The Bureau of Meteorology did what to February?
Wow, just wow. Look what the Bureau of Meteorology has covertly done to February? Something like one third of a degree has been added to the average Australian summer maximum anomalies over the past few years according to the “expert” data from the worlds-best-practise equipment.
In the BOM Whopper Part 1 we revealed that in the BOM’s latest round of unannounced adjustments there were big increases in the rate of Australian summer warming. It turns out a lot of the summer rise comes from changes to February. Mysteriously, there were large changes to the national average of the last three years. Let that sink in.
These changes were incomprehensible because while the averaged “whole nation” got warmer, there were no changes to the data in any of the 104 individual stations.
It’s all rather spooky… but what it isn’t, is scientific.
The two main points in Bob Fernley Jones’ work:
- There are big increases to measurements recorded in the last three years? Why? Yet again, the adjustments are down in the early years, up in the latter years, and overall, the rate of warming, surprise, increases thanks to man-made adjustments. He points out that it makes no sense that modern equipment needs more adjustments than equipment from the 1960s? Somehow our thermometers today are under-reporting temperature? Seriously?
- Mysteriously, the national averages in recent years are different even though the individual site data is identical v1 versus v2 (and the raw Climate Data Online). Bobs Fernley Jones says “I can find no evidence that the area-weighting in itself has been changed, including my reading of The Second Book of Trewin (as in the V2 official doc). Perhaps there has been a covert change in how it is applied (different exponentially from the centre points maybe?), but that’s pure speculation on my part.“
The BoM needs to explain…
Despite urbanisation of modern sites, which would artificially warm them, we see adjustments that work in the opposite direction, effectively warming the modern era recorded in built-up sites, and cooling the past that was recorded with sites in fields and gardens.
Jo
—————————————————————————————————
Australia’s Bureau of Meteorology has covertly made summer hotter:
In 2011, the BoM adjusted (homogenised) Australia’s temperature records with the objective of making corrections for the varying measurement conditions back in time. Controversially, they deleted all data before 1910 by ruling them unreliable (especially the inconveniently hotter records) and by adjusting the surviving early 1900’s values generally downwards. Amid the controversy, this program received much official praise and publicity that culminated in a third and final government sponsored report in 2017 declaring it to be among the world’s best practice. Despite that acclaim, in October 2018 they found it necessary to quietly launch a new program that further increased warming rates.
The BoM has long had a propensity to issue many media releases and special reports that emphasise hot weather events, and yet strangely, they were silent in this matter. Consequently, the vast majority of Australians are unaware of the big changes to the already modelled data. It was presumably well received by the IPCC in time for their coming sixth assessment report though.
The two adjustment programs employ a methodology known as homogenization which is described under the acronyms ACORN-SAT (2011), and then by ACORN-SAT version 2 (hereafter v1 & v2). The discovery of what follows arose from enquiries to the BoM made last year over some already existing concerns with v1 data and the fortuitous archiving of BoM data and graphics that no longer exist on line (but which are easily proven to be genuine). It resulted in citizen researcher awareness of v2 and hence in recent interest to compare outcomes.
In monthly terms, the most extreme warming increase in the all-of-Australia summer average was in February, as seen in the following animation. It is derived from a BoM online download archived in early 2018 compared with the replacing v2 copy in 2019:
Australian February summer maximum temperatures, adjusted again? Animation by Chris Gillham at waclimate.net
Typically, (as seen elsewhere, regionally and temporally) the greatest adjustments are increasingly negative towards 1910 (cooling) and increasingly positive towards 2019. Typically, they are netting rather flat with minimal change in the centre and over the full range they are in a random magnitude pattern.
The v1 data were also archived which has enabled determination of comparative linear trend rates of the modeled data as follows:
Adjustments mean the trend in February maximums has risen (yet again). | Click to enlarge.
The v2 over v1 warming rate 1910 – 2018 is increased by 87%! (v1 data not available in 2019). Similar calculations for Summer (DJF) give an increased trend of 57% for 1910 – 2018.
Notice that the two trend lines merge and cross close to the centre, which is typical of what was seen in Part 1 (and elsewhere). In effect, this can be called ‘rotating the dataset anti-clockwise,’ with minimal net change around the centre. Strangely, equipment in the middle period as highlighted in the chart has seemingly worked more accurately than modern equipment. The reason behind this paradox is elaborated next.
The biggest surprise is seen in the following figure:
Adjustments mean February maximums have cooled in the distant past and risen in recent times. (Yet again). | Click to enlarge.
Of particular interest is that the final three years show increased warming on top of that already existing under “world’s best practice” in v1:
2016 = 0.34 °C
2017 = 0.33 °C
2018 = 0.30 °C
The site specific data did not change. To make this very clear, look at the table below with eight randomly chosen sites. The values are identical — there are no changes from the raw to V1 to v2. The mystery remains — if the data are the same, yet the averages are different, what has changed?
Thanks to Chris Gillham for review of draft, the animation and this table (and for Joanne’s edits).
ACORN-SAT national annual temperatures are the area-weighted * average of the 104 network stations. By definition, homogenisation should be a thing of the past and certainly shouldn’t cause adjustments to observed temperature over the past decade or so.
It will also be shown in Part 3 that there is an additional warming distortion in the data as a consequence of increasing the number of hot stations in the latter decades. A third of a degree in the summer months here, and another third there, all add up of course.
There is more analysis and information in the February and Summer Supplementary File
* ‘Area-weighting’ is the process of modelling the contribution of individual stations to the national/regional averages. Stations that are in remote areas have higher contributions than those concentrated closer together. There have been no changes to area-weightings in recent years.
REFERENCES:
- Australian Climate Observations Reference Network – Surface Air Temperature (ACORN-SAT) v2
- Technical report defining ACORN-SAT v1 (Trewin 2012)
- Technical report defining ACORN-SAT v2 (Trewin 2018)
- “Bureau’s procedures and data analysis as amongst the best in the world” (3-year Technical Advisory Forum final report 2017)
- Online time-series graphics and data v2.
- Part 1 of this series Who knew? The Australian Bureau of Met just made last summer hotter, and history colder (again)
- About BoM time-series charts and station area-weighting