climate change wa

RAW temperatures at ACORN locations
1910-2013

Methods
Results
Analysis
BoM drift correction for expanding ACORN

Also see Comparison of RAW and ACORN temperatures 1910-2013

Methods

  • Daily minimum and maximum RAW temperatures were downloaded from the Bureau of Meteorology's Climate Data Online (CDO) for the 112 locations and 209 weather stations that have contributed to the ACORN network since 1910 (eg. post office originally and then moved to nearby airport).
  • This analysis uses only CDO RAW temperatures and not ACORN temperatures, which are homogenised and adjusted by the BoM. RAW temperatures have had comparatively little BoM adjustment.
  • BoM CDO RAW data in this analysis are calculated from the annual average of daily temperatures for each year and are subject to the usual discrepancies such as missing days, but are identical to BoM annual estimates except for occasional 0.1C variations caused by the BoM averaging monthly averages for annual calculations instead of averaging all available days per year for annual calculations.
  • RAW mean temperatures are calculated as an average of RAW minima and maxima.
  • RAW data is included at nine locations where CDO temperatures are not available for different periods after 1910 as daily temperatures but are available as annual temperatures (Gabo Island 1910-1957, Kerang-1910-1962, Miles 1910-1957, Moruya Heads 1910-1957, Normanton 1910-1957, Pamerville 1910-1957, Port Macquaurie 1910-1957, Wilson Promontory 1910-1957, Yamba 1910-1957). An example snapshot below from Climate Data Online shows how Yamba daily temperatures are unavailable as blue linked daily data before 1957 but yearly temperatures are extracted from the right Annual column before that year:

climate data online example

  • RAW annual network temperatures are calculated after correcting all RAW data so that if there is a minimum but not a maximum at any station any year, or vice versa, the remaining temperature is deleted so there can be no possibility of a high number of missing minima or maxima at various stations in any given year which might create a bias in the annual calculation for all stations in that year, nor an imbalance of minima or maxima in calculating mean from min+max/2. Example of source data correction:

temperature correction example

  • Shaded background areas in the charts represent the average long-term annual minimum, maximum and mean temperatures from 1910 to 2013 of all stations operating in that decade of the ACORN timeline. They represent the average of all years at those stations from start to 2013 and not the average temperature in that particular decade. The shaded areas include a proportion of temperatures that hadn't yet happened at those locations but are indicative of the averaged overall temperature at stations within the ACORN network that were open at the time, largely dependent on latitude and coastal proximity.
  • The ACORN network had 53 contributing weather stations in 1910, increasing to 61 by 1920, 65 by 1930, 70 by 1940, 86 by 1950, 104 by 1960, 110 by 1970 and 112 by 1980, the last two stations opening at Rabbit Flat in 1970 and Learmonth in 1976. Rabbit Flat and Learmonth are located in the hot climates of the Northern Territory and WA's north-west coast, and are used as a variable in this analysis to examine the influence of local climates at new weather locations on Australia's averaged temperature trends.
  • The spreadsheet source (Excel 442kb) is available for charts on this page and data sorting of temperatures in the source Excel can be downloaded here (Excel 90kb)

Results

australian raw minimum temperature record

australian raw maximum temperature record

australian raw mean temperature record

  • Comparing the first 10 years of the record against the last 10 years, the RAW minimum temperature at all 112 stations has increased 0.05C (13.46C > 13.51C) and excluding Learmonth and Rabbit Flat has decreased 0.02C (13.46C > 13.44C).
  • Comparing the first 10 years of the record against the last 10 years, the RAW maximum temperature at all 112 stations has increased 0.57C (25.07C > 25.64C) and excluding Learmonth and Rabbit Flat has increased 0.44C (25.07C > 25.51C).
  • Comparing the first 10 years of the record against the last 10 years, the RAW mean temperature at all 112 stations has increased 0.32C (19.26C > 19.58C) and excluding Learmonth and Rabbit Flat has increased 0.21C (19.27C > 19.48C).
  • Comparing the first 20 years of the record against the last 20 years, the RAW minimum temperature at all 112 stations has increased 0.07C (13.32C > 13.39C) and excluding Learmonth and Rabbit Flat has remained static (13.32C > 13.32C).
  • Comparing the first 20 years of the record against the last 20 years, the RAW maximum temperature at all 112 stations has increased 0.47C (25.01C > 25.48C) and excluding Learmonth and Rabbit Flat has increased 0.33C (25.01C > 25.34C).
  • Comparing the first 20 years of the record against the last 20 years, the RAW mean temperature at all 112 stations has increased 0.28C (19.16C > 19.44C) and excluding Learmonth and Rabbit Flat has increased 0.17C (19.16C > 19.33C).
  • Comparing the first 30 years of the record against the last 30 years, the RAW minimum temperature at all 112 stations has increased 0.11C (13.28C > 13.39C) and excluding Learmonth and Rabbit Flat has increased 0.04C (13.28C > 13.32C).
  • Comparing the first 30 years of the record against the last 30 years, the RAW maximum temperature at all 112 stations has increased 0.36C (24.96C > 25.32C) and excluding Learmonth and Rabbit Flat has increased 0.22C (24.96C > 25.18C).
  • Comparing the first 30 years of the record against the last 30 years, the RAW mean temperature at all 112 stations has increased 0.23C (19.12C > 19.35C) and excluding Learmonth and Rabbit Flat has increased 0.13C (19.12C > 19.25C).

