Australasian Science: Australia's authority on science since 1938

Climate Change or Natural Variability?

Image of barometer

The long-term trend in annual rainfall for Australia from 1900 to 2009 is upwards at a linear rate of 6.33 mm/decade.

By Robert E. White

Meteorological records since the 1950s reveal a decrease in rainfall that is consistent with anthropogenic climate change, but a different picture emerges when looking at records since 1900.

The Fourth Assessment Report of the Intergovernmental Panel on Climate Change concluded that the increase in the Earth’s surface temperature during the second half of the 20th century can only be simulated by models that include anthropogenic forcing, and not just natural factors. However, the report acknowledges that there are difficulties in simulating and attributing observed temperature changes at smaller than continental scales, expecially when attempting to model the amount and distribution of rainfall.

As part of the Australian Climate Change Science Program, CSIRO and the Bureau of Meteorology (BOM) scientists have modelled a number of attributes of the Australian climate at subcontinental scales, and concluded in their State of the Climate statement that, relative to the period 1981–2000, “decreases in rainfall are likely in the decades to come in southern areas of Australia during winter (and) in southern and eastern areas during spring”.

The statement makes the point that “models make assumptions about future events such as CO2 emissions, and are designed to paint a picture of a series of possible future states based on known facts... as opposed to observations which are accurate measures of an event that has already occurred”. This latter statement should be examined more closely because records of events can be used selectively.

For example, the Federal Department of Climate Change and Energy Efficiency has used a comparison of measured rainfall distribution in eastern Australia from the 1950s to 2008 to imply that climate change is responsible for decreased rainfall over this period. Similarly, in discussing climate change impacts, the Victorian government’s White Paper, Securing our Natural Future, uses the rainfall record from the 1950s to 2008 to infer that anthropogenic factors have caused a major change in rainfall patterns, with eastern Australia becoming considerably drier.

However, the BOM’s records reveal that, despite considerable variability, the long-term trend in annual rainfall for Australia from 1900 to 2009 is upwards at a linear rate of 6.33 mm/decade. A similar positive trend, but of lower gradient, is observed for eastern Australia, southern Australia and the Murray–Darling Basin, while for south-eastern Australia the trend is slightly negative at less than 1 mm/decade.

The BOM’s long-term record also shows that the two decades of the 1950s and 1970s were wetter than average, so it is misleading to use either the 1950s or 1970s as the starting point in calculating changes in regional rainfall distribution up to the present time (Fig. 1), and disingenuous then to infer that these changes are caused by human activities.

Rainfall records show that the drought of 2000–09 was prolonged and severe in eastern Australia. However, this drought was comparable in severity to the Federation drought at the end of the 19th century.

Indeed, when the Bureau’s complete record of rainfall from 1900 is examined, a different picture emerges of the decadal changes in rainfall (Fig. 2). In this case, most of the Murray–Darling Basin shows an increase in rainfall ranging from 0–5 mm/decade to as much as 15–20 mm/decade.

Given the known variability in Australian rainfall acknowledged in the CSIRO report, surely the longer the period examined the more reliable is the picture that might emerge of a possible long-term trend in rainfall that reflects natural variation and is not the result of anthropogenic forcing.

The joint CSIRO–BOM statement on climate also states that “because models are representations of the future based on a range of emission scenarios, they tend to produce a range of results”. True, but in his book Climate Change: The Science, Impacts and Solutions, Barrie Pittock estimates that about half the uncertainty in model projections is due to uncertainty in the actual emission scenarios in the future, and half due to uncertainty in the modellers’ understanding of how physical processes interact to determine the climate.

With respect to the latter point, modellers need to know the full range of variables that influence climate and how they operate, now and in the future; and also to apply numerical values to these variables and parameterise the relationships between these variables and the specific outputs sought from the model. Models can be only as good as the spatial data available for the input variables and the assumptions that are made, not only about future CO2 emissions but also about the complex physical input–output relationships.

For example, the El Niño-Southern Oscillation (ENSO) is an important influence on the incidence and severity of drought in eastern Australia. However, climate scientists acknowledge that ENSO effects on the Australian climate are not well-simulated by current models. Additional uncertainty comes from an inadequate understanding of the interaction between ENSO and the Indian Ocean Dipole, which has recently been shown to influence rainfall in central and southern Australia.

Thus, while the CSIRO/BOM statement that “increased computer capacity allows us to make projections with increased accuracy” is fair, we should realise that a better range and quality of input data, especially for the Southern Hemisphere, and improved understanding of the physical relationships, especially as they determine rainfall, will be required to bring increased certainty to model projections.

Achieving this will require much greater government investment in the infrastructure necessary for collecting data at a high resolution, and the funding of research, than has been the case in the past. For a country such as Australia, with its record of climate variability, greater certainty in model projections for rainfall amounts and their regional distribution is of critical importance in planning to meet the competing demands of agriculture, urban communities and the environment.

Irrespective of improvements in model projections through increased computing power and better physical inputs, it is worthwhile to remember a lesson from the first half of the 20th century, when the determinism of classical Newtonian mechanics was displaced by the statistical uncertainty of quantum mechanical events. As Niels Bohr, one of the great physicists of that period said: “Prediction is very difficult, especially of the future”.

Robert E. White is Professor Emeritus of The University of Melbourne’s School of Land and Environment.