Page 332 - Materials Chemistry, Second Edition

L1644_C07.fm Page 300 Monday, October 20, 2003 12:10 PM

more differentiated with respect to meteorological conditions and population density
than the EcoSense results for each grid cell. Therefore, compliance between BEEST
and EcoSense is not achievable by saying that the population exposure subtracted
by EcoSense must be equal to the I near results.
In multimedia models for ERA, additional sources of uncertainty must be con-
sidered — in particular, site characteristics and transfer factors for transport among
media. The site characteristics include human life-styles because different diets and
media composition show important variations from one site to another.
The evaluation of the dispersion conditions of pollutants in air has been carried
out using meteorological data of a limited number of years (site-specific assessment)
and measurement stations (site-dependent assessment). The meteorological condi-
tions are strongly determined locally and therefore a reduction of data always
increases the uncertainties. Also, a smaller number of years and stations makes the
results less representative for the site or Catalonia as a whole. However, the derivation
of the meteorological data files in this study has been carried out on this basis
because no more data were available. Of course, the statistical evidence would
increase if more data were available; however, for pragmatic reasons, the limited
number of data was accepted.
If one considers the derivation of the statistical meteorological data files by
Harthan (2001) mentioned in Section 7.4.1, it must be said that the formation of
classes of wind speed (0 to 2 m/s, 2 to 3 m/s and 3 to 4 m/s) has not been undertaken
according to statistical reasoning. The limits of these classes are chosen according
to the limits defined by the Deutscher Wetterdienst (DWD) and constitute a good
compromise between the concept of classes and the meaningfulness of the classes.
If broader class limits were chosen, the handiness of the results would increase
because the overall number of classes would be reduced. However, this would lead
to a large variation of actually occurring wind speeds within each class, i.e., the
statistical determination of the classes — describing the wind speed of all locations
lying in this class with a reasonable standard deviation — would no longer be well
founded. If narrower class limits were chosen, this would lead to a smaller standard
deviation within each class and would therefore decrease the uncertainties. Never-
theless, the number of classes would increase and the number of districts lying in
each class would decrease and a statistical reasoning combining several districts in
one class would no longer be possible. The class limits chosen here seem to be
appropriate because they allow a minimum differentiation of wind speed (into three
classes), but still with a reasonable number of districts per class.
Neglect of terrain elevations and precipitation is necessary in the site-dependent
impact assessment due to the absence of statistical reasoning for this parameter;
nevertheless, this leads to uncertainties. In particular, the concentration increment
of particles calculated is overestimated because wet deposition is not considered.
An evaluation of different temperatures has led to the conclusion that the results for
the concentration increment are not sensitive to temperature. Therefore, it is valid
to choose the country average for all data sets and throughout the whole year. The
neglect of wind direction leads to uncertainties, especially on a local level, because
wind direction strongly influences dispersion on that level. However, in order to
form class averages accounting for several wind directions, the neglect of wind

© 2004 CRC Press LLC