Climate Forecasts: Intrinsically All For Nought?

It is commonly accepted that all it will take for us to be able to predict future climate, is faster computers with gigantic computing power, in a progression analogous to meteorology’s.

I find that unlikely. And that’s why climate forecasting is likely to go nowhere, just like…alchemy. Not due to anybody’s fault: rather, because it is intrinsically impossible for it to do otherwise.

=============

Climatology is the study of the long term behavior of something that is stable, and predictable, but only until it goes through a state change, perhaps all of a sudden.

Imagine what if anything would nuclear physicists be able to study if there were not even sure if today’s proton-proton accelerators would or would not transform themselves into proton-neutron accelerators, in ten, thirty or a hundred years, thereby completely changing all results and all predictions?

Add to that the climatological possibility, or shall I say the absolute certainty, that in any meaningful (i.e. multi-decadal) period of study, there will be external, uncontrollable, unpredictable inputs such as volcano eruptions and changes in the Sun…as if the energy available to power the LHC would vary at random.

Check for example this statement from German climate scientists, just published on the Süddeutsche Zeitung:

“If we had had 10% more cloudiness over Germany, that would have compensated for the warming of the past 30 years”

In other words, a minor change in a climate detail is enough to modify the end result altogether.

=============

How do you study in a scientifically appropriate manner a system whose simplest scientifically appropriate representation is…itself?

You don’t. You cannot even follow the usual statistical route, because in the long term every possible solution is equally probable. And if you don’t work on the long term, on the decadal or secular scale, then you are not doing climatology.

The problem of climate forecasting is therefore unassailable, just as it is not possible to predict the stock market, another system that is heavily influenced by external factors.

Think of the money thrown for nothing in the financial forecasting route. Then, think of the results.

=============

Of course, the above does not mean that we can not do any climatological study, for example to determine which crops appear to be more suitable for a certain territory…just as one can play the stock market using reasonably objective parameters and computer models without falling necessarily into financial ruin.

But climate forecasting might be the one and only science where “blacks swans”, the events that throw all predictions up in the air, are ironically the one thing that can be predicted.

ADDENDUM 00:27GMT June 5: Isn’t it beautiful to write something on your own in the middle of the night, and then to discover that even Roger Pielke Sr. has just been dealing with a very similar topic?

ADDENDUM #2: There is one point that needs to be clarified in the above.

The climate forecasts I am talking about are multi-decadal. The stuff just criticised by Pielke Sr.  Those, I am educately guessing, are impossible, even if we knew all the physics and we had vast amounts of computing power.

The simplest way to compute the climate of 20-30 or more years in the future, is to build a system at least as complex as the climate own’s . In other words, the Earth’s climate is its own simplest multidecadal computer.

Since “climate” is usually taken as a multi-decadal concept, then perhaps we can move the forecast of what next season will bring, into meteorology.  Of course, before anybody says anything, no, I do not think meteorology is “inferior” to climatology.

ADDENDUM #3: many thanks to Douglas Hoyt for pointing out that Roger Pielke Jr. has just published a blog and article along the same lines of thought

Rather than basing decision-making on a predict (probabilistically of course) then act model, we may have to face up to the fact that skillful prediction of variables of interest to decision makers may simply not be possible. And even if it were possible, we would not be able to identify skill on the same time scales as decisions need to be made. The consequence of this line of argument is that if stationarity is indeed dead, then it has likely taken along with it fanciful notions of foreseeing the future as the basis for optimal actions. Instead, it may be time to rethink how we make decisions in the face of not simply uncertainty, but fundamental and irreducible ignorance