Analysis

  • The inclusion of Rabbit Flat in 1970 and Learmonth in 1976 added 0.07C to the minimum, 0.13C to the maximum and 0.11C to the mean temperature of Australia's ambient climate within the ACORN network (first decade vs last decade), demonstrating how local climate at just one or a few stations can affect the annual averaged temperature of all 112 locations.
  • In the 52 years from 1962 to 2013, Australia's mean RAW temperature at all 112 ACORN location was .17C warmer than in the 52 years from 1910 to 1961 (19.00C > 19.17C).
  • In the 52 years from 1962 to 2013, Australia's mean RAW temperature at 110 ACORN location (excluding Rabbit Flat and Learmonth) was .09C warmer than in the 52 years from 1910 to 1961 (19.00C > 19.09C).
  • Maximum temperatures in particular have risen sharply since 2002. Australia's maximum temperature within all 112 ACORN stations averaged 25.66C from 2002 to 2013, up .49C on the 25.17C average from 1990 to 2001. Australia's minimum temperature averaged 13.46C from 2002 to 2013, up .10C on the 13.36C average from 1990 to 2001, and Australia's mean temperature averaged 19.56C from 2002 to 2013, up .28C on the 19.26C average from 1990 to 2001.
  • There has been no addition or change to the number of locations in the ACORN network since 1976 but many locations have seen their relevant weather stations shifted during that time including the 1990s and since the year 2000. Environmental influences such as the Urban Heat Island in cities and regional towns may be contributing to the increase in temperatures since 2002, as may instrument influences such as the advent of Automatic Weather Stations at many ACORN locations since the early 1990s.
  • The mean RAW temperature at all 112 ACORN locations in 1961-90 was 18.99C compared to the ACORN tracker's 1961-90 mean temp baseline of 21.8C.
  • Australia's mean RAW temperature at 112 stations (1910-2013 average 19.09C) in the 12 years from 2002 to 2013 was 19.56C, which compares to the mean at 61 stations (1910-2013 average 19.36C) in the 12 years from 1910 to 1921 of 19.27C, a 0.29C increase.
  • This analysis uses ACORN as an accepted network of locations but does not use ACORN adjusted temperatures and is not a critique of ACORN adjustments or claimed temperature trends based on ACORN.
  • The Top 20 hottest years in RAW are tabulated:

hottest australian years

BoM drift correction for expanding ACORN

A significant source of inhomogeneity in spatial averages computed from these analyses arises from spurious changes in the climatology as the mean location of the network changes over time. In Australia, this may occur (for example) during periods where the network has expanded into warmer northern and central locations across the continent. This transient drift in the mean climatology of the network must be estimated and removed from the spatial-average time series in order to perform any meaningful comparison with homogenised datasets.

We perform a somewhat simple but objective adjustment for this network non-stationarity in a `drift-corrected' grid set. We first generate a full set of monthly analyses from the raw data for the period 1911-2010, and in the process calculate 1981-2010 monthly climatologies from the raw analysis grids. The second step is to generate a parallel set of monthly analyses for the period 1911-2010, in which the raw data fed into the analysis are replaced by climatological values at each site reporting in the particular month. These climatological values at each site are interpolated from the 1981-2010 monthly climatology grids, using the bi-cubic polynomial interpolation technique mentioned above. If the network were completely static, no changes over time would result in this parallel set of analyses, apart from the normal annual temperature cycle, but the network is obviously not static and so some variation over time results. The drift-corrected analysis is then obtained by subtracting the parallel analysis from the raw analysis.

The nationally averaged annual time series of the difference between the drift-corrected analysis and the raw analysis is plotted in Fig. 2, anomalised with respect to the 1981-2010 period. The magnitude of the drift correction is not very large for minimum temperature, but is quite large (nearly 1°C) for maximum temperature in the early years. The drift correction serves to “reduce” the 100-year trend in the raw analyses by a substantial amount. It is important to note here that the warming trend in maximum temperature from the raw analyses is much larger than that in the existing homogenised analyses. The drift correction over the last thirty years is very small, indicating a degree of network stability over that period.

bom acorn analysis


Top of page

Also see Comparison of RAW and ACORN temperatures 1910-2013

Stations tableAustralia HQ auditAustralia ACORN-SAT audit
Western Australia ACORN-SAT auditNew Zealand 11 Station auditNew Zealand Seven Station audit
Website home


Bookmark and Share



This website produced by Scribeworks 2009-2014

Back To Top