note: In this post I am not claiming that Acorn only cools past temperatures and warms recent ones. On the contrary, there are many examples of the opposite, e.g. Tennant Creek. There is a predominance of early cooling. However in this post I am looking at the actual adjustments.

First, Alice Springs. Here is a table showing 1943, when there was overlap between the old Post Office observations and the new airfield built for the war effort. Note that Acorn bears little resemblance to either of the raw data records.

Yr	Month	Day	Old	Air	Acorn	Diff old	Diff air
1943	9	1	1.1	2.6	0.6	-0.5	-2
1943	9	2	2.8	3.8	2.2	-0.6	-1.6
1943	9	3	5	7	3.7	-1.3	-3.3
1943	9	4	5.6	5.1	4.7	-0.9	-0.4
1943	9	5	4.4	3.8	3.1	-1.3	-0.7
1943	9	6	2.8	3.7	2.2	-0.6	-1.5
1943	9	7	2.2	5.7	1.3	-0.9	-4.4
1943	9	8	2.8	6	2.2	-0.6	-3.8
1943	9	9	8.9	12.6	7.4	-1.5	-5.2
1943	9	10	10.6	13.3	9.8	-0.8	-3.5
1943	9	11	11.7	11.3	10.3	-1.4	-1
1943	9	12	8.9	8.7	7.4	-1.5	-1.3
1943	9	13	8.9	14.9	7.4	-1.5	-7.5
1943	9	14	14.4	12.6	13.4	-1	0.8
1943	9	15	8.9	5.7	7.4	-1.5	1.7
1943	9	16	14.4	14.4	13.4	-1	-1
1943	9	17	11.7	7.9	10.3	-1.4	2.4
1943	9	18	9.4	7.8	8	-1.4	0.2
1943	9	19	9.4	12.1	8	-1.4	-4.1
1943	9	20	17.2	17.3	15.7	-1.5	-1.6
1943	9	21	18.3	21.4	13.8	-4.5	-7.6
1943	9	22	12.2	12.1	10.3	-1.9	-1.8
1943	9	23	8.9	7.4	7.4	-1.5	0
1943	9	24	12.2	9.1	10.3	-1.9	1.2
1943	9	25	13.9	9.7	13.5	-0.4	3.8
1943	9	26	8.3	13.4	7.2	-1.1	-6.2
1943	9	27	7.2	8.7	5.9	-1.3	-2.8
1943	9	28	7.8	10.9	6.5	-1.3	-4.4
1943	9	29	6.7	7.7	5.6	-1.1	-2.1
1943	9	30	7.2	3.6	5.9	-1.3	2.3
1943	10	1	9.4	3.8
1943	10	2	5	5.7
1943	10	3	6.1	7.8
1943	10	4	5	8.5
1943	10	5	5.6	9
1943	10	6	8.3	7.9
1943	10	7	9.4	10.8
1943	10	8	13.9	17.2
1943	10	9	14.4	16.9
1943	10	10	15	16.3
1943	10	11	15	21.6
1943	10	12	13.3	16.3
1943	10	13	11.7	18.7
1943	10	14	11.1	17.1
1943	10	15	12.8	16.7
1943	10	16	11.7	17.1
1943	10	17	11.7	18.9
1943	10	18	12.2	17.7
1943	10	19	15	20.3
1943	10	20	15.6	20.1
1943	10	21	14.4	20.7
1943	10	22	12.2	20.7
1943	10	23	11.7	16.7
1943	10	24	12.2	13.3
1943	10	25	12.2	15.5
1943	10	26	15	15.3
1943	10	27	15	15.2
1943	10	28	15	16.2
1943	10	29	7.8	10.2
1943	10	30	12.2	15.6
1943	10	31	12.2	18.8
1943	11	1	12.2	20.4
1943	11	2	17.2	19.4	18.5	1.3	-0.9
1943	11	3	12.2	12.2	13.4	1.2	1.2
1943	11	4	12.2	10.7	13.4	1.2	2.7
1943	11	5	11.7	10.4	11.8	0.1	1.4
1943	11	6	12.8	12.7	13.6	0.8	0.9
1943	11	7	11.1	9.5	9.8	-1.3	0.3
1943	11	8	11.1	10.4	9.8	-1.3	-0.6
1943	11	9	10.6	11.3	9.2	-1.4	-2.1
1943	11	10	11.1	14.8	9.8	-1.3	-5
1943	11	11	12.2	13.6	13.4	1.2	-0.2
1943	11	12	15.6	14.9	15.2	-0.4	0.3
1943	11	13	12.8	14.6	13.6	0.8	-1
1943	11	14	15	14.4	15.2	0.2	0.8
1943	11	15	18.9	20	19.1	0.2	-0.9

What happened to October?

Here’s April and May 1931, just before a site shift in 1932. Note the average adjustment for these two months is -2.8 degrees, with two days adjusted down by 8 degrees, and not accidentally.

