358
J. Mak et al.
AERMOD will not underestimate maximum concentrations but will sometimes overpredict as high as by a factor of 5 [5, 10]. The occurrence of maximum concentrations
is often at times of low-wind speeds (when dispersion is poor). During these stagnation conditions, model errors have been over-emphasised utilizing both AERMOD
[8] and CALPUFF [3].
Dispersion models are more adequate at predicting longer-timed averaged concentrations (e.g. annual) than they are shorter-timed averages (e.g. hourly) [1]. They
are also not accurate at estimating specific spatial or temporal occurrences, which
can pose issues for pollutants that range drastically with time and location [9]. As a
result, it is a common practice to evaluate the monitored and modelled results based
on unpaired basis. The treatment of volume sources in AERMOD is another area of
conservatism, as they do not include upward velocity [11].
Dispersion models are sensitive to the inputted meteorological data, with variability in the meteorological dataset carrying through to the dispersion model
results. These arise due to assumptions that simplify the model but are not always
well-representative of actual atmospheric conditions during each modelling hour.
One example of this that occurs in both the meteorological pre-processors AERMET and CALMET (preprocessors of regulatory dispersion models AERMOD and
CALPUFF, respectively) is the Bowen Ratio, which is estimated by month or season
and land-use type. Assuming consistent monthly or seasonal values does not represent actual conditions. Similar assumptions occur for parameters such as Albedo and
surface roughness.
The objective of this study is to assess the model performance of both AERMOD
and CALPUFF by comparing modelled concentrations to measured concentrations
at two Facilities within Ontario. The bias from each of these dispersion models were
analyzed to see whether these models produce inconsistent results.
57.2 Methodology
AERMOD v14134 and CALPUFF v.6.263 level 080827, the two Ontario approved
regulatory models at the time of the experiment, were used to model 24 h average
Nickel (Ni). For AERMOD, the modelling was performed as per the Ontario Ministry
of the Environment, Conservation and Parks’ (Ministry) “Guideline A-11: Air Dispersion Modelling Guideline for Ontario,” [6]. The Ministry’s “Technical Bulletin:
Using Combined Assessment of Modelled and Monitored (CAMM) Results to
Refine Emission-Rate Estimates” [7], was followed to conduct the assessment. The
CALPUFF modelling was performed as per the British Columbia Ministry of Environment’s “Air Quality Dispersion Modelling Guideline” [2].
Emissions rates and source characterization for these Facilities were obtained
from confidential Facility specific data. The emission rate sources include: direct
measurements, estimates from emission factors (such as those from the U.S. EPA’s
AP-42), mass balances, and engineering calculations. The Facilities’ emission data
J. Mak et al.
AERMOD will not underestimate maximum concentrations but will sometimes overpredict as high as by a factor of 5 [5, 10]. The occurrence of maximum concentrations
is often at times of low-wind speeds (when dispersion is poor). During these stagnation conditions, model errors have been over-emphasised utilizing both AERMOD
[8] and CALPUFF [3].
Dispersion models are more adequate at predicting longer-timed averaged concentrations (e.g. annual) than they are shorter-timed averages (e.g. hourly) [1]. They
are also not accurate at estimating specific spatial or temporal occurrences, which
can pose issues for pollutants that range drastically with time and location [9]. As a
result, it is a common practice to evaluate the monitored and modelled results based
on unpaired basis. The treatment of volume sources in AERMOD is another area of
conservatism, as they do not include upward velocity [11].
Dispersion models are sensitive to the inputted meteorological data, with variability in the meteorological dataset carrying through to the dispersion model
results. These arise due to assumptions that simplify the model but are not always
well-representative of actual atmospheric conditions during each modelling hour.
One example of this that occurs in both the meteorological pre-processors AERMET and CALMET (preprocessors of regulatory dispersion models AERMOD and
CALPUFF, respectively) is the Bowen Ratio, which is estimated by month or season
and land-use type. Assuming consistent monthly or seasonal values does not represent actual conditions. Similar assumptions occur for parameters such as Albedo and
surface roughness.
The objective of this study is to assess the model performance of both AERMOD
and CALPUFF by comparing modelled concentrations to measured concentrations
at two Facilities within Ontario. The bias from each of these dispersion models were
analyzed to see whether these models produce inconsistent results.
57.2 Methodology
AERMOD v14134 and CALPUFF v.6.263 level 080827, the two Ontario approved
regulatory models at the time of the experiment, were used to model 24 h average
Nickel (Ni). For AERMOD, the modelling was performed as per the Ontario Ministry
of the Environment, Conservation and Parks’ (Ministry) “Guideline A-11: Air Dispersion Modelling Guideline for Ontario,” [6]. The Ministry’s “Technical Bulletin:
Using Combined Assessment of Modelled and Monitored (CAMM) Results to
Refine Emission-Rate Estimates” [7], was followed to conduct the assessment. The
CALPUFF modelling was performed as per the British Columbia Ministry of Environment’s “Air Quality Dispersion Modelling Guideline” [2].
Emissions rates and source characterization for these Facilities were obtained
from confidential Facility specific data. The emission rate sources include: direct
measurements, estimates from emission factors (such as those from the U.S. EPA’s
AP-42), mass balances, and engineering calculations. The Facilities’ emission data