Dissecting a Real Climate Text by Hendrik Tennekes

I understand that Gavin Schmidt was upset by my essay of January 29 . I admit that I neglected to mention that I responded to his long exposition of January 6 on Real Climate. The part of his text that deals with the difference between weather models and climate models reads:

Conceptually they are very similar, but in practice they are used very differently. Weather models use as much data as there is available to start off close to the current weather situation and then use their knowledge of physics to step forward in time. This has good skill for a few days and some skill for a little longer. Because they are run for short periods of time only, they tend to have much higher resolution and more detailed physics than climate models (but note that the Hadley Centre for instance, uses the same model for climate and weather purposes). Weather models develop in ways that improve the short term predictions, though the impact for long term statistics or the climatology needs to be assessed independently. Curiously, the best weather models often have a much worse climatology than the best climate models. There are many current attempts to improve the short-term predictability in climate models in line with the best weather models, though it is unclear what impact that will have on projections.”

What to make of this? I will dissect this paragraph line by line.

“Conceptually they are very similar……”

In practice, they are. However, as I have argued time and again, this apparent similarity is a serious defect. A crude representation of the ocean is all that is needed for a weather model, but in a climate model the ocean should share center stage with deforestation and other land use changes.

“Weather models …use their knowledge of physics to step forward in time.”

What Gavin leaves unsaid here is that most of the physics in a weather model deals with the atmosphere. Also, most of the physics is parameterized and the reliability of the parameterizations continues to be debated. I don’t want to pick nits, else I would query how models can possess knowledge of any kind.

“This has good skill for a few days…….”

Yes, Gavin is aware of Lorenz’ butterfly. He fails to state, however, that the average prediction horizon of weather forecasts is comparable to the lifetime of synoptic weather systems. I would not mind this omission, were it not for the fact that the (unknown) prediction horizon of climate models is determined in part by the life time of circulation systems in the ocean, such as the Pacific Decadal Oscillation. Since  weather models and climate models are conceptually similar, one must expect similar predictability problems.

“Because they are run for short periods of time only……”

The logic in this sentence is inverted. The development of weather and climate models is driven by the desire to employ the latest supercomputers available. It is conceptually a small matter to fill these computers with parameterizations operating at higher resolution. My interactions with Tim Palmer of ECMWF (see my weblog of June 24, 2008) focused on his claim for Seamless Prediction Systems. His advocacy boiled down to a quest for a computer facility that could run climate models at the resolution now feasible for weather models. I submit that no conceptual progress can be expected if the modeling community fails to reconsider the architecture of their software.

“Weather models develop in ways that improve…….”

This line ends with the need to independently assess the impact of model improvements on long-term statistics. I agree with the need, but not with Gavin’s off-hand way of letting this problem pass by without explaining how such assessments can or should be performed. Throughout this text Gavin avoids matters of methodology. That, to me, misleads all readers who are not professionals themselves.

“Curiously, the best weather models…….”

At this point, a Dutchman would say “Nu breekt mijn klomp” (now my clog breaks). Gavin Schmidt is a professional climate modeler, but he appears surprised that the climatology of weather models is inferior. Of course it is. Weather models deal with the atmosphere, climate models with the entire climate system.

“There are many current attempts to improve the short-term predictability …….”

Climate modelers are responding to public opinion and have chosen to develop “seamless” or “unified” prediction systems. The present skill of  seasonal forecasts is marginal at best; why should the public and their governments have confidence in forecasts many ten of years ahead? Conceptually, this is indeed a crucial question. It cannot be answered by increasing computer power. Gavin admits as much:

“….. it is unclear what impact that will have on projections.”

So why should one base climate policy on forecasts made by climate models? 

Curiously, Gavin’s text is conceptually vague. He should be able to do better.

It is up to you, Gavin. I am waiting.

Comments Off on Dissecting a Real Climate Text by Hendrik Tennekes

Filed under Guest Weblogs

Comments are closed.