Monday, February 8, 2010

Assumptions regarding Canberra rainfall predictions

I have made the prediction on this blog that Canberra inflow - the water that ends up in dams - will effectively drop to zero by around 2017. It should be noted that this is not a solid number - there is a considerable statistical range over which this could happen. But 2017 is the mean.

There are a number of assumptions underpinning this prediction. Any one of those assumptions might be wrong.

The basic argument is that rainfall is declining due to climate change and that declines in rainfall cause declines in runoff. This decline has been observed.

There are three sets of arguments running here.

Argument 1

Premise 1: Climate change will continue.
Premise 2: If climate change continues, daytime temperatures in Canberra will continue to increase.
Conclusion 1: Therefore, daytime temperatures in Canberra will continue to increase.

Argument 2

Premise 3: Increases in daytime temperatures reduce rainfall.
Premise 4: Reductions in rainfall lead to reductions in runoff.
Conclusion 2: Increases in daytime temperatures reduce runoff.

Argument 3:

Premise 5: Daytime temperatures in Canberra will continue to increase.
Premise 6: Increases in day time temperatures reduce runoff.
Conclusion 3: Runoff will reduce.

(Note how conclusions 1 and 2 became premises 5 and 6.)

There are other assumptions here - the premises have not been spelt out completely. The main assumptions involve the rate at which day time temperatures will increase and the rate at which those temperature increases will cause rainfall, and hence runoff, to decline. The rates predicted are based on empirical observation.

The problem with this is that the future may well differ from the past in this respect. Just as it is possible that we have undergone a step change in climate here in Canberra, so it is also possible that we will undergo another one soon.

Further, it is possible that there is a limit at which daytime temperature ceases to become the dominant driver of rainfall. Thus, a one degree rise in temperature in the future may not have the same effect as a one degree rise in the past.

So, my prediction could well be wrong. But that is how science advances: by linking empirical observation to a theoretical model and then testing that theoretical model against further empirical observation. At the moment, my theoretical model - based in empirical observation - is predicting an end to Canberra runoff by 2017.


  1. Shouldn't there be an Argument 2(b):
    increases in daytime temperature should increase evaporation.
    Increased evaporation leads to reduced runoff.

    Also, while I see that you have shown a historical correlation between increased temperature and reduced rainfall, do you know how good the theoretical underpinning is? eg, all else being equal, global warming leads to more rainfall. In certain areas there are theoretical reasons to believe that the global average increase won't apply - eg, the southwest U.S. may be encompassed by enlarging Hadley cells which cause more drying. Is there a parallel reason to believe that Canberra should see a reduction of rainfall, or is it just the historical correlation?

  2. Anonymous,

    I have ignored evaporation and transpiration. But yes, that will tend to reduce runoff. One of the reasons that the CSIRO has put forward for why this dry period throughout the Murray Darling Basin has seen a greater decline in runoff than in previous dry periods - the Federation drought and the drought in World War 2 - is that the temperature is warmer.

    Re a theoretical underpinning, I am just using the statistics. I have no theoretical underpinning of my own.

    However, re increased rainfall generally due to global warming, this is true. But regional predictions (which are likely to be much less accurate, due to resolution issues) say that Canberra and the south of Australia generally will experience less rainfall, with the north experiencing more.

    One of the reasons for this is that increased temperatures are predicted to increase precipitation in Antarctica, which will mean that the weather systems that the south of Australia will get will contain less moisture.

    It has been shown that there is a direct link between observed increases in snowfall in Antarctica and the decline in rainfall in south-west WA. While no such direct link has been demonstrated for Canberra, this is a potential reason for our loss in rainfall.

  3. Re the regional predictions: I should point out that the predictions for declining rainfall did not predict as much decline as we have seen. Indeed, we have surpassed the worst-case scenarios for 2030 and we are approaching the worst-case scenarios for 2070.

    This is why the CSIRO suspect that Canberra may have undergone a step change in climate, with us moving to a state of permanent lower rainfall.