By this means, the relative extent to which MAXTEMP increases in those cities can be attributed to urbanisation, the enhanced greenhouse effect, and other causes, is quantified.
The influence of cities on overnight temperatures is well documented. However, their influence on daytime temperatures is less well documented.
Sydney and Melbourne MAXTEMP data are compared with other ADAM data sets and are found to be increasing at a faster rate than elsewhere.
For example, Sydney's MAXTEMP is increasing at a linear rate that is +0.065 deg C per decade faster than that of Newcastle, whilst Melbourne's MAXTEMP is increasing at a linear rate that is +0.050 deg C per decade faster than that of Ballarat.
More generally, Sydney and Melbourne MAXTEMP data are compared with ADAM data sets for the 73 Australian localities (excluding Sydney and Melbourne) with at least 80 years of MAXTEMP data during the 100-year period 1910 to 2009 inclusive.
MAXTEMPs at Sydney and Melbourne are found to be increasing, respectively, at rates 0.080 deg C per decade and 0.071 deg C per decade faster than the average temperature at the 73 sites. The probabilities of such large differences occurring by chance is <<0.1% in both cases. That the average MAXTEMPs in the two cities are rising faster than at more rural localities is therefore largely attributed to urbanisation.
For both localities, annual MAXTEMP data are statistically modelled over various control periods using MAXTEMP data at surrounding non-urban stations as input. Thereby, sequences of non-urbanised MAXTEMP can be constructed for the hypothetical circumstance of the cities not being built.
Synoptic stratification of daily data shows that a recent "jump" in the Melbourne series is due to buildings constructed immediately to the south of the site.
In contrast to the current study, Torok's (1996) PhD work identified, and adjusted for, "' jumps in the time series due to non-climatic changes' (and this consequently removed) ...urbanisation signals from the time series". Torok's adjustments have been applied to the derivation of the Bureau of Meteorology's high quality data sets (HQDS).
The current study's approach (using ADAM data sets) has been to identify, and preserve, the urbanisation signals in the time series.
As would be expected, MAXTEMP rising trends in the Melbourne and Sydney HQDS (with the urbanisation signals removed) are found to be slower than those in the corresponding ADAM data sets (without the urbanisation signal removed).
Supplementary URL: http://www.bom.gov.au