Monte carlo simulation of regional aerosol: transport and kinetics

Published in Air & Waste Manage. Assoc: Schichtel, B.A.; Husar, R.B. J. of Air & Waste Manage. Assoc. 1996, 47, 331-343.

Regional Simulation of Atmospheric Pollutants with the CAPITA Monte Carlo Model

by
Bret A. Schichtel and Rudolf B. Husar

Center for Air Pollution and Trend Analysis (CAPITA)

Campus Box 1124

Washington University

St. Louis, MO 63130

September 25, 1995

ABSTRACT

A Monte Carlo model for the simulation of regional scale transport, transformation, and dry and wet removal is presented. The model was newly re-designed in a modular framework, separating the emissions, transport and kinetics calculations. The transport module employs a quantized Monte Carlo technique for the simulation of atmospheric boundary layer physics. Kinetic processes are simulated using rate equations where the rate coefficients are dependent upon meteorological variables, and thus vary in space and time. The rate coefficient equations are determined via a tuning process in which simulated values are compared to observed measurements. Results from simulations of and over the eastern US are presented for 1992. Comparisons of simulated daily concentrations to observations had r² = 0.35 - 0.83 depending on season and location, while for weekly wet deposition rates r² = 0.5 - 0.95. The utility of the model to investigate the source receptor relationship is demonstrated by explicitly examining the role of emission rates, transport, and kinetic processes in the attribution of sulfur dioxide and sulfate at a receptor in Massachusetts during the summer of 1992.

INTRODUCTION

During the 1970’s, air quality measurements demonstrated that pollutant emissions could be transported over large distances, significantly influencing distant receptor concentrations. The realization of regional scale transport of pollutants led to numerous air quality networks, models, and large scale field studies to relate the contributions from near as well as distant sources to receptor concentrations, i.e. to quantify the source receptor relationship. Many of the initial studies focused on sulfur oxides, and were designed to attain better understandings of the physical/chemical processes governing regional scale transport. For example, in Europe, the OECD established the Long Range Transport of Air Pollutants (LRTAP) program in 1972 (1), and in the US, the Sulfate Regional Experiment (SURE) was conducted (2), along with numerous plume studies.

Associated with these studies were the development of models. The initial models tended to be simple, using aggregated and parameterized meteorology and rate coefficients to simulate the physical/chemical processes responsible for regional transport. These included trajectory models (3, 4), statistical models (5, 6), as well as simple diffusion and box models (7, 8). As the understanding of the atmospheric processes increased, these types of models were generalized to include processes such as non-linear chemistry and atmospheric diffusion (9, 10).

During the 1980’s, with the concern over acid rain, substantial effort was devoted to the development of elaborate Eulerian models, such as the Regional Acid Deposition Model (RADM) (11) and the Acid Deposition and Oxidant Model (ADOM) (12). These models attempt to simulate atmospheric processes, such as in-cloud oxidation, with great detail. However, the data and computer resource needs have restricted their use to research activities with runs limited in time and space. Also, it is difficult to separate the contributions from the various physical/chemical processes involved in the source receptor relationship (13). As a result these models tend to be operated in a “black box” approach.

In this paper, an IBM-PC based Monte Carlo model for the simulation of regional scale transport, transformation, and removal of atmospheric pollutants is presented. This model is an extension of the CAPITA Monte Carlo model developed in the 1980’s (9, 14). The model is “data driven,” and was designed to take advantage of multiple meteorological, emission, concentration, and deposition databases that are available from various data suppliers. Beyond the calculation of ambient concentrations and deposition fluxes, the model was designed to be used as a diagnostic tool for the investigation and quantification of the physical/chemical processes governing the source receptor relationship, as well as for the estimation of unknown emission fields given ambient concentration data.

The paper presents the simulation methodology, along with the underlying physical and chemical processes. The suitability of the model for regional scale simulation is shown through the simulation of sulfate and the total sulfate wet deposition over the Eastern US. Also, the utility of the model to investigate the source receptor relation is demonstrated by examining the role of emission rates, transport, and kinetic processes in the attribution of sulfur dioxide and sulfate at a receptor in Massachusetts.

METHODOLOGY

Monte Carlo Approach

In the Monte Carlo approach to air pollution simulation, individual quanta or particles are subjected to the same physical and chemical processes and events in the computer as in the physical world. It can be thought of as a direct simulation of atmospheric pollutants. This approach contrasts with more customary numerical simulations in that it does not require explicit conservation equations and their solution as the basis for the simulation.

The Monte Carlo approach to the simulation of atmospheric pollutants has the following key characteristics. First, each emitted quantum contains a fixed quantity of mass of various pollutants based on the source's emission rate. Individual quanta are then subjected to transport, transformation, and removal processes. Mass conservation is maintained at the quantum level by accounting for the mass that has been chemically transformed and physically removed as an assembly of quanta travel through three dimensional space (Figure 1). An airmass is depicted as an ensemble of multiple quanta. The dispersion is represented by the spread of the quanta arising from the integration of the equations of motion, where horizontal and vertical mixing are based on mean and random wind components. This approach can be considered to be a Lagrangian simulation since the trajectories and the fate of individual quanta are followed. In the limit of infinite quanta, this approach results in a stochastic solution of the dispersion equation, yielding results equivalent to an Eulerian solution to the governing equations.

In the Monte Carlo approach, the simulation of the meteorological dispersion of pollutants proceeds independently from the nature and chemistry of the pollutants. This is based upon the assumption that meteorology is not significantly influenced by trace constituents in the atmosphere; however, it is understood that kinetic processes are dependent upon the pollutant transport. Consequently, the calculation of the dispersion of the pollutants can proceed independently from the calculation of kinetic processes. This allows for the separate examination of the influence of the transport and kinetics processes upon receptor concentrations.

Monte Carlo simulations of atmospheric processes were developed in the 1970’s (15, 16). However, this technique lost favor in the 1980’s as other approaches were pursued. Due to the multiple benefits of this modeling approach, it regained popularity during the 1990’s for mescoscale simulations (17, 18, 19), and is believed to offer considerable improvements over conventional Gaussain-plume models (20). The use of the Monte Carlo technique for regional scale simulations has received less attention with little development since the initial studies were conducted by Patterson et al. (14) and Husar (9).