Weather generators such as Cligen use a large database of historical
weather readings to produce a likely model of weather at a particular site
on a particular day. Because of the amount of data, the models can take a
long time to generate. Parallelization of the software could result in
significant speed-ups. The core equations involve summations and nested
polynomials, so I would expect to get fairly significant speed-up.
In addition to improving run-time, I would also like to design a graphical
front-end for the simulation. The front-end would not only display a
human-readable summary of the data, but would allow the user to tweak the
local conditions to see what small (or large) changes to the weather have
on the model.
This document was translated from LATEX by