7a. Larrimah, NT

Is homogenisation of Australian temperature data any good?

Part 7a. Larrimah, Northern Territory, Australia

Bureau of Meteorology ID 014612. Located 430 km SE of Darwin at Latitude -15.5748 Longitude 133.2137. Maximum temperature data from 1 January 1965 to 29 June 2012.

Dr Bill Johnston[1]


Data quality is poor. The use of faulty data that are not homogeneous, to adjust faults in ACORN-SAT data is unscientific and likely to result in biased outcomes. As the ACORN-SAT project does not ensure comparator datasets are homogeneous, it is deeply flawed and should be abandoned. Read on …   


While data quality is poor, accounting for rainfall and the step-change simultaneously using multiple linear regression left no additional trend or change that could be attributed to the climate, CO2, coalmining or anything else.

1.    Background

Situated on the Stuart Highway 430 km southeast of Darwin (Figure 1) maximum temperature (Tmax) observations at Larrimah ceased on 29 June 2012. However, Larrimah Tmax was used by ACORN-SAT (the Australian Climate Observations Reference Network – Surface Air Temperature project) to homogenise Tmax data at Victoria River Downs (1/1/1976 and 1/8/1987, AcV1 and 1/1/2007, AcV2.x) and Burketown airport (1/1/2002, AcV1 and 1/1/1986, AcV2.x).

Figure 1. Larimah Tmax was used to homogenise Tmax data for Victoria River Downs (256km away), and Burketown airport (762 km distant). ACORN-SAT sites are indicated by red buttons, and sites having more than 10-years of data, by grey circles.

This raises the question: is Tmax data for Larrimah any good?

2.    Methods

Daily temperature and monthly rainfall were downloaded from the Bureau of Meteorology, Climate Data Online facility (Climate Data Online – Map search (bom.gov.au)). Data were summarised into an annual dataset and analysed using the same BomWatch protocols described previously in the Parafield case study[2] and subsequent reports, including for Victoria River Downs[3] of which this report is a subset.

Only maximum temperature (Tmax) data are used in the study.

3.    Results

Ignoring outliers (red squares in Figure 2(a)), rainfall, which is the deterministic portion of the Tmax signal explains 39.5% of Tmax variation (Radj = 0.395). As this is less than the R2adj BomWatch benchmark of 0.50, either data quality is exceedingly poor, or an influential variable has not been accounted for by the naïve Tmax ~ rainfall case (Table 1(i)).

(Note: R2adj calculated by the statistical package R, adjusts variation explained for the number of terms in the linear model, as well as for the number of observations[4]. It is therefore more robust (less biased) than unadjusted R2 calculated by spreadsheet programs such as Excel.)  

Rescaled so values are comparable, Tmax ~ rainfall residuals were evaluated for inhomogeneities using STARS, which objectively tests whether the mean of subsequent data is significantly different (P<0.05) to that before, using a t-test of the difference. Indicated by the horizontal line in Figure 2(b), STARS detected an up-step of 0.41oC in 1986 (= 0.015). Segments defined by the step-change were examined separately in Figure 2(c) and Figure 2(d).

The spread of points about each line, and influence plots (not shown) confirmed that data for 1982 and 2011 were likely outliers. Also, as R2adj was <0.50 data quality was generally poor.

Categorical multiple linear regression of the form Tmax ~ Sh(ift)factor + Rainfall showed rainfall reduces Tmax 0.174oC/100mm, and that segmented regressions were offset by a rainfall adjusted difference of 0.41oC (Figure 2(e)), which is the same as that detected by STARS. Post hoc tests confirmed that data consisted of two non-trending segments interrupted by the step-change in 1986 (Table 1(iv)).

Figure 2. Composite analysis of Larrimah Tmax. A statistical summary for each phase of the analysis is provide in Table 1.