Yr	Month	Day	PO	Acorn	Diff air	365d av PO	Acorn	365d diff
1931	4	1	23.3	20.5	-2.8	13.5	12.0	-1.5
1931	4	2	19.2	15.3	-3.9	13.5	12.0	-1.5
1931	4	3	11.4	9.3	-2.1	13.5	12.0	-1.5
1931	4	4	8.3	5.9	-2.4	13.5	12.0	-1.5
1931	4	5	8.3	5.9	-2.4	13.5	12.0	-1.5
1931	4	6	8.3	5.9	-2.4	13.5	12.0	-1.5
1931	4	7	10.6	9.1	-1.5	13.5	12.0	-1.5
1931	4	8	10.3	8.3	-2	13.5	12.0	-1.5
1931	4	9	10.6	9.1	-1.5	13.5	12.0	-1.5
1931	4	10	7.2	5.3	-1.9	13.5	12.0	-1.5
1931	4	11	9.4	7.7	-1.7	13.4	12.0	-1.5
1931	4	12	17.5	14.8	-2.7	13.5	12.0	-1.5
1931	4	13	20.6	17.4	-3.2	13.5	12.0	-1.5
1931	4	14	18.3	14.5	-3.8	13.5	12.1	-1.5
1931	4	15	14.6	12.4	-2.2	13.5	12.1	-1.5
1931	4	16	13.3	10	-3.3	13.5	12.1	-1.5
1931	4	17	10.6	9.1	-1.5	13.5	12.1	-1.5
1931	4	18	8.9	6.5	-2.4	13.5	12.1	-1.5
1931	4	19	11.7	9.4	-2.3	13.5	12.1	-1.5
1931	4	20	13.3	10	-3.3	13.5	12.1	-1.5
1931	4	21	10.7	9.1	-1.6	13.5	12.0	-1.5
1931	4	22	7.8	5.6	-2.2	13.5	12.0	-1.4
1931	4	23	8.9	6.5	-2.4	13.4	12.0	-1.4
1931	4	24	8.1	5.9	-2.2	13.4	11.9	-1.4
1931	4	25	7.8	5.6	-2.2	13.4	11.9	-1.4
1931	4	26	8.9	6.5	-2.4	13.4	11.9	-1.4
1931	4	27	10	8.3	-1.7	13.4	11.9	-1.4
1931	4	28	18	14.4	-3.6	13.4	11.9	-1.4
1931	4	29	10.6	9.1	-1.5	13.4	11.9	-1.4
1931	4	30	10.6	9.1	-1.5	13.4	11.9	-1.4
1931	5	1	8.1	5.8	-2.3	13.3	11.9	-1.4
1931	5	2	8.5	6.1	-2.4	13.3	11.9	-1.4
1931	5	3	8.4	6.1	-2.3	13.3	11.9	-1.4
1931	5	4	10.6	8.5	-2.1	13.3	11.9	-1.4
1931	5	5	17.7	9.7	-8	13.3	11.9	-1.5
1931	5	6	18.3	10.3	-8	13.3	11.9	-1.5
1931	5	7	12.2	8.8	-3.4	13.3	11.9	-1.5
1931	5	8	11.4	8.8	-2.6	13.4	11.9	-1.5
1931	5	9	11.7	8.8	-2.9	13.4	11.9	-1.5
1931	5	10	16.1	11.3	-4.8	13.4	11.9	-1.5
1931	5	11	21.7	14.8	-6.9	13.4	11.9	-1.5
1931	5	12	23.3	16.4	-6.9	13.4	11.9	-1.5
1931	5	13	20.6	13.7	-6.9	13.5	12.0	-1.5
1931	5	14	10.2	8.1	-2.1	13.5	12.0	-1.5
1931	5	15	3.9	1.9	-2	13.5	12.0	-1.5
1931	5	16	8.8	6.6	-2.2	13.5	12.0	-1.5
1931	5	17	11.1	8.8	-2.3	13.5	12.0	-1.5
1931	5	18	9.4	6.7	-2.7	13.5	12.0	-1.5
1931	5	19	12.2	8.8	-3.4	13.6	12.0	-1.5
1931	5	20	9.4	6.7	-2.7	13.6	12.0	-1.5
1931	5	21	11.7	8.8	-2.9	13.6	12.1	-1.5
1931	5	22	14.4	11	-3.4	13.6	12.1	-1.5
1931	5	23	10.6	8.5	-2.1	13.6	12.1	-1.5
1931	5	24	10	7.4	-2.6	13.7	12.1	-1.5
1931	5	25	9.4	6.7	-2.7	13.7	12.1	-1.5
1931	5	26	7.8	5.8	-2	13.7	12.1	-1.6
1931	5	27	5.1	3.4	-1.7	13.7	12.1	-1.6
1931	5	28	5.3	3.5	-1.8	13.7	12.1	-1.6
1931	5	29	2.2	0.6	-1.6	13.7	12.1	-1.6
1931	5	30	2.8	1	-1.8	13.6	12.1	-1.6
1931	5	31	6.2	5.7	-0.5	13.6	12.1	-1.5
		Avg	11.4	8.6	-2.8

Now here is December 2004, when the adjustments end. Note the 365 day running means show there is a 0.8 degree adjustment on average. (Don’t be confused by the difference- numbers are rounded to one decimal place.) Note the sudden change from a large adjustment to no adjustment on 1 January 2005. Happy New Year Richard! Your comments at Jo Nova are spot on!

Yr	Month	Day	Air	Acorn	Diff air	365d av air	Acorn	365d diff
2004	12	1	23.3	23.9	0.6	14.9	14.1	0.8
2004	12	2	11.5	12.7	1.2	14.9	14.1	0.8
2004	12	3	13	14.2	1.2	14.9	14.0	0.8
2004	12	4	20.4	20.9	0.5	14.9	14.0	0.8
2004	12	5	21.1	21.7	0.6	14.9	14.0	0.8
2004	12	6	20.6	21.1	0.5	14.9	14.1	0.8
2004	12	7	20.4	20.9	0.5	14.9	14.0	0.8
2004	12	8	15.3	16.1	0.8	14.8	14.0	0.8
2004	12	9	12.9	14.1	1.2	14.8	14.0	0.8
2004	12	10	16.5	17.5	1	14.8	14.0	0.8
2004	12	11	16.5	17.5	1	14.8	14.0	0.8
2004	12	12	17.8	18.9	1.1	14.8	14.0	0.8
2004	12	13	19.5	20.2	0.7	14.8	14.0	0.8
2004	12	14	26.1	26.8	0.7	14.8	14.0	0.8
2004	12	15	17.7	18.9	1.2	14.8	13.9	0.8
2004	12	16	21.9	22.6	0.7	14.8	13.9	0.8
2004	12	17	16.6	17.7	1.1	14.7	13.9	0.8
2004	12	18	18.8	19.8	1	14.7	13.9	0.8
2004	12	19	24.7	25.6	0.9	14.7	13.9	0.8
2004	12	20	20.9	21.4	0.5	14.7	13.9	0.8
2004	12	21	21.2	21.8	0.6	14.7	13.9	0.8
2004	12	22	22.9	23.6	0.7	14.7	13.9	0.8
2004	12	23	23	23.7	0.7	14.7	13.9	0.8
2004	12	24	20.7	21.1	0.4	14.7	13.9	0.8
2004	12	25	24.8	25.7	0.9	14.7	13.9	0.8
2004	12	26	19	20	1	14.7	13.9	0.8
2004	12	27	24.6	25.3	0.7	14.7	13.9	0.8
2004	12	28	24.2	24.9	0.7	14.7	13.9	0.8
2004	12	29	20.7	21.1	0.4	14.7	13.9	0.8
2004	12	30	12.7	13.9	1.2	14.7	13.8	0.8
2004	12	31	15.8	16.7	0.9
2005	1	1	19.4	19.4	0
2005	1	2	29.8	29.8	0
2005	1	3	24.9	24.9	0
2005	1	4	21.5	21.5	0
2005	1	5	16	16	0
2005	1	6	14.1	14.1	0
2005	1	7	15	15	0
2005	1	8	15.7	15.7	0
2005	1	9	12.8	12.8	0
2005	1	10	14.8	14.8	0

Now Rutherglen. First, winter 1913, when the largest cooling adjustments were made.

