Environmental Engineering Reference
In-Depth Information
4.4 Evaluation of the Air Quality Forecasting System
Using Satellite and In-situ Data
Marje Prank, Milla Lotjonen, Joana Soares, and Mikhail Sofiev
Finnish Meteorological Institute, Erik Palmenin Aukio 1, Helsinki, Finland
Abstract
The paper presents the evaluation of the regional air quality forecasting
system based on chemical transport model SILAM, using three pollution cases
selected in the CO ST Action 728 - a heavy PM episode in Central Europe during
January-April 2003, mainly fire-induced multi-pollutant episode in April-May
2006, and another fire- and anthropogenic- emission induced episode in August
2006. Modelling results for these cases were compared with both in-situ and
satellite measurements. Comparison with in-situ data shows the model well
reproducing or over-estimating PM concentrations while under-estimating most of
gases. Comparison with AOD data from MODIS shows that the model correctly
reproduces the overall spatial patterns being in most cases within a factor of 1.5
from the total spatially-averaged AOD. For total modelled AOD at 550 nm, the
largest contributions came from sulphates, nitrates and primary PM2.5, while
coarse particles and sea salt contributions were negligible.
Keywords
Model evaluation, model inter-comparison
1. Introduction and Setup of the Evaluation Exercise
The current paper presents model evaluation exercises performed within the scope
of the COST Action 728 “Enhancing mesoscale meteorological modelling capabilities
for air pollution and dispersion applications”. A specific exercise performed within
the Action was the multi-model evaluation and inter-comparison using three
selected challenging pollution cases: a set of heavy PM episodes in Central
Europe, February-March 2003; multi-pollutant episode in April-May 2006, largely
created by wild-land fires in western Russia, and fire- and anthropogenic- emission
induced episode in August 2006.
There were up to seven models performing the specific episode analysis. This
multi-model ensemble was evaluated against in-situ and remote-sensing observations
as shown in
Fig. 1.
Additionally, the individual models were compared with each
other and the ensemble average was computed with the Joint Research Centre
Ispra ENSEMBLE software.
