The study is concerned with the significant and historically well-known problem of winter air pollution in Christchurch and its vicinity (especially associated with particulate matter-PMlO) during nights dominated by stagnant synoptic weather situations. This pollution results from active use of domestic log-burners and open-fires, on nights of strong near-surface temperature inversion. In this study, a numerical modelling research approach is used with the application of a regional atmospheric model (Mesoscale Model, 5th generation-MM5) and an air quality-air pollution model (Comprehensive Air quality Model with extensions, version 4-CAMx4), coupled together for more sophisticated investigation of space-time PMlO dispersion over the Christchurch region during heavy smog nights. In the thesis, the theoretical and practical background of limited area models is described, with reference to basic limited area meteorological models such as MM5, RAMS and WRF. The potential of limited area meteorological modelling (MM5) , air pollution chemical modelling (CAMx4), and the use of coupled numerical systems (MM5-CAMx4) for numerical investigations are also discussed. The basic types of coupled meteorological-chemical models, the application of the main limited area models, and the level of precision of the various modelling systems are briefly discussed. The factors controlling atmospheric circulation over Canterbury and the Christchurch region are described, and their influence on heavy night-time air pollution in Christchurch and its vicinity, with particular reference to the winter situation and the possibility of prediction and control of the aerosol air pollution. The study analyses input data for numerical modelling, including global analysis data, the CAPS2000 and winter 2003 field experimental sites, and particulate monitoring observations. Several methods of MM5 modelling are applied to replicate the local air circulation over Christchurch and its environs: single run modelling, series of runs with different levels of nesting with use of the CAPS2000 observed data for MM5 fine tuning. Particulars of 2003 winter MM5 modelling using global analysis input data, and Meteorological Service input data are covered separately. The numerical modelling system (meteorological-chemical) was created with application of CAMx4 as a basic air quality model. The procedures for assimilation of MM5 data as basic input meteorology for the chemistry model, and use of particulate matter monitoring data were investigated. Different versions of MM5-CAMx4 optimal combination have been compared, and quasi-operation possibilities of the MM5-CAMx4 modelling system have also been considered on the basis of winter 2000 and winter 2003 PMlO experimental modelling.