200
N. Karvosenoja et al.
31.1 Introduction
Spatially resolved emission inventories are an important input for air quality models.
The quality of emission data depends on the uncertainties of several components
of the inventory, e.g. activity data, emission factors and spatial representation of
emissions. The uncertainties associated with activity data and emission factors in
different sectors have been relatively widely assessed at regional and national level
(e.g. [3, 5]). However, the impact of the accuracy of spatial disaggregation of emissions, i.e. gridding, has been less studied, but it has been identified to be potentially
considerable [6].
The required spatial resolution of an emission inventory depends on the scale and
impact end-point considered in an air quality model. With global to regional level
models that describe mainly the long-range transboundary effects, an emission grid
resolution of from 25 to 100 km is often considered to be adequate. As for regional
to local level modeling that describe also the impact of local emission sources, it is
commonly necessary to use spatial representation at grid resolutions down to 1 km or
finer. Especially for primary pollutants that show strong concentration gradients, e.g.
fine particulate matter (PM 2.5 ), gridding of emissions to fine horizontal resolution is
required for reliable impact assessment.
Gridding of emissions from disperse emission sources (e.g. road transport) are
typically performed using GIS-based spatial information, i.e. spatial proxies (e.g.
road network and traffic volume data) to represent the spatial distribution of emissions. Therefore it is crucial to select appropriate proxies for different emission
source sectors. Most commonly used proxies are road network data for road transport, industrial land use for industrial emissions and population density for a wide
variety of emission sources [7]. However, especially population density data rarely
represents adequately the spatial distribution of emissions at fine spatial resolution.
Zheng et al. [9] showed that the importance of more sophisticated proxy choices than
population density increased at spatial resolutions below 25 km.
Some of the more challenging sectors for emission gridding include residential
wood combustion and diesel machinery [7, 8]. For example, for residential wood
combustion information about the location and use of wood stoves is often scarce or
lacking. Another spatially challenging emission source category is diesel machinery
used in urban areas, which includes various types of equipment operating in construction and maintenance works at multitude of environments. These sources are
often spatially allocated in emission inventories using population density information as a proxy. This study presents an improvement of the gridding procedure for
machinery in the Finnish regional emission model FRES. The influence of using
more sophisticated gridding proxies than population density was demonstrated with
modeled population exposure results at 250 m grid resolution.
N. Karvosenoja et al.
31.1 Introduction
Spatially resolved emission inventories are an important input for air quality models.
The quality of emission data depends on the uncertainties of several components
of the inventory, e.g. activity data, emission factors and spatial representation of
emissions. The uncertainties associated with activity data and emission factors in
different sectors have been relatively widely assessed at regional and national level
(e.g. [3, 5]). However, the impact of the accuracy of spatial disaggregation of emissions, i.e. gridding, has been less studied, but it has been identified to be potentially
considerable [6].
The required spatial resolution of an emission inventory depends on the scale and
impact end-point considered in an air quality model. With global to regional level
models that describe mainly the long-range transboundary effects, an emission grid
resolution of from 25 to 100 km is often considered to be adequate. As for regional
to local level modeling that describe also the impact of local emission sources, it is
commonly necessary to use spatial representation at grid resolutions down to 1 km or
finer. Especially for primary pollutants that show strong concentration gradients, e.g.
fine particulate matter (PM 2.5 ), gridding of emissions to fine horizontal resolution is
required for reliable impact assessment.
Gridding of emissions from disperse emission sources (e.g. road transport) are
typically performed using GIS-based spatial information, i.e. spatial proxies (e.g.
road network and traffic volume data) to represent the spatial distribution of emissions. Therefore it is crucial to select appropriate proxies for different emission
source sectors. Most commonly used proxies are road network data for road transport, industrial land use for industrial emissions and population density for a wide
variety of emission sources [7]. However, especially population density data rarely
represents adequately the spatial distribution of emissions at fine spatial resolution.
Zheng et al. [9] showed that the importance of more sophisticated proxy choices than
population density increased at spatial resolutions below 25 km.
Some of the more challenging sectors for emission gridding include residential
wood combustion and diesel machinery [7, 8]. For example, for residential wood
combustion information about the location and use of wood stoves is often scarce or
lacking. Another spatially challenging emission source category is diesel machinery
used in urban areas, which includes various types of equipment operating in construction and maintenance works at multitude of environments. These sources are
often spatially allocated in emission inventories using population density information as a proxy. This study presents an improvement of the gridding procedure for
machinery in the Finnish regional emission model FRES. The influence of using
more sophisticated gridding proxies than population density was demonstrated with
modeled population exposure results at 250 m grid resolution.