4.     Discussion and conclusions

Diagrams in site-summary metadata for 1 September 1964 and 7 December 1968, show the site was originally at the rear of the post office on the western side of the Stuart Highway. However, by 8 June 1992 it had moved to the disused WWII rail terminus of the North Australian Railway, which ceased operations and closed in February 1981.

While the alleged move within the confines of the terminus in November 1998 was not influential, the 1986 step-change was probably due to relocating the site there and replacing the original 230-litre Stevenson screen with a 60-litre one. It seems the move to the trucking-yard, allegedly in 1998, was erroneously reported. 

Google Earth Pro satellite images from October 2004 show surrounds of the trucking-yard site were generally bereft of ground cover, which would be sufficient to cause Tmax data to be warmer than at the previous site behind the post office.

Although categorical multiple liner regression confirmed that the up-step in the mean occurred in 1986 (Table 1(iii)), low R2adj and apparent overlap in scatter between the series (Figure 2(e)) show data were of exceptionally poor quality, particularly after 1986. This may have been due to lackadaisical observing practices, including excessive numbers of missing data/year. (Fewer than 340 observations were noted in 1970, 1975 and 1976, and from 1994 to 1998). Although given its role in ACORN-SAT and that the site is isolated, it is surprising that the BoM did not install an automatic weather station at Larrimah. The current site probably closed due to a lack of local interest in making observations and undertaking maintenance.  

It was concluded that while data quality is poor, accounting for rainfall and the step-change simultaneously using categorical multiple linear regression, left no additional trend or change that could be attributed to the climate, CO2, coalmining or anything else.

Table 1. Statistical summary. RSS refers to residual sum of squares. Partial R-square (R2partial) estimates the proportion of variation explained by the Sh(ift)factor that is not explained by rainfall alone (calculated as: [(RSS­full – RSSrain)/ RSSfull)*100].

ModelCoef. (oC/100mm)PR2adjSegmentRainAdj (SE) (oC)(1)RSS (R2partial) 
(i) Tmax ~ rain (all)      Not 1982, 2011(2)-0.176 -0.171<0.001 <0.0010.379 0.395    10.649 
(ii) Tmax ~ rain(2) 1965-1985 1986-2011  -0.175 -0.175  <0.001 0.004  0.482 0.400    
(iii) Tmax ~ Shres + rain(2)-0.175<0.0010.4731965-1985 1986-2011 Delta(1 vs 2)33.7(a) (0.113) 34.2(b) (0.101) 0.41 (0.151)  12.507 (14.9%)Interaction Tmax ~ Shres * rain ns
(iv) Tmax ~ Year(2) 1965-2011 1965-1985 1986-2011(oC/decade) 0.028 0.240 -0.426(3)  0.720 0.329 0.025  ns ns 0.165          
(1) Letters in parenthesis indicate differences between means (2) Outliers omitted from the analysis (3) Affected by a single extreme value

Bill Johnston

17 February 2024

Click here for the full Larrimah data

Preferred citation:

Johnston, Bill 2024. Is homogenisation of Australian temperature data any good? Part 7a. Larrimah, Northern Territory, Australia http://www.bomwatch.com.au/ 3 pp.


Unethical scientific practices including the homogenisation of data to support political narratives undermines trust in science. While the persons mentioned or critiqued may be upstanding citizens, which is not in question, the problem lies with their approach to data, use of poor data or their portrayal of data in their cited and referenceable publications as representing facts that are unsubstantiated, statistically questionable or not true. The debate is therefore a scientific one, not a personal one.


David Mason-Jones is gratefully acknowledged for providing invaluable editorial assistance. Research includes intellectual property that is copyright (©).

[1] Former NSW Department of Natural Resources research scientist and weather observer.

[2] Parafield Ref Welcome to homogenisation – arguably the greatest scam of 20th century science (bomwatch.com.au)

[3] Victoria River Downs Ref VictoriaRiverDowns-16-Feb-2024-1.pdf (bomwatch.com.au)

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.