Accuracy of computer modelling depends on multiple sets of data and figuring out mathematical ways to combine them. The air flow models developed at the Geography Department, University of Canterbury, use geographical and topographical data and measurements of pollution, given certain conditions. Professor Simon Kingham and Associate Professor Peyman Zawar-Reza discuss how models are continually refined to make them better.
PROFESSOR SIMON KINGHAM
The models rely on two sources of data. They rely on very good meteorological data, so weather data: they rely, to some extent, on good topographical data, in terms of the shape of the land; and the third form of data, is data about emissions, so how much pollution is being produced. So how much air pollution is coming out of cars, how much air pollution is coming out of chimneys, how much air pollution is being produced by industry? So Peyman's model was able to use… because we had all the good quality emissions data, meteorology data and topographical data, he is able to run very accurate models and accurately predict pollution values for Christchurch. For other parts of the country, that information is not available, and secondly, it uses quite a lot of computer power to run those models.
ASSOCIATE PROFESSOR PEYMAN ZAWAR-REZA
Initially, the models did have a lot of errors, but as time has gone in the past 20 years, there has been an enormous amount of effort trying to build models that are more accurate. And these models are constantly refined and reformulated, reworked, to try and make them better and better.