Year	Month	Day	Rutherglen	Acorn	diff
1913	6	1	3.1	2.2	-0.9
1913	6	2	6.1	4.5	-1.6
1913	6	3	10.6
1913	6	4	4.7	3.5	-1.2
1913	6	5	-2.5	-3.7	-1.2
1913	6	6	-2.2	-3.3	-1.1
1913	6	7	-3.1	-4.9	-1.8
1913	6	8	-1.9	-3	-1.1
1913	6	9	-2.5	-3.7	-1.2
1913	6	10	-3.3	-5.5	-2.2
1913	6	11	-3.3	-5.5	-2.2
1913	6	12	-3.3	-5.5	-2.2
1913	6	13	-2.2	-3.3	-1.1
1913	6	14	-1.1	-2.1	-1
1913	6	15	1.1	0.1	-1
1913	6	16	0.3	-1	-1.3
1913	6	17	3.9	2.8	-1.1
1913	6	18	-2.8	-4.4	-1.6
1913	6	19	2.2	1.5	-0.7
1913	6	20	3.9	2.8	-1.1
1913	6	21	-1.4	-2.6	-1.2
1913	6	22	1.1	0.1	-1
1913	6	23	6.7	5.1	-1.6
1913	6	24	8.9	6.3	-2.6
1913	6	25	8.9	6.3	-2.6
1913	6	26	2.2	1.5	-0.7
1913	6	27	-1.9	-3	-1.1
1913	6	28	-2.2	-3.3	-1.1
1913	6	29	-1.1	-2.1	-1
1913	6	30	3.9	2.8	-1.1
1913	7	1	1.7	1.1	-0.6
1913	7	2	0.6	-0.3	-0.9
1913	7	3	-3.1	-5.4	-2.3
1913	7	4	-4.2	-6.7	-2.5
1913	7	5	4.4	3	-1.4
1913	7	6	4.4	3	-1.4
1913	7	7	1.9	1.2	-0.7
1913	7	8	2.2	1.4	-0.8
1913	7	9	-1.1	-2.7	-1.6
1913	7	10	-3.9	-6.4	-2.5
1913	7	11	-1.4	-3.1	-1.7
1913	7	12	2.2	1.4	-0.8
1913	7	13	3.6	2.5	-1.1
1913	7	14	1.9	1.2	-0.7
1913	7	15	5	3.5	-1.5
1913	7	16	7.2	5.3	-1.9
1913	7	17	4.7	3.2	-1.5
1913	7	18	6.9	5.1	-1.8
1913	7	19	6.9	5.1	-1.8
1913	7	20	3.9	2.7	-1.2
1913	7	21	3.9	2.7	-1.2
1913	7	22	-1.7	-3.3	-1.6
1913	7	23	-2.6	-4.5	-1.9
1913	7	24	-2.4	-4.4	-2
1913	7	25	-3.9	-6.4	-2.5
1913	7	26	-2.1	-3.9	-1.8
1913	7	27	1.5	0.9	-0.6
1913	7	28	3.3	2.2	-1.1
1913	7	29	3.4	2.4	-1
1913	7	30	7.3	5.5	-1.8
1913	7	31	8.3	6.7	-1.6
		AVERAGE	1.50	-0.07	-1.42

Now 1973, when the adjustments end. Note the 365 day average shows an adjustment of -0.5, which ends suddenly on New Year’s Day again.

