INTRODUCTION

Ranjeet S. Sokhi

1.1 Advent of Mesoscale Models for Air Pollution Applications

Over the last four decades, mesoscale modelling of meteorology and air pollution has developed into an essential tool for research and the scientific investigation of atmospheric processes and for designing emission control strategies to protect human health and the environment (Anthes and Warner, 1978; Shuman, 1989; Dudhia, 2014). Developments in computational methods and techniques, particularly in numerical weather predictions (NWP), has provided the opportunity to employ sophisticated modelling approaches for operational and policy applications (McGregor et al., 1978; Tapp and White, 1976; Grell et al., 1994).

The scale of the air pollution problem has moved away from considering only the impact of a local source or even emissions just within a city region. The appreciation that air pollution at a given point is determined by a combination of local, urban, regional and even global influences has grown with our understanding of atmospheric processes and source contributions (e.g. NRC, 1991; Hov et al., 1978; Rao et al., 1997 and 2003). While measurements play a key role in monitoring the current state and the past trends of air pollutants, they also have a key role in model evaluation and development (e.g. Porter et al., 2015). Multiscale mesoscale models now offer a number of unique capabilities, which include providing comprehensive spatial and temporal description of air pollution, providing the ability to analyse scenarios, explaining changes in pollutant concentrations and their behaviour in terms of the underlying atmospheric processes and producing forecasts of air pollution in the short and long term.

Mesoscale models for air pollution applications have a history that intertwines developments in numerical weather prediction and air pollution dispersion modelling approaches. Since the 1970s, these two stands have gradually merged with stronger cooperation. While simpler modelling approaches still have a role in air quality assessment, there has been a greater shift towards full process-oriented 'comprehensive' or 'one atmosphere' models (e.g. Russell, 1997). These types of models were a major shift, for example from simpler Gaussian approaches, and include more sophisticated treatment of meteorology and chemistry as well as emissions processing. Earlier models were developed to study long-range transport leading to acid deposition (e.g. Elliassen, 1980; Fisher, 1982) and photochemical pollution affecting urban areas (McRae et al., 1982; Seinfeld, 1988; Scheffe and Morris, 1993; Peters et al., 1995).

A significant advantage of mesoscale models is their modular and flexible structure. The main components involve the following:

(i) sources and emissions – anthropogenic and natural including biogenic species

(ii) meteorological processes, such as advection and diffusion, land-surface exchange, radiation, turbulence, microphysics and precipitation

(iii) geophysical processes involving terrain and land cover;

(iv) chemical transformations, such as, gas, aqueous and heterogeneous reactions and pathways

(v) physical transformations including deposition and aerosol processes

(vi) numerical solvers

(vii) analysis and visualization tools

The term 'model' implies a single computer software or a core set of algorithms that lead to a mathematical prediction of a quantity after taking account of the inputs, governing processes and any post-processing to deliver the output fields. However, in the case of mesoscale models, a more appropriate term would be 'mesoscale modelling system' as they are a collection of models and modules for physical and chemical parameterization schemes. While they offer considerable flexibility and robustness of their scientific basis, the number of components within a system also adds to their complexity and hence requires a greater level of expertise to operate them effectively for analysis air pollution problems.

1.2 Growth of Air Pollution Mesoscale Models

The move towards more sophisticated process models has been motivated by the need for more reliable predictions taking account of complex non-linear dynamical and chemical processes and interactions that affect air pollutants in the atmosphere. For example, emission rates of biogenic species will be affected by meteorological conditions and have to be calculated and integrated into the emission inputs before running the model. Terrain characteristics will affect meteorological fields, and both land cover and meteorology will influence surface exchange processes. Meteorological interactions with chemistry also have to be treated within models, for example, where clouds affect the radiation in the atmosphere (Ramanathan et al., 1989; Kylling et al., 2005) and hence change photolysis rates (Thompson, 1984; Madronich, 1987).

There are numerous...