In this paper we present a methodology that has been used to give an insight into the impact of climate change on ozone levels in Sydney in 20 and 50 years time. The methodology comprises a dynamical downscaling system which takes 200 km resolution global climate simulations and through a two stage process, generates 3 km resolution mesoscale meteorological and trace gas concentration fields over populated areas. The system was assessed for Sydney and was found to perform well in the prediction of the historical ozone climatology, mesoscale meteorology and peak ozone concentrations. The system was used to downscale an A2 climate scenario for 2021-2030 and 2051-2060. Although the simulated ozone climatology did not change significantly, increases in ambient temperature and the resultant increases in ozone precursor concentrations and photochemical transformation rates lead to increases in the frequency and magnitude of peak ozone concentrations and health impacts. The tools that were developed and assessed in this project are intended to provide a capability which can aid policy makers in formulating long term air pollution policies where the impact of climate change has to be considered. However, when applied for this purpose, it is recommended that the system be operated in an ensemble mode whereby a range of model projections are generated and an estimate of likelihood can be calculated. It is further recommended that the system be enhanced to consider the formation and fate of fine particles (primary and secondary) as this air pollutant is generally considered to cause the largest air quality related health impacts in Australia

This paper presents results from a study to investigate the extent to which NO(2) data from ambient network monitoring, air quality modelling, or a combination of both, can improve estimates of personal exposure across a city. As it is not practical to measure the personal exposure of every individual, a common assumption in most epidemiological studies for urban areas has been that people are exposed to a spatially-homogenous pollutant, ignoring variations in concentrations across an airshed and in various micro-environments. Our conceptual model of an individual's personal exposure to NO(2) is based on time-weighted sums of exposure in the microenvironments of home, transit and work. Personal exposure in each microenvironment is linked to ambient concentration by indoor-outdoor concentration ratios. To allow us to both develop and evaluate the model, we designed a measurement program involving volunteers across Melbourne wearing personal passive samplers Participants also wore additional samplers for sub-periods of each 48-hour exposure, at home, at work and in transit. Diaries were designed to record details of time and activities in each microenvironment, especially those associated with cooking and ventilation. Three methods of estimating indoor-outdoor ratios and three approaches to calculating ambient exposure were evaluated. For estimation of the personal exposure to NO(2) of a large number of people, it is recommended that best results would be obtained with the I/O ratio calculated from a mass balance method. This requires participants to record daily gas cooking periods and approximate house age, although a simpler but slightly less accurate method dependent only on the existence or not of a gas cooking appliance also produces satisfactory results. The recommended method for calculating the required ambient outdoor concentration is to use values from the network monitor nearest to a person's microenvironment. Evaluation statistics were considerably poorer for a commonly-used method whereby each person is assigned the same ambient concentration, taken to be the mean concentration across all network monitors.

Perhaps the weakest aspect of epidemiological studies is the exposure component. How accurate is our assessment of the dose received by a person or population on short- or long-term scales? Most epidemiological analyses involving air quality use data from a fixed location monitor(s), with the implicit assumption that pollutant concentration is spatially uniform across the study domain. This is not the situation in reality, nor do individuals remain at the one location, even over short periods. We have developed a methodology which takes account of spatial variation in air quality and reduces uncertainty in ambient exposure estimates. This approach, using elliptical influence functions, involves the blending of observations from a monitoring network with gridded meteorological and pollution fields predicted by the complex air quality model TAPM. Examples from exposure fields developed on a 1.5 km-spaced grid for each day of a 6-year period (1998-2003) for Brisbane will be shown.

In recent years childhood asthma has increased. Although the precipitants of childhood asthma are yet to be established possible contributing factors are local ambient air pollutants. This study aims to assess associations of regional ambient air pollutants on emergency department childhood asthma presentations across four regions of the city of Melbourne, Australia. Daily emergency department (ED) presentations for asthma in children were studied for the years 2000 and 2001. Estimates of local air pollutant levels were obtained using simulation modelling techniques. Generalized Additive Models were used to examine associations between combined local levels of air pollutants and childhood asthma ED presentations adjusting for seasonal variation, day of week effects, and meteorological variables. There was consistent associations between childhood ED asthma presentations and regional concentration of PM10, with a strongest association of RR = 1.17 (95% CI 1.05 to 1.31) in the central district of Melbourne. NO2 and Ozone was associated with increased childhood asthma ED presentations in the Western districts. This study suggests that regional concentrations of PM10 may have a significant effect on childhood asthma morbidity. In addition, ozone may play a role however, its effect may vary by geographical region.

Meteorological simulations have been performed with two different mesoscale models: MM5 (PSU/NCAR-USA) and TAPM (CSIRO-Australia). Both models have been run under two different meteorological situations, typical for summertime and wintertime, allowing the models to be tested under different synoptic forcings. The study region for these mesoscale simulations is located in the interior of Catalonia, Spain, north of Barcelona, and can attain high pollution episodes despite not including important pollutant sources within the domain. Simulations have shown that the origin of the summertime pollution peaks in the region is the advection of air masses loaded with pollutants from the densely-populated region of Barcelona. Comparison of meteorological fields obtained with the two models have shown that, despite the much smaller computational resources needed by TAPM and being almost 10 times faster than MM5 in simulation time, both gave similar results, demonstrating thus the usefulness of TAPM as an alternative tool for air pollution management.

For the prediction of fog, INM is currently using a 1-D version of the model HIRLAM. In this model, terms which are dependent on the horizontal structure of the atmosphere are estimated from the output of the operational run of HIRLAM at a resolution of 0.2º. However, at this resolution, katabatic winds are not always well reproduced by the 3-D model and, consequently, are not reproduced by the 1-D model either. In order to fix this problem, and under some conditions, forcing from the 3-D model are substituted by others of climatologic origin, estimated from a conceptual model of katabatic winds developed for the region. In order to check the quality of the conceptual model, a simulation has been conducted with a mesoscale model at high resolution. The model used was the Australian model TAPM, and a grid spacing of 2-km was used for the innermost of its nested domains. The simulation was able to reproduce very well the generation of katabatic winds in the region, and has confirmed the main characteristics of the circulatory patterns described in the conceptual model. In particular, the simulation has identified areas of convergent flows, with upward movement of air, close to the airport of Madrid-Barajas. GEOGRAPHIC FRAMEWORK AND CONCEPTUAL MODEL OF KATABATIC WINDS The Madrid airport is located in the centre of the Iberian Peninsula. The surrounding region is characterized by the presence of several mountain ranges and river valleys. The main valley corresponds to the Tajo river and has a general orientation NE-SW, channelled by the Central and the Iberian Mountain Ranges. Four tributaries merge in the lower part. Figure 1 shows the complex orography of the region, with mountains well over 2000 m above sea level. Studies conducted at the INM have shown that the development of mountain breezes is an extended phenomenon in this region. In particular, down slope winds (katabatic winds) due to differential cooling over complex orography.