Year	Month	Day	Rutherglen	Acorn	diff	365d av	Acorn	365d diff
1973	11	1	10.5	9.9	-0.6	8.5	8.1	-0.5
1973	11	2	9.5	9.1	-0.4	8.6	8.1	-0.5
1973	11	3	6	5.6	-0.4	8.6	8.1	-0.5
1973	11	4	10	9.4	-0.6	8.5	8.1	-0.5
1973	11	5	10	9.4	-0.6	8.6	8.1	-0.5
1973	11	6				8.6	8.1	-0.5
1973	11	7	5.5			8.5	8.1	-0.5
1973	11	8	7	6.8	-0.2	8.6	8.1	-0.5
1973	11	9	12	11.9	-0.1	8.6	8.1	-0.5
1973	11	10	15.5	14.9	-0.6	8.6	8.1	-0.5
1973	11	11	7	6.8	-0.2	8.6	8.1	-0.5
1973	11	12	8.5	7.6	-0.9	8.6	8.1	-0.5
1973	11	13	6.5	6	-0.5	8.6	8.1	-0.5
1973	11	14	5	4.7	-0.3	8.6	8.1	-0.5
1973	11	15	6.5	6	-0.5	8.6	8.1	-0.5
1973	11	16	5	4.7	-0.3	8.6	8.1	-0.5
1973	11	17	7.5	7	-0.5	8.6	8.1	-0.5
1973	11	18	13	12.6	-0.4	8.6	8.1	-0.5
1973	11	19	12	11.9	-0.1	8.6	8.1	-0.5
1973	11	20	19.5	19	-0.5	8.6	8.1	-0.5
1973	11	21	5.5	5.2	-0.3	8.6	8.1	-0.5
1973	11	22	5	4.7	-0.3	8.6	8.1	-0.5
1973	11	23	6	5.6	-0.4	8.6	8.1	-0.5
1973	11	24	7.6	7.1	-0.5	8.6	8.1	-0.5
1973	11	25	8.5	7.6	-0.9	8.6	8.0	-0.5
1973	11	26	9.5	9.1	-0.4	8.6	8.0	-0.5
1973	11	27	13.5	13	-0.5	8.6	8.1	-0.5
1973	11	28	12.5	12.2	-0.3	8.6	8.1	-0.5
1973	11	29	10	9.4	-0.6	8.6	8.1	-0.5
1973	11	30	10	9.4	-0.6	8.6	8.1	-0.5
1973	12	1	10	9.4	-0.6	8.6	8.1	-0.5
1973	12	2	21	20.5	-0.5	8.6	8.1	-0.5
1973	12	3	10.5	9.7	-0.8	8.6	8.1	-0.5
1973	12	4	11	10.1	-0.9	8.6	8.1	-0.5
1973	12	5	6	5.4	-0.6	8.6	8.1	-0.5
1973	12	6	7	6.3	-0.7	8.6	8.1	-0.5
1973	12	7	11	10.1	-0.9	8.6	8.1	-0.5
1973	12	8	10	9.4	-0.6	8.6	8.1	-0.5
1973	12	9	10	9.4	-0.6	8.6	8.1	-0.5
1973	12	10	14	13.2	-0.8	8.6	8.1	-0.5
1973	12	11	19	18.3	-0.7	8.6	8.1	-0.5
1973	12	12	16.5	15.9	-0.6	8.6	8.1	-0.5
1973	12	13	16	15.4	-0.6	8.6	8.1	-0.5
1973	12	14	21	20.5	-0.5	8.7	8.1	-0.5
1973	12	15	18	16.9	-1.1	8.7	8.1	-0.5
1973	12	16	15	14.4	-0.6	8.7	8.1	-0.5
1973	12	17	8.5	8	-0.5	8.6	8.1	-0.5
1973	12	18	11	10.1	-0.9	8.6	8.1	-0.5
1973	12	19	9	8.5	-0.5	8.6	8.1	-0.5
1973	12	20	11.5	10.9	-0.6	8.6	8.1	-0.5
1973	12	21	17.5	16.8	-0.7	8.6	8.1	-0.5
1973	12	22	9	8.5	-0.5	8.6	8.1	-0.5
1973	12	23	5.4	4.8	-0.6	8.5	8.0	-0.5
1973	12	24	7	6.3	-0.7	8.5	8.0	-0.5
1973	12	25	13.8	13.1	-0.7	8.5	8.0	-0.5
1973	12	26	7.5	6.9	-0.6	8.5	8.0	-0.5
1973	12	27	10.5	9.7	-0.8	8.5	8.0	-0.5
1973	12	28	15.5	14.8	-0.7	8.5	8.0	-0.5
1973	12	29	14	13.2	-0.8	8.5	8.0	-0.5
1973	12	30	18	16.9	-1.1	8.5	8.0	-0.5
1973	12	31	10	9.4	-0.6	8.5	8.0	-0.5
1974	1	1	10	10	0
1974	1	2	10.5	10.5	0
1974	1	3	15	15	0
1974	1	4	20	20	0
1974	1	5	17.5	17.5	0
1974	1	6	12	12	0
1974	1	7	14.5	14.5	0
1974	1	8	12.3	12.3	0
1974	1	9	12.5	12.5	0
1974	1	10	18.3	18.3	0
1974	1	11	17.7	17.7	0

		AVERAGE	10.81	10.33	-0.57

Finally, Brisbane Aero. A commenter claimed the 1960s were very cold in Brisbane, and it is much warmer now. Note what Acorn has done to the data.

The raw data consists of the old Regional Office data, which appears affected by UHI, and the Old Airport (Eagle Farm), which is several kilometres closer to Moreton Bay. The new Aero site is several kilometres closer to the Bay again.

