Media scientist Sebastian Vehlken: climate models have developed a life of their own, where hard measured data only disrupt
By Sebastian Lüning and Fritz Vahrenholt
(Translated/edited by P Gosselin)
Media scientist Dr. Sebastian Vehlken works on the theory and history of computer simulation, and studies crowd research and super-computing. In the radio show “Voices of the Sciences” he made very critical statements on computer models’ capability to make prognoses. Quentin Quencher put up a part of the interview at vimeo. In a blog posting at Glitzerwasser, Quencher sorted out an interesting part:
[Vehlken in interview]:
Computer simulations are not necessarily based on sound data that I then project. Here there is always the difference: projections are not prognoses. When I’m dealing with the future with respect to computer simulations, it’s always about generating projections, scenarios – which are always in plural. They are not the future in the system. Here, for example, we can look at climatology.
Recently there was a presentation at our institute. It was about, okay, the claim about climate models is always: ‘Your database isn’t broad enough.’ And then the scientist who made the presentation told us: ‘On one hand the simulators tells that the models run very well, and then we put in data from some sensor in Antarctica…and suddenly it just doesn’t fit.’ Why doesn’t it fit in? Often it’s because the sensor was defective, or the measured data are simply faulty. And so that means I often have systems that are hardly able to produce reliable data.
Yet, I need, so to say, models. And in the meantime, with 20 -30 years of climate research that are simulation-based, the models have become so complex that they have in a sense developed a life of their own – and sometimes measured data simply just disrupt them. In the models there at times assumptions that simply cannot be physically validated. That is they have articifical factors, parameterizations without end. For these there are no empirical databases for them. There is…a complex system..and then there is experimentation with this complex system, which is what my climate model is. And then one sees, okay, when I implement this parameter, which I don’t even know whether or not it really exists.”
In the comments that follows the blog article a friend of models comments and attempts to convince Quencher of their supposedly great use for them. Very amusing. Read for yourself.
At his Glitzerwasser site here, Quencher sums up what Vehlken says above:
In short, the models, especially the climate models, may be good tools but for making prognoses they are not worth anything, and the projections are fraught with so much uncertainty that they absolutely should not be used as a basis for making policy decisions.”
Flashback: Prof. Christopher Essex