Environmental Engineering Reference
In-Depth Information
When only initial conditions of the forecast are improved, the additional
information from the assimilation is lost rapidly. Therefore, the noise settings from
the analysis were used to adapt the emissions, boundary condition and deposition
velocities also in the forecast. The noise settings from the time of the ozone
maximum of the previous day appeared to give the best results for the 1 day
forecast. For a longer forecast time, the free-running model was slightly better.
This possibly relates to the question of how long the adapted emission settings are
expected to be valid. Furthermore, since the free-running model is not able to
reproduce the extremes, it is possible that the model is biased for high temperatures,
which makes data assimilation effective in the sense that it improves the initial
conditions but not in the sense that it compensates for inherent model uncertainties.
Therefore, a temperature-dependent bias correction is being tested.
Further model improvements include the assimilation of NO2 column observations
from OMI and the combination of the ground- and satellite-based data sets.
5. Conclusions
CTMs can be used successfully in forecasts during episodes of high concentrations of
Particulate Matter and ozone. They are superior to statistical models when meteoro-
logical conditions change, especially at the beginning and end of the episode. Data
assimilation of ground-based monitoring network data can improve the forecasts
at least for 1 day ahead. Work is under way to include ground-based PM obser-
vations and satellite observations of PM and NO
2
in the assimilation process.
References
Schaap et al.: The LOTOS-EUROS model: description, validation and latest developments. Int.
J. Environment and Pollution
32
no 2. (2008)
Manders et al.: Testing the capability of the chemistry transport model LOTOS-EUROS to
forecast PM10 levels in the Netherlands. Atmospheric Environment,
43
(2009)