Year	Mon	Day	RO	Air old	Acorn	diff	RO 365d	Old	Acorn	old diff
1964	6	1	13.6	13.5	13.7	0.2	15.7	15.6	14.7	-1.0
1964	6	2	9.4	9.3	7.3	-2	15.7	15.6	14.7	-1.0
1964	6	3	12.9	12.8	12.5	-0.3	15.7	15.6	14.7	-1.0
1964	6	4	13.4	13.6	13.7	0.1	15.7	15.6	14.6	-1.0
1964	6	5	10.9	11	9.7	-1.3	15.7	15.6	14.6	-1.0
1964	6	6	12.8	13.5	13.7	0.2	15.7	15.6	14.6	-1.0
1964	6	7	13.7	12.3	11.5	-0.8	15.7	15.6	14.6	-1.0
1964	6	8	14.1	14.3	14.7	0.4	15.7	15.6	14.6	-1.0
1964	6	9	11.7	12.1	11.1	-1	15.7	15.6	14.7	-1.0
1964	6	10	11.4	11.5	10.2	-1.3	15.7	15.6	14.7	-1.0
1964	6	11	9.6	7.9	5.1	-2.8	15.7	15.6	14.7	-1.0
1964	6	12	11.1	9.4	7.4	-2	15.7	15.6	14.6	-1.0
1964	6	13	13.4	12	11	-1	15.7	15.6	14.6	-1.0
1964	6	14	8.6	8.7	6.1	-2.6	15.7	15.6	14.6	-1.0
1964	6	15	8.8	9.6	7.9	-1.7	15.7	15.6	14.6	-1.0
1964	6	16	8.4	9.4	7.4	-2	15.7	15.6	14.6	-1.0
1964	6	17	8	8.8	6.4	-2.4	15.7	15.6	14.6	-1.0
1964	6	18	11.3	11.2	9.9	-1.3	15.7	15.6	14.6	-1.0
1964	6	19	13.9	13.9	14.1	0.2	15.7	15.6	14.6	-1.0
1964	6	20	13.1	11.6	10.5	-1.1	15.7	15.6	14.6	-1.0
1964	6	21	10.7	7.8	5	-2.8	15.7	15.6	14.6	-1.0
1964	6	22	9.9	8.3	5.3	-3	15.7	15.6	14.6	-1.0
1964	6	23	9.6	8.7	6.1	-2.6	15.7	15.6	14.6	-1.0
1964	6	24	11.6	8.9	6.7	-2.2	15.7	15.6	14.6	-1.0
1964	6	25	7.4	7.8	5	-2.8	15.7	15.6	14.6	-1.0
1964	6	26	9.7	9.8	8.2	-1.6	15.7	15.6	14.6	-1.0
1964	6	27	9.1	8.7	6.1	-2.6	15.7	15.6	14.6	-1.0
1964	6	28	5.5	6.2	3	-3.2	15.7	15.6	14.5	-1.0
1964	6	29	7.7	8.3	5.3	-3	15.7	15.5	14.5	-1.0
1964	6	30	11.8	11.7	10.5	-1.2	15.7	15.5	14.5	-1.0
1964	7	1	11.7	11.3	10.8	-0.5	15.7	15.5	14.5	-1.0
1964	7	2	13.9	13.3	14	0.7	15.7	15.6	14.5	-1.0
1964	7	3	15.6	15.7	16.2	0.5	15.7	15.6	14.6	-1.0
1964	7	4	10.8	10.4	9.2	-1.2	15.7	15.6	14.6	-1.0
1964	7	5	6.3	4.4	0.5	-3.9	15.7	15.6	14.6	-1.0
1964	7	6	5.5	6.2	2.5	-3.7	15.7	15.6	14.6	-1.0
1964	7	7	8.7	8.8	7.4	-1.4	15.7	15.6	14.5	-1.0
1964	7	8	6.3	7.3	4.9	-2.4	15.7	15.6	14.5	-1.0
1964	7	9	6.3	7.1	4.6	-2.5	15.7	15.6	14.5	-1.0
1964	7	10	8.8	8.9	7.4	-1.5	15.7	15.6	14.5	-1.0
1964	7	11	11.5	12.1	12.1	0	15.7	15.6	14.5	-1.0
1964	7	12	12.5	9.2	7.8	-1.4	15.7	15.6	14.6	-1.0
1964	7	13	16.7	15.1	15.6	0.5	15.7	15.6	14.6	-1.0
1964	7	14	17.4	14.4	14.9	0.5	15.7	15.6	14.6	-1.0
1964	7	15	11.3	10.9	10	-0.9	15.7	15.6	14.6	-1.0
1964	7	16	10.2	8.9	7.4	-1.5	15.8	15.6	14.6	-1.0
1964	7	17	11.7	11.7	11.5	-0.2	15.8	15.6	14.7	-1.0
1964	7	18	13.5	13.6	14.2	0.6	15.8	15.7	14.7	-1.0
1964	7	19	10.3	8.9	7.4	-1.5	15.8	15.7	14.7	-1.0
1964	7	20	12	12.2	12.2	0	15.8	15.7	14.7	-1.0
1964	7	21	8.9	6.5	2.7	-3.8	15.8	15.7	14.8	-1.0
1964	7	22	7.6	5.8	2.2	-3.6	15.8	15.7	14.7	-1.0
1964	7	23	6.7	5.6	1.9	-3.7	15.8	15.7	14.7	-1.0
1964	7	24	4.9	5.6	1.9	-3.7	15.8	15.7	14.7	-1.0
1964	7	25	6.4	6.8	3.2	-3.6	15.8	15.7	14.7	-1.0
1964	7	26	9.1	8.7	6.8	-1.9	15.8	15.7	14.7	-1.0
1964	7	27	8.4	8.7	6.8	-1.9	15.8	15.7	14.7	-1.0
1964	7	28	8.8	9.4	7.9	-1.5	15.8	15.7	14.7	-1.0
1964	7	29	10.2	10.5	9.6	-0.9	15.8	15.6	14.6	-1.0
1964	7	30	10.2	9.9	8.6	-1.3	15.8	15.7	14.7	-1.0
1964	7	31	9.7	10.4	9.2	-1.2	15.8	15.7	14.7	-1.0
1964	8	1	11.5	11.1	9.8	-1.3	15.8	15.7	14.7	-1.0
1964	8	2	7.3	6.7	3.3	-3.4	15.8	15.7	14.7	-1.0
1964	8	3	7.7	7.8	5.3	-2.5	15.8	15.7	14.7	-1.0
1964	8	4	10.8	7.7	5	-2.7	15.8	15.7	14.6	-1.0
1964	8	5	9.6	6.3	2.9	-3.4	15.8	15.6	14.6	-1.0
1964	8	6	12.6	11	9.6	-1.4	15.8	15.6	14.6	-1.0
1964	8	7	11.7	12.5	12.5	0	15.8	15.6	14.6	-1.0
1964	8	8	8.7	9.2	7.2	-2	15.8	15.6	14.6	-1.0
1964	8	9	8.7	8.2	5.5	-2.7	15.7	15.6	14.6	-1.0
1964	8	10	5.9	5.4	2	-3.4	15.7	15.6	14.6	-1.0
1964	8	11	10.2	9.5	8	-1.5	15.7	15.6	14.6	-1.0
1964	8	12	8.6	7.3	4	-3.3	15.7	15.6	14.5	-1.0
1964	8	13	2.7	3.3	-0.1	-3.4	15.7	15.6	14.5	-1.1
1964	8	14	8.1	9.4	8	-1.4	15.7	15.6	14.5	-1.1
1964	8	15	13.2	13.6	14	0.4	15.7	15.6	14.5	-1.1
1964	8	16	8.6	6.7	3.3	-3.4	15.7	15.6	14.5	-1.1
1964	8	17	9.8	8.3	5.7	-2.6	15.7	15.6	14.5	-1.1
1964	8	18	6.2	6.3	2.9	-3.4	15.7	15.5	14.5	-1.1
1964	8	19	7.6	7.8	5.3	-2.5	15.7	15.5	14.5	-1.1
1964	8	20	8.4	9.3	7.4	-1.9	15.6	15.5	14.4	-1.1
		Avg.	10.04	9.61	7.88	-1.73

Note the -0.1 in August 1964! The average cooling adjustment is -1.73 degrees Celsius.

The lowest temperature for the new Airport was -0.1 on 19/07/2007, the only time the raw temperature for Brisbane has been below zero. Acorn has several below zero temperatures in the 1960s, and has the record low as -2.2 on 22/07/1951, an adjustment of -4 degrees.

The Percentile Matching algorithm may, in theory, be an excellent method of homogenizing temperatures. However its effect in practice is to garble our temperature record, to make a nonsense of it.

