Estimating future air quality under the warming climate is an urgent task for all populated regions. Often, climate models are evaluated with respect to air temperature and precipitation, but without a focus on other air quality-related meteorological variables. This study evaluated a regional ensemble system over the southeast Australian region driven by five selected CMIP6 global climate models (downscaled by two regional models, making the ensemble size ten) in terms of a range of surface variables relevant for air quality from seasonal to diurnal timescales. Results showed that the two regional climate models, although only differing in their planetary boundary layer (PBL) parameterizations, performed quite differently. In general, the regional model with the MYNN2 PBL scheme (named R3) performed better than the other. While most meteorological variables, including surface wind speed, were verified well, wind direction showed large biases and variability among models. When downscaled (~4 km resolution) atmospheric variables were applied to drive the Community Multiscale Air Quality (CMAQ) model, the ensemble members, particularly the two versions of the regional model, resulted in different chemical species concentrations. A model ranking scheme was developed based on various spatiotemporal timescales and identified slightly superior performance by the regional model R3. The findings provide a valuable reference for selecting optimized model members for future air quality projections.
Cheung et al. (2026) studied this question.