San José State University
Thayer Watkins
Silicon Valley
& Tornado Alley

Some History of Climate Computing

While researching material on the history of climatology I came across an article in Science of May of 1982 written before the time journalists accepted without reservation the use of computer models in climatology. (Before they drank the koolaid.) The full title of the article was

Global forecast: increasing
temperatures, rising seas,
and more rain.
Chances of accuracy: 0 to 100%

The healthy skepticism of that last line is refreshing. The author was Steve Olson.

The article focuses on two researchers. One is Syukuro Manabe, a well respected meteorologist at the Geophysical Fluid Dynamics Laboratory in Princeton, New Jersey. The other is James (Jim) Hansen, the director of NASA's Goddard Institute for Space Studies.

Hansen is identified in the article as a climatologist but his actual training was as a physicist. The author of the article was probably not aware of this. Hansen wrote his dissertation on the greenhouse effect for the planet Venus. Venus has an albedo (proportion of solar energy reflected) of 0.65 to 0.75. The Earth's albedo is 0.35 to 0.40. Because of the high albedo of Venus less solar energy is absorbed on Venus than is for the Earth. Nevertheless the surface temperature of Venus is above 400°C, hot enough to melt lead. The atmosphere of Venus is 96 percent carbon dioxide. Hansen's dissertation objective was to show how the greenhouse effect from the atmosphere could account for the high surface temperature despite the solar energy inflow being less than for Earth. Of course, with such an atmosphere and sufficiently high absorption of infrared radiation from the surface such a high temperature could explained. What Hansen apparently never considered is that the high surface temperature could also arise from the reflection of the infrared radiation off the under side of the clouds which totally enclose Venus 100 percent of the time. When a greenhouse gas absorbs radiation it reradiates it half upward and half downward. On the other hand the clouds may reflect back down 75 percent of the radiation. This is the cloud blanket effect and it is more potent at raising the surface temperature than the greenhouse effect. Apparently Hansen never even considered the cloud blanket effect. Yet most people know that a clear winter night is colder than one with cloud cover despite the greenhouse effect being the same on the clear night as the night with cloud cover. Apparently Hansen's tunnel vision concerning carbon dioxide takes precedence over everyday experience. Conditions on Venus are in all likelihood the result of a combination of a greenhouse effect and a cloud blanket effect. If Hansen's dissertation explained the temperature as due entirely to a greenhouse effect then the parameters he used were wrong.

Syukuro Manabe was well aware of the difficulties of building a model that adequately represents the Earth's climate system. Manabe characterized the climate models such as those Hansen was developing as being very primitive. Manabe is known for his efforts to include such things as atmospheric convection in meteorological models. Hansen, on the other hand, was content to use a modified version of his Venusian work. It seemed to be the case of computer jockies rushing in where trained meteorologists hesitated to tread. The article emphasized how difficult it was to get the models to replicate the seasonality of Earth's climate.

Hansen put together a model involving carbon dioxide but also volcanic erruptions and other factors. He felt that he had successfully replicated the average global temperature data of the past hundred years. Manabe commented on this

[...] the data Hansen uses are very nebulous, very controversial. [...]and how much of each ingredient he is putting in is not clear. {...] The data he had to use are simply not good enough to convince everyone that he has simulated past temperatures.

The phenomenon that makes pernicious the entry of computer jockies into fields where the trained professionals hesitate to to tread is that the models have a large number of parameters which can be tweaked until a desired result is obtained. The modeler then announces to the world that this result is the result of scientific modeling. Instead the agenda-driven result determined the parameters of the model. Often the models are so complex that no one can determine what the parameters have to be to get the desired result. Instead the model, by a process of trial and error, is tweaked and tweaked until the desired result appears. Then the so-called research ends and the result is reported to the world as the end result of scientific enquiry. This is very likely the process by which the infamous hockey stick graph of global temperatures was derived. Some of the tweaking in that case were simply gross mathematical errors. This process of tweaking a model until a desired result is achieved subverts scientific inquiry. Instead of independent data X giving rise to result R, it is a matter of desired result R generating the supposed data X.

(To be continued.)

HOME PAGE OF applet-magic
HOME PAGE OF Thayer Watkins