We’re being conned.

This entry was posted on June 23, 2012 at 11:03 am and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

15 Responses to “Acorn or A Con: Acorn’s Daily Adjustments”

Richard C (NZ) Says:
June 23, 2012 at 1:28 pm
Thanks for tabulating these adjustments Ken, I’m still focussed on Alice Springs Min so I’ll confine my comment to there for now.

You say:-

Note that Acorn bears little resemblance to either of the raw data records.

I think this is because the daily adjustments are carried out AFTER adjusting for inhomogeneities (breakpoints, 345 breakpoint steps in Alice Springs Min) although I can’t be sure of that because I can’t find a specific note in CAWCR Technical Report No. 049 by Blair Trewin (Trewin049).

Breakpoint adjustments result in steps as plotted in Trewin049 Fig 29 page 96 pdf whereby ranges of data are adjusted up or down due to site moves of seasonal variations (I have a problem with the latter, see below) with the same adjustment value being applied to each datapoint in the range. Then I think the individual adjustments are made that give the final values you have tabulated although I will have to do more reading to confirm this or otherwise because it’s difficult to work out BOM’s adjustment sequence from 7.3 Adjustment of data to remove inhomogeneities – an overview or 7.7 Implementation of data adjustment in the ACORN-SAT data set, meantime Trewin049 says in 7.7 page 71 pdf:-

After the first round of homogenisation, the homogenised data sets were evaluated, using the following tools:

This indicates there were “rounds” in the adjustment sequence but where the daily adjustments fit in in is unclear. Reference to “daily adjustments” can be found by searching the pdf but there’s nothing describing how BOM applied daily adjustments. All of the implementation notes seem to describe application of step change adjustments.

As I’ve commented elsewhere, I no longer think the daily adjustments have made the radical change to the Alice Springs Min trend but that it is the step changes that have done so and in particular, BOM’s very questionable practice of what I see as adjusting out local climate. I left this comment at JoNova:-

An important distinction.

The 1st +0.8 C Alice Springs Min step at 2004/05 is due to CLIMATE (seasonal vegetation change)

The 2nd -0.3 C Alice Springs Min step at 2001/02 is due to SITE MOVE (“a few metres west”).

Therefore, BOM has adjusted local climate OUT of ACORN – SAT. This is unnecessary, NIWA did not do that in the NZT7.

The -0.3 site move step at 2001/02 is perfectly acceptable however.

http://joannenova.com.au/2012/06/threat-of-anao-audit-means-australias-bom-throws-out-temperature-set-starts-again-gets-same-results/#comment-1072495

From what I recall without looking up the numbers, a 0.3 adjustment is consistent with other AWS site moves and not worth quibbling over. BOM doesn’t make step adjustments for anything less than 0.3.

Adjusting for weather (climate) is a different situation altogether – why do it? Trewin049 discusses weather dependent inhomogeneities in 7.3 Adjustment of data to remove inhomogeneities – an
overview on page 58 pdf and have gone ahead with it gung-ho but without the necessary conclusive or rational justification that I can see. This has resulted in an inordinately large number of step changes (as above, 345 breakpoint steps in Alice Springs Min). This is literally hundreds more step adjustments than were made by NIWA in the NZT7 for any one location series, typical would be 5 to 10 I think offhand.

You seem to have isolated the +0.8 step change made to the 2002 – 2004 range of 365d datapoints in the 365d diff column of the 2004/05 Alice Springs table although I wonder if you have the headings “Acorn” and “365d av air” transposed?

Surely the 2004/12/1 365d difference would be 14.9 (Acorn) – 14.1 (365d av air) = 0.8 because the ACORN data has had 0.8 ADDED TO IT over that range to adjust for the step change?

My calculation to isolate the daily adjustment to 2004/12/1 would be 23.9 (ACORN) – 0.8 (step adj) = 23.1 (step removed). 23.1 – 23.3 (Air) = -0.2. This gives a NET step/daily adjustment to 2004/12/1 of +0.6 (+0.8 + -0.2 = +0.6) and tallies with your “Diff air” value of +0.6.

Similarly for 2004/12/2 the NET is:-
12.7 – 0.8 = 11.9
11.9 – 11.5 = 0.4
0.8 + 0.4 = 1.2

This tallies with your “Diff air” value of 1.2

In both cases the step adjustment is much greater than the daily adjustment and would therefore have a greater effect on a linear trend.

2005 and 31 Dec 2004 remain unchanged because 2005 is the last “open” site used as the “reference” benchmark for step change adjustments. I dispute the justification for this step however as already outlined.

I look forward to what you make of my analysis – agreement or otherwise.

Cheers

Rich.
Richard C (NZ) Says:
June 23, 2012 at 1:34 pm
Dang, 3rd para should read:-

“data are adjusted up or down due to site moves [or] seasonal variations
Richard C (NZ) Says:
June 23, 2012 at 1:44 pm
Dang again, last para should read:-

“I dispute the justification for [the first] step however as already outlined”
Richard C (NZ) Says:
June 23, 2012 at 2:26 pm
Ken, I haven’t seen any reply from you to the series of comments I sprayed around both at the previous ACORN post here and the one at JoNova.

I am particularly interested to know if I’ve managed to change your position wrt step change methodology (“accumulated” or “cumulative”) i.e. do you now see that BOM have in fact applied step changes beginning with my detail of the methodology here:-

http://joannenova.com.au/2012/06/threat-of-anao-audit-means-australias-bom-throws-out-temperature-set-starts-again-gets-same-results/#comment-1071405

Here

http://joannenova.com.au/2012/06/threat-of-anao-audit-means-australias-bom-throws-out-temperature-set-starts-again-gets-same-results/#comment-1071936

Here

http://joannenova.com.au/2012/06/threat-of-anao-audit-means-australias-bom-throws-out-temperature-set-starts-again-gets-same-results/#comment-1071983

And up-thread from all that here

http://joannenova.com.au/2012/06/threat-of-anao-audit-means-australias-bom-throws-out-temperature-set-starts-again-gets-same-results/#comment-1071363

I would appreciate your reply at JoNova for those comments on methodology beyond anything brief leaving this forum free for exploration of BOM’s actual application of methods and adjustments (but this is your blog so as you wish).
- kenskingdom Says:
  June 27, 2012 at 11:39 am
  Gday Richard
  
  My apologies for not following up on your many detailed comments. I’ve been away for a week and except for a half hour window haven’t had internet access. I’ll be away again from tomorrow for a few days. A pretty ordinary way to run a blog but I have a life too.
  
