Modelling of indoor exposure to nitrogen dioxide in the UK

Abstract

A dynamic multi-compartment computer model has been developed to describe the physical processes determining indoor pollutant concentrations as a function of outdoor concentrations, indoor emission rates and building characteristics. The model has been parameterised for typical UK homes and workplaces and linked to a time-activity model to calculate exposures for a representative homemaker, schoolchild and office worker, with respect to NO₂. The estimates of population exposures, for selected urban and rural sites, are expressed in terms of annual means and frequency of hours in which air quality standards are exceeded. The annual mean exposures are estimated to fall within the range of 5–21 ppb for homes with no source, and 21–27 ppb for homes with gas cooking, varying across sites and population groups. The contribution of outdoor exposure to annual mean NO₂ exposure varied from 5 to 24%, that of indoor penetration of outdoor air from 17 to 86% and that of gas cooking from 0 to 78%. The frequency of exposure to 1 h mean concentrations above 150 ppb was very low, except for people cooking with gas.

Introduction

The real health effects of air pollution depend on the concentrations experienced by people rather than those recorded by stationary air quality monitors located outdoors. Two key features of people which strongly influence their exposures are their mobility and the time spent indoors (Loth and Ashmore, 1994). Recent estimates suggest that the average proportion of time spent indoors by the population in developed countries is about 90%, with considerable variation between individuals (GB Parliament House of Commons Environment Committee, 1991). Although the standards designed to protect the population from exposure to pollutants traditionally derive from ambient concentrations, the relation between these and personal exposure is not well defined. Effective strategies to reduce human exposures to air pollution should consider the contribution from both outdoor and indoor sources. Thus, recent policy encourages more research on the role of indoor pollution and the assessment of the total exposure to pollutants (DoE, 1992, DoE, 1995; IEH, 1996).

This paper is focussed on exposure to nitrogen dioxide. Several monitoring studies have measured indoor NO₂ concentrations in order to understand the contribution of indoor sources to human exposure. These studies show that indoor NO₂ concentrations may be substantially higher and more variable in homes with gas cookers, and gas or kerosene space heaters, than in homes without these appliances (e.g. Spengler et al., 1979, Spengler et al., 1983; Quackenboss et al., 1982; Marbury et al., 1988; Melia et al., 1990).

In the United States, personal exposures of both individuals and populations have been extensively studied over the past 15 years, and have been supported by work in the indoor environment. Personal exposure has been most intensively investigated in a series of large-scale field NO₂ exposure studies (Ryan et al., 1988a, Ryan et al., 1988b; Spengler et al., 1994). The objectives of these direct studies were to quantify the relative influence of indoor and outdoor concentrations on personal exposure, based on activity patterns, personal and household characteristics and other seasonal and spatial variables.

Direct personal exposure measurements are expensive and technically difficult, and applicable only to samples. An alternative approach to estimating population exposures is to use appropriate computer models. A key concept developed for such modelling work in the US is the microenvironment (ME), defined as a generic location with homogeneous pollutant concentrations where people spend time (i.e. offices with smoking activity or a living room in a house with gas cookers) (Duan, 1982). To estimate exposures, two types of models are needed: (i) physical models, to predict the pollutant concentrations in different MEs, (ii) exposure models, to simulate the movement of individuals between these MEs through time (NRC, 1991).

Most of the models developed to predict exposures to date use empirical data on indoor/outdoor ratios (I/O) to determine indoor ME concentrations as a function of time (e.g. Sexton et al., 1983; Noy et al., 1986; Lee et al., 1998). Furthermore, the current models do not link activity patterns to mechanistic modelling of ME concentrations. Therefore, they are not able to separate the role of indoor and outdoor sources, or to predict the effects of changing emission rates both indoors and outdoors.

This paper describes a new modelling approach, which incorporates the description of physical processes. In this way, it allows the relative importance of key factors influencing personal exposures to be assessed in different situations. This model can be used to distinguish three types of exposures: (i) outdoor exposure, (ii) indoor exposure resulting from penetration of outdoor air and (iii) indoor exposure resulting from indoor sources. In the current study, the model predicts the above types of personal exposures, for a representative homemaker, schoolchild and office worker, based on indicative activity patterns. The model predictions for annual mean exposures to NO₂, and the proportion of hours in which short-term air quality standards for NO₂ are exceeded, are presented and discussed.