4.1 Methods and data
Our projections of drought are based on
simulations from 13 climate models that perform
acceptably well in the Australian region and for
which potential evaporation data exist. The models are described in the Supplementary Information.

This is all the more puzzling given the purpose of the (dubious) selection process in Suppiah was to identify the best performing models in the Australian region. So no, I haven’t raised that issue. I am waiting on an answer regarding the validation failures.

4.1 Methods and data
Our projections of drought are based on
simulations from 13 climate models that perform
acceptably well in the Australian region and for
which potential evaporation data exist. The models are described in the Supplementary Information.

This is all the more puzzling given the purpose of the (dubious) selection process in Suppiah was to identify the best performing models in the Australian region. So no, I haven’t raised that issue. I am waiting on an answer regarding the validation failures.

And the Norwegian (BCC), Meteo-France (CNRM), Geophysical Fluid Dynamics (GDFL 2.0 and 2.1), German/Korean MPI-ECHAM 5 and Hadley Centre HADGEM1 models were all used in Suppiah et al 2007 but were not among the 13 used in the DECR.

Has there been any explanation for this large difference between the suite of models used in the two studies?

And the Norwegian (BCC), Meteo-France (CNRM), Geophysical Fluid Dynamics (GDFL 2.0 and 2.1), German/Korean MPI-ECHAM 5 and Hadley Centre HADGEM1 models were all used in Suppiah et al 2007 but were not among the 13 used in the DECR.

Has there been any explanation for this large difference between the suite of models used in the two studies?

This sentence alone is a scientific disgrace.

“Evidently…”. Evident to whom? Show the evidence. The paper uses models whose performance is under test; it finds that the model graphs are offset from reality. It concludes that the models have “skill” because model results can differ from actual.

Aha! But these models have the skill to bias as well. The more skilfull show a (rainfall) decrease in future years.

The models, although they deal with future projections and so can never be shown right in advance, are ranked in table 4.1, with scores on a scale of 0 to 1. Of course, none is negative. The range is lowest .304 to highest at .700.

Would not a student of elementary mathematics be amused by the use of 3 significant figures to “guess” as”skill” that can never be show correct?

But there are more steak knives. The models, according to chic mantra, also forecast global temperature changes. Of course, none of these is negative, that is not chic. The relation between model skill and temperature forecast is not strikingly correlated. The lowest temperature change is predicted at 1.96 degrees by a model with skill of 0.508, the highest change of 4.31 degrees from a model skill of 0.608. So the models with middle skill produce both the highest and lowest forecast of global warming. How odd.

It becomes even stranger. The CSIRO models mark 3.0 and 3.5 have skills of 0.601 and 0.607. What a disappointment to spend all that effort for an improvement of 0.1 percent. BUT, the forecasts of global warming have improved, from 2.11 degrees in mark 3.0 to 3.17 degrees in mark 3.5.

In a move that would surely please the policy makers, we have gained 30% severity in Global warming – with no more skill than cherry picking.

This sentence alone is a scientific disgrace.

“Evidently…”. Evident to whom? Show the evidence. The paper uses models whose performance is under test; it finds that the model graphs are offset from reality. It concludes that the models have “skill” because model results can differ from actual.

Aha! But these models have the skill to bias as well. The more skilfull show a (rainfall) decrease in future years.

The models, although they deal with future projections and so can never be shown right in advance, are ranked in table 4.1, with scores on a scale of 0 to 1. Of course, none is negative. The range is lowest .304 to highest at .700.

Would not a student of elementary mathematics be amused by the use of 3 significant figures to “guess” as”skill” that can never be show correct?

But there are more steak knives. The models, according to chic mantra, also forecast global temperature changes. Of course, none of these is negative, that is not chic. The relation between model skill and temperature forecast is not strikingly correlated. The lowest temperature change is predicted at 1.96 degrees by a model with skill of 0.508, the highest change of 4.31 degrees from a model skill of 0.608. So the models with middle skill produce both the highest and lowest forecast of global warming. How odd.

It becomes even stranger. The CSIRO models mark 3.0 and 3.5 have skills of 0.601 and 0.607. What a disappointment to spend all that effort for an improvement of 0.1 percent. BUT, the forecasts of global warming have improved, from 2.11 degrees in mark 3.0 to 3.17 degrees in mark 3.5.

In a move that would surely please the policy makers, we have gained 30% severity in Global warming – with no more skill than cherry picking.