  Your suggestion that Acorn results from 2 adjustments- step change and then daily- though not documented in Trewin’s paper seems implicit from what he says. I do agree with you that adjustments were made from step points, but these varied for each season and frequently for each month. Rutherglen shows this clearly. Moreover, the Percentile Matching algorithm appears to have made DAILY adjustments which average out over a year or a season to be the amount of the step change adjustment. Therefore, there are no step changes as such (except at the beginning and end of adjustment periods) visible in Acorn at the daily level: it is only by 365 day smoothing that I was able to see what was going on. The sudden ceasing of the adjustments from points such as January 1 2005 is ludicrous but shows they were adjusting from breakpoints.
  
  My focus has been to show that Acorn is messy and full of errors or at least spurious data; and the adjustments greatly affect climate trends, and, as you say, edit out real climate. I am fast losing interest in surface temperatures as they don’t tell us much except at a local or regional level, and the Australian record is so poor as to be embarassing. Acorn was too obviously poor to ignore however.
  
  I’ve fixed the column headings- thanks for pointing out that pretty obvious error!
  
  Ken
  - thefutureshere Says:
    June 27, 2012 at 1:47 pm
    Ken, thanks for responding. I was becoming concerned that you had vanished off the planet.
    
    I hope you don’t throw Acorn out for the wrong reasons because unless, as I’ve said elsewhere, you can back up your claims statistically as NZCSC has with the NZT7 you wont get much traction with people that matter.
    
    I can’t check the site move steps statistically because I don’t have the expertise and neither did NZCSC. They engaged the services of a specialist professional statistician. I suspect disputing BOM’s ACORN site move steps would require that expertise and as the ‘Statistical Audit of the New Zealand Temperature Series’ shows, there’s a lot of exacting work that has to be done. It may even be that the site move steps check out OK.
    
    I’m still concerned that you haven’t got to grips with step change methodology going by your comment, especially this dead giveaway:-
    
    The sudden ceasing of the adjustments from points such as January 1 2005 is ludicrous but shows they were adjusting from breakpoints
    
    BOM have done exactly as NIWA has done in most of the NZT7 locations using conventional method, that is: the last open site (or a recently open one) becomes the location “reference” site and all other sites are adjusted into terms with the reference site using cumulative steps. Trewin touches on this in CAWCR Technical Report No. 049 page 57 pdf:-
    
    7.3 Adjustment of data to remove inhomogeneities – an
    overview
    
    Once potential inhomogeneities have been identified, the next step is to adjust the data to remove the effects of the inhomogeneity (normally by adjusting data prior to the inhomogeneity to make it homogeneous with the most recent data, although the reverse is also possible) and make the data set homogeneous. The practice of homogenising to the most recent data has clear advantages for ongoing monitoring as it allows new data to be simply appended to the location time series (until such time as the next inhomogeneity occurs).
    
    Airport AWS January 1 2005 – present is the Alice Springs location reference site to which ALL other prior sites are adjusted into terms with. The adjustment process employed moves backwards in time from the reference site and forwards in time from it as data is added to the series. Therefore, Airport AWS post Jan 1 2005 until the next step (for whatever reason arises) will remain unadjusted for posterity and the raw data for it will be EXACTLY the same as the final series data. There is nothing untoward or ludicrous about that.
    
    The weather dependent steps are no different in terms of cumulative step change but the justification for them is another question entirely. I would have thought weather and seasons is exactly what should be recorded because the attendant temperature, rainfall, humidity, wind changes make grass and vegetation grow or recede all over the world. To adjust those breakpoints out is daft IMO and distorts the series unnecessarily.
    
    BTW I had a think about the number of steps in the ACORN 112 station series (max + min = 660) and if you say the series is just mean temps of 660 steps this works out at just under 6 steps per location on average. That’s about on par with the NZT7 and not “inordinately large” as I first thought.
    - thefutureshere Says:
      June 27, 2012 at 1:50 pm
      “thefutureshere” is Richard C (NZ). I forgot to change my login.
    - kenskingdom Says:
      June 27, 2012 at 3:06 pm
      Gday Richard
      
      You might like to check out my understanding of step changes at the annual time frame in my previous posts re HQ data (see ‘The Australian Temperature Record” Parts 1 – 8 for example). I understand that 2005 to now is the reference period, what I find ludicrous is that they didn’t taper off the adjustments during 2004 but continued right up to 31/12, so there is an abrupt change rather than a tapering off.
      Not much hope of statistical analysis until we have access to the full data, comparison sites, adjustment explanations etc in a few months time.
      Ken
      - Richard C (NZ) Says:
        June 27, 2012 at 5:44 pm
        OK Ken, some points. You say:-
        
        You might like to check out my understanding of step changes at the annual time frame in my previous posts re HQ data (see ‘The Australian Temperature Record” Parts 1 – 8 for example)
        
        I’ve looked at 1 – 3 and see nothing to indicate you are any where near where NZCSC got to. A group of us (myself non-affiliated) got together by email and compiled NZT7 location series case studies from scratch every way we could think of, both backward and forward in time to check step change methodology and the actual steps. That is the scientific test: given the raw data and the established method, reconstruct what NIWA produced. We were able to generally but it became obvious that a statistician was required to check some specific steps that were disputable in the early part of a number of locations thereby creating a composite trend 0.56 C/century too steep in view of the statistical audit (0.9 vs 0.34 C/century). Such an ACORN audit will have to wait as you point out.
        
        We did NOT however, do as you did and splice together disparate site datasets to create unadjusted “raw” location series with which to make comparisons. That is unscientific and any trend derived from such a series has no validity whatsoever. Richard Treadgold at CCG did so a few times a couple of years ago but I think by now his realization is that doing so is not the correct approach (I hope). He was not part of the group but did follow the email exchange. I have taken Ian Bryce to task at JN for doing the same thing at Echuca (among other things) but he has not responded with any substance.
        
