I\'ve looked around the site, but haven\'t found an answer to this question:

Why does it run models for dates in the past? Both of my currently running models are for 1974 and 1976. Why is it modeling something that already happend?

It seems illogical to me, but surely there is some good reason?

Disclaimer: I know virtually nothing about climate modeling, but I am curious.

Thanks!

Welcome to the project Alec

Basically, the early years are for calibration purposes to check that the values used to start each model are producing a valid result.

You can find some information about the science and the models from near the front of this web site.
Start here, then use the blue menu on the right to go to: Climate science, from where you can read about the basics of climate science, and also about the Experiment Strategy.
The part that you\'re asking about is in Experiment strategy - the basics, with more detail in Experiment strategy - advanced

PS
Some of the climate models actually run mostly in the past, to research certain events.
Starting again from the front page, and this time clicking on Experiments, you can see a short description of some of the past, present, and future experiments, with a link in each to a fuller description.

Look at: Seasonal Attribution Experiment for some research on the United Kingdom floods of Autumn 2000, (now finished),
and at: Millennium experiment, still in testing, which will start way back in the year 800, and run for 1200+ years to the present.

These huge models are expected to take perhaps 4 months on a fast desk top, running continuously and doing nothing else. A lot of long time modellers have been waiting, slightly impatiently, for a year or so for these.

---

Backups: Here

Hi Alec

The HadAM3P models which you are running will soon have an extension to run data for very recent years using real-world data values soon after they\'ve been produced; extra two-year periods will gradually be added. The beta version is running well on Windows but still isn\'t right on Linux so I hope we\'ll see the extension on this main project early in 2010.

Model results have to be checked against real-world values to see how accurate they are. Of course that\'s just one of their functions.

---

Cpdn news