Testing hypotheses

I have presented the hypothesis on this blog that Canberra will become a desert by mid century if greenhouse gas emissions are not dramatically reduced.

A test of whether a claim is scientific or not is whether there are empirical observations that would demonstrate that the claim is unlikely to be true. In the case of my hypothesis, there are indeed empirical observations that would demonstrate this, and I will list some of them here.

1.) Obviously, if the five-year average rainfall has not fallen to 250 mm or lower by 2050, my hypothesis would be 'unlikely to be true'. I would suggest that there is some wiggle-room here - if five-year average rainfall has fallen to 253 mm, then I would still claim that my hypothesis is likely to be correct. However, if we set 265 mm as the cut-off, that makes things very clear.

2.) The above empirical observation cannot be made until 2050, which is a fair way off. There are, however, observations that can be made sooner that would demonstrate that the claim is unlikely to be true. A year with rainfall of 850 mm or above would be just such an empirical observation. Further, I would suggest that for every five years that passes, we can reduce this by 30 mm. In other words, for the years 2010-2014, if any one of those has 850 mm of rainfall or above, my hypothesis is unlikely to be true; for the years 2015-2019, if any one of those has 820 mm rainfall or above, my hypothesis is unlikely to be true.

Five years out of 70 have had rainfall over 850 mm, so on average one year in 14 should meet that criteria.

It should be noted that these are the current numbers - they will not increase, but they may decrease as more data comes in that shows definitely that the variance in annual rainfall has declined - at the moment, these figures rely on the current observed variance. By way of an explanation, currently, the standard deviation is about 170 mm, meaning that 95 per cent of the time the rainfall will be within the range of the average plus or minus 340 mm.

3.) If any five-year average is 670 mm or greater. This figure will again decline by 30 mm for every five years that passes.

21 five-year periods out of 66 have had an average of 670 mm or greater, so one in three.

4.) If any 10-year average is 600 mm or greater. This figure will again decline by 30 mm for every five years that passes.

44 10-year periods out of 61 have had an average of 600 mm or greater, so more than two in three.


This should be sufficient for now. There are other possible criteria, such as rainfall in multiple five-year periods in a row, but they are mostly covered by the 10-year average situation.