Combining deterministic and statistical models for ill-defined systems: Advantages for air quality assessment

Uncertainty pervades the description of ill-defined systems and the data collected from the for model development. Young (1) has used a systems theoretic framework to espouse a general theory of modeling based upon the scientific method to cope with uncertainty. We show a hybrid deterministic/statistical approach consistent with this general theory can be used when such systems have a phenomenological property which can be simply characterised. The methodology is especially relevant to the assessment of air quality systems and details are provided of a comprehensive program within the Centre for Resource and Environmental studies (CRES) to develop a suite of algorithms for predicting the probability distribution of ambient pollutant concentrations from a range of emission sources.