        Our group comparison was: NIWA steps where they used remote comparators vs group steps using neighbouring comparators as per Rhoades and Salinger 1993 methodology that NIWA promised to use but did not, both using cumulative step change methodology.
        
        I understand that 2005 to now is the reference period
        
        OK, that’s settled.
        
        what I find ludicrous is that they didn’t taper off the adjustments during 2004 but continued right up to 31/12, so there is an abrupt change rather than a tapering off
        
        This is the weather dependent step so I find it ludicrous that there’s a step at all. If a weather event created a break well so be it – record it, it’s weather, it’s seasons, it’s climate. Even if there is a good reason for the step I don’t see the need to “taper” the pre-break data at all because the nature of a break is that it is abrupt. Trewin 7.2:-
        
        Since the ability to detect a breakpoint in a time series is a function of the ratio of the size of the breakpoint to the standard deviation of the data,
        
        What concerns me most is that the weather dependent step is a significant (to my mind – Trewin doesn’t think so) +0.8 C. By comparison, the next step along is -0.3 C for a site move. BOM doesn’t adjust for steps less than 0.3 C. That +0.8 being at the end of the series has an effect on the slope of a linear regression. The same step in the middle of the series effects the y intersept more than the slope I think.
        
        My other bone-of contention is that linear regression is the wrong trend method for fluctuating data anyway but ‘nuther story. A best fit polynomial or moving average/Fourier is appropriate for fluctuations, linear for data resembling a line. I’ve taken Ian Bryce to task at JN also over this but no response.
Richard C (NZ) Says:
June 27, 2012 at 6:38 pm
….from scratch every way we could think of, both backward and forward in time to check step change methodology and the actual steps.

We spent some time initially just averaging raw site data over various spans either side of site moves and finding the difference as a quick and dirty way to get a feel for whether a NIWA step was valid or not.

You don’t need comparator locations for that nor do you need the middle raw data between steps at each site to compile an adjusted series. That way you’re focused on the steps and we plotted several step change charts like Fig 29 in Trewin049 (or least I did, not sure about the others).
kenskingdom Says:
June 27, 2012 at 8:07 pm
Gday Richard

Last comment as I have to go off and do other things for a few days.
I see nothing at all wrong with splicing station data to make a continuous series: this is the only defensible adjustment in my opinion. Provided there is sufficient overlap (which BOM neglect) very close neighbours can be compared and combined. Not perfect, but good enough for Torok and Nichols, Della-Marta, and others, including me. From your description I followed a similar process to you in looking at adjustments, including looking at diurnal range. Here’s another link: https://kenskingdom.wordpress.com/2011/10/24/the-australian-temperature-record-part-10-bom%e2%80%99s-%e2%80%9cexplanations%e2%80%9d/

Richard, I’ve really got better things to do so I’m going to leave it at that .

Ken
- Richard C (NZ) Says:
  June 28, 2012 at 9:34 am
  I understand your time constraints Ken. I’m off into long night shifts next week so I’ll be in the same boat. We’ll probably have to agree to disagree but here’s my responses.
  
  I see nothing at all wrong with splicing station data to make a continuous series: this is the only defensible adjustment in my opinion.
  
  If you are referring to raw data, neither do I. And neither does BOM, they only adjust for steps greater than or equal to 0.3. But it is indefensible NOT to adjust for larger steps when those steps can be justified by established methodology. What you have then is disparate data sets otherwise.
  
  Provided there is sufficient overlap (which BOM neglect) very close neighbours can be compared and combined.
  
  Exactly, and that is where the dispute arises in the NZT7 and should also in ACORN if it is found that the size of the step resulting from BOM’s (or NIWA’s) procedure of overlap, neighbour vs remote, statistical analysis, deference to established literature etc is outside what can be determined by independent reconstruction. That is the essence of science: to be able to reproduce another’s (or even your own) results using the established method. If the results cannot be reproduced (or even can be with random data as in Mann’s hockey stick) questions must be asked.
  
  BTW, by “reproduce” I know what I’m talking about. Among other things I’ve spent time as a materials testing QC using enclosed electronic weighing instruments that can take 5 – 10 minutes to settle at an accuracy of 9 decimal places of a gram. All the lab windows and doors must be closed to prevent draughts. To reproduce results in that environment means following procedures to the letter and especially when results outside the specification envelope mean failure of a batch and $$$ lost by the organization providing the sample. Your procedure has to stand up to the inevitable questions that will be asked.
  
  NSCSC are requesting a judge to decide that those questions wrt the NZT7 be asked in the form of a judicial review.
  - Richard C (NZ) Says:
    June 28, 2012 at 10:12 am
    I see you have investigated HQ step adjustments extensively in The Australian Temperature Record- Part 10: BOM’s “Explanations”. I would have had a much better understanding of your position if I’d seen that first.That’s some very dodgy stuff, especially this re reproducing results:-
    
    Della-Marta et al go on to lament that it was impossible to reproduce exactly Torok and Nichols’ adjustments, as slightly different techniques, reference stations, and source data “can apparently produce different results”. It seems they were surprised by this.
    
    I’ve got that post bookmarked for future reference, thanks for the pointer.
    
    BOM seems to have changed their criteria for ACORN (e.g. no 0.1 step adjustments). Possibly site moves are given better treatment in ACORN (?) but the raft of weather dependent breaks seem to be the overriding problem common to both HQ and ACORN.
    
    Missing data and UHI highly questionable too (as you have).
    
    But as I’ve been banging on, use polynomial trends and the linear “warming” is of no consequence, especially now that 21st century temperatures are flat or cooling.
ACORN-SAT 2.0: Western Australia- A State of Confusion | kenskingdom Says:
February 14, 2019 at 11:31 am
[…] with others) found to have very many severe problems. (If you like, check these posts, here, here, here, and here. There are many […]
ACORN-SAT 2.0: Nation-wide Summary | kenskingdom Says:
May 20, 2019 at 7:23 pm
[…] with others) found to have very many severe problems. (If you like, check these posts, here, here, here, and here. There are many […]

Comments are closed.

kenskingdom

Acorn or A Con: Acorn’s Daily Adjustments

Share this:

Related

15 Responses to “Acorn or A Con: Acorn’s Daily Adjustments”