Why does the program continue to use more disk space when the results are being sent out regularly?
(Look out for the idiot....ME!)

It's a very reasonable question, but the answer is that the amount of data generated in the course of a model run is huge, and only a small part is ever returned to the project routinely. As the phases run sequentially, and each uses data generated earlier, so the files grow.

1. The regular trickles contain very little data - essentially only a report on how far you have got.

2. At the end of phases 1 and 2 (one third and two thirds of the way through) there is a 'mega trickle' which contains some data (and enables the generation of the temperature and precipitation graphs on your results page, for example).

3. At the end of the run there is a more substantial package of data (8MB?) which gives the scientists information on a number of variables, which indicate how your model did in comparison to others.

Occasionally the scientists ask for more data in order to study a run more fully, and they have been doing this recently in relation to some recent runs in classic CPDN. You also need the full data if you want to study it yourself, and there are some tools (eg Martin Sykes' CPView) for this. It is more interesting to be able to compare more than one run. For these reasons you should keep the folders of models already completed - saving onto CD/DVD is fine if you are short of HD space.

There is a lot of information on the study on the PHP board and the website if you are interested - ask for pointers if you want to follow it up.