Of course I’m talking about the science behind it, not the effects… Steve Mcentyre over at climate audit (www.climateaudit.org) has finally been given the source code that Steve Hanson used for determining temperature time series that are used by many, many climate scientists. Despite the fact that Hanson works for the GISS (a division of NASA) and that his numbers were being used to shape policy around the globe, he was reluctant to release his methods. I won’t get into what kind of shoddy science that is, but I have to give my 2 cents about his attitude. He has repeatedly said that climate science should be left to the professionals. that amateurs are more likely to be steered by agendas and not by science. Well, OK, I can see some of that argument, but the trouble is that most of his “science” isn’t climate based at all, it is statistical. He wasn’t just making theories based on the data, he was creating the data as well. After Steve reverse engineered some of his methods and showed how Hanson needed to correct figures, Hanson had little choice but to release his code.
By all accounts, it is a statistical nightmare. What Hanson was doing was using several sets of data and combining them, “adjusting” assumed biases in the measurements, and glossing over site irregularities all the time admitting that he was not a statistician. Folks, there are some serious problems with this, if he had only taken his own advice! He should have had statistical professionals go over his numbers and then he could use that data to theorize on what it means.
Some of the code is over twenty years old! It has routines for writing output data, REWINDING THE TAPE, and then using that output as an input for the next calculation. Fortran anyone? Anyway, there are some things in there that even I could see would cause problems. Converting from floating point calculations to integer in Fortran leads to some wild rounding errors, “correcting” obviously flawed measurements instead of throwing them out, and his method of calculating running averages is…. interesting. A recent study (also on climate audit) showed that only 1/3 of the measurement stations actually met with the standards that ensure accurate readings. There’s a lot of things wrong, and this is the most often used set of numbers for determining if the planet is warming up. Keep in mind that most theories about CO2 driven global warming expect around a .7 increase due to man’s CO2 contribution. SOme of the things that have been uncovered in Hanson’s code show that he may have overestimated the warming by .2-.4 degrees! I highly recommend the site, the guys really know their statistics and they are doing a great job of trying to get less biased data to work with for climate change.