Sunday, November 29, 2009

A Scientific Experiment

Experiments

Step three of the Scientific Method is to perform an experiment to test your Hypothesis. That sounds easy but it is anything but. To have a valid test you must control all but one of the variables. In the laboratory and for a basic experiment that may be possible but once you’ve opened the door and gone into the real world it can get complicated very, very quickly.

Let’s say you wanted to test the Hypothesis: “Increased temperature will cause Joshua trees to grow more robustly as measured by the thickness of the rings in the trunk at chest height.”

First you must control for all the things that might affect the growth of the tree: water, amount of sunlight, gases in the air, nutrients in the ground, organisms in the ground, etc. In the lab you might be able to start with a closed cabinet in which you’ve placed a sterile loam soil. Controlled amounts of nutrients (fertilizer) could be provided. Controlled amount of water could be provided. The amount of “sunlight” can be controlled by artificial lights. The very air within the cabinet can be controlled by hermetically sealing the dang thing. Make ten or more of these cabinets all of them identical in content. Use genetically identical seeds or cuttings of the tree species to be tested. NOW you can vary the temperature in each of those cabinets so each is, say two degrees C different ranging from 0 degrees C on up to 20 degrees C. To ensure things are truly like the natural world, you turn all the temperatures down in all the cabinets to -5 degrees C for six months out of every twelve.

Oops, better do three of each--just in case.

So you’ve got 36 or so identical cabinets in which everything is identical except the temperature. There are three that are at 0 degrees C, three at 2 degrees C, three at 4 degrees C, etc., etc., up to three at 20 degrees C. Easy.

Now all you have to do is monitor your cabinets for a couple of years for your trees to grow several new rings. Then you take core samples from each tree (or cut them in half to take a disk out) at chest height and measure the thickness of those rings. Did the trees produce thicker rings (more growth) at higher temperatures?

Regardless of your results, you submit your paper describing your experiment and its results to a peer reviewed journal so others can examine your work and duplicated it if they so desire.

That’s the way it’s supposed to be.

Now, take your experiment outside. Can you control the numerous variables? Unlikely. The genetic variability in the trees alone is a problem. Add the variability of soils, sunlight, yearly differences in precipitation, disease and insect infestation, etc. and you’ve got one heck of a head ache.

Yet that is precisely what some of the scientists at the Climate Research Unit have tried to do in reverse. They are trying to say that increased growth in tree rings is indicative of higher temperatures. (Or, conversely, smaller growth rings are indicative of colder temperatures.) From this they have constructed a computer model to predict the future.

The problem is, as I understand it, they have used a small sample of trees from a limited number of very localized sites. A group of trees, it must be noted, that confirms their hypothesis when certain “corrections” have been made. (Larger samples were available and other researchers have used them to refute the hypothesis.) And they then produced a computer model that can’t account for recent past temperatures. Oops!

UPDATE: About those computer models: Climategate Computer Codes Are the Real Story
"I think there’s a good reason the CRU didn’t want to give their data to people trying to replicate their work.

It’s in such a mess that they can’t replicate their own results."





No comments: