Special Report: Climate Change and River Flooding

4.3. Australia

4.3.1. Projections

Using an improved version of the mixed-layer model of CSIRO, Whetton et al. (1993) found an increase in the frequency of high-rainfall events and a decrease in the frequency of low-rainfall events. This results applies over most of Australia. The frequency of the simulated heavy rain events and increased by factors ranging from 2 to 5 under 2xCO2 conditions. These findings are in agreement with the global scale modelling by Gordon et al. (1992), using an earlier version of the same model. Current investigations with a limited-area model (island of Tasmania) nested in the global CSIRO model predicts that the total number of rain days decreases while the frequency of high-intensity rainfall events increases.

4.3.2. Observations

Suppiah (1994) concluded that the synoptic climatological patterns during anomalous wet and dry conditions in central Australia agree fairly well with those simulated by the CSIRO climate model. No trends in rainfall were investigated.

Drought and moisture indices calculated for Australia showed no systematic long-term changes (IPCC, 1990).

The relation between rainfall in Australia and changes in ENSO remains uncertain. ENSO is said to account for one third of the total variance in seasonal rainfall (Pittock, 1984). On the other hand, Cordery and Yao (1993) conclude that although strong, consistent relationships between ENSO and rainfall have existed for many years, they are subject to sudden changes that do not appear to be linked with any other phenomena.

4.3.3. Flooding

A study performed by Chiew et al. (1993) describes the impacts of climate change on Australian catchments by use of a hydrologic daily rainfall-runoff model. The first part of this study consisted of a sensitivity analysis. In general, changes in rainfall are amplified in the simulated runoff. In the wet and temperate areas, the percentage change in runoff is about twice as much as the percentage of change in rainfall. In the drier areas, large increases in rainfall can enhance runoff by more than five times the change in runoff. In these catchments, runoff also reduces very quickly as rainfall decreases. This is mainly because the percentage of rainfall that becomes runoff is very low in these catchments. The change in rainfall has little effect on the soil moisture in wet catchments but in drier catchments, the percentage of change in soil moisture levels can be greater than the percentage of change in rainfall. Compared to precipitation, temperature increases alone have negligible effects.

The second part of this study used results from five global climate models to provide regional climate change scenarios. This information served as the input for the hydrologic model. The range of changes in runoff and soil moisture by the year 2030 were then estimated. The GCMs reasonably agree on a general increase of temperature and rainfall across most of Australia, but quantitative predictions vary considerably on the regional scale.

In the summer, rainfall increases of up to 5% per degree of global warming are predicted in southeast Australia and of up to 10% per degree of global warming elsewhere. In the winter, three separate regions are identified: the models agree on an increase (of up to 5% per degree warming) in Tasmania, agree on a decrease (of up to 5%) in central and south Australia, or fail to agree (-5% to +5%) in eastern and southwest Australia. The average global warming falls in the range of 0.6 K to 1.7 K by 2030. For example, the maximum increase of summer precipitation on the northeast coast will then range from 6% (0.6x10) to 17% (1.7x10).

The model simulations with the mentioned precipitation increases as input indicate a nonlinear increase in annual runoff: up to 25% by 2030 in the wet-tropical catchments near the northeast coast of Australia. Because the GCMs do not agree in the direction of rainfall change in south-east Australia, the direction of the runoff is also uncertain: ą20%. For Tasmanian catchments, a 10% increase in runoff is simulated whereas for catchments in the South Australian Gulf, up to 35% decrease in annual runoff is simulated for 2030. Near the western coast, simulations show runoff changes of up to ą50%. It is concluded that the runoff changes simulated for many parts of Australia are large enough to affect the prevailing flooding regimes.

Another approach is demonstrated by Bates et al. (1994). The effects of doubled CO2 concentrations on a small South Australian catchment were evaluated using the results from the CSIRO GCM, a stochastic weather model and a water balance model. Changed climate sequences are produced by adjusting the weather generator according to GCM trends. The long-term sequences (1000 years) of artificial daily rain records are then used to the drive the hydrologic model.

Although results show only a marginal decrease in median monthly runoff during winter months, large increases in monthly runoff maxima are indicated for August and September due to large increases in extreme monthly rainfall. Modest increases in evapotranspiration were also indicated for these months.

In a case study of urban flooding damage, Minnery and Smith (1994) links flood probability and flood damage for two urban river catchments in New South Wales. He estimated that the financial damage would be between 4.5 and 11 times higher under doubled CO2 concentrations (the connection between the increasing rainfall intensity and probable maximum flood value was not considered).