388
o. Kriiger et al.
using six measurements, three at 2000 m and three at 3000 m, of 0.99 shows
quite good agreement between model and measurements. Increasing the number
of measurements to a density of 15, it can be seen that the accuracy of the
individually determined source strengths does not significantly increase. The use
of 15 measurements shows no significant further improvement in the results. A
much stronger variability in the results can be seen by changing the location of
the four measurements inside the plume. From all model experiments it can be
summarized that the inverse modelling procedure reaches a minimum difference
between modelled and theoretical source strengths of less than 5% by the using
only four measurements. Therefore, relative to the source configuration at the
BSL Werk Schkopau, four measurements of gaseous mercury at a distance of
1500 m appeared to be satisfactory. With further increasing distances, the
accuracy of the results for individual emission sources decreases. An important
result for all test runs was that the total source strength at BSL Werk Schkopau
was almost reproduced by this procedure with uncertainties less than 1%.
Following these findings, a second series of test runs was performed in order to
reproduce the results using four measurements at a distance of 1500 m for the
other meteorological conditions indicated in Table 2. Now all cases were
reproduced by the inverse modelling procedure within the same low uncertainties as shown in Table 2. The differences between modelled and theoretical
concentrations in nearly all runs were lower than 1 ng/m3. The results above
confirmed the good performance of the inverse modelling technique and suggest
an optimal distance of 1000 m for obtaining in situ measurements.
6
Emissions of Gaseous Mercury
For applying the inverse modelling procedure described above, data from several
measurement campaigns in the period from December 1993 to April 1994 were
used to estimate the range of source strengths i"n a first guess, which were used as
start values for the minimizing procedure. These estimates are based on vertical
wind- and temperature fields interpolated from neighbouring meteorological
stations. The measurements which were used to determine source strengths were
detected with peak values of 235 ng/m3 within a distance of 1000 m from the
sources. The total source strength of the BSL Werk Schkopau including the three
closed chlor-alkali factories, the one chlor-alkali factory still in operation and the
partly removed acetaldehyde production plant was roughly estimated to be lower
than 10 kg/day. In this estimation of the source strength an error of a factor of 2
was assumed.
In June 1994 a 3-day measurement campaign was undertaken in order to
estimate the source strengths at the BSL Werk Schkopau with higher accuracy.
The meteorological data of this dry episode were simultaneously measured by
radio-sonde ascents and theodolite observations to obtain more reliable data at
the BSL Werk Schkopau for the dispersion modelling.
o. Kriiger et al.
using six measurements, three at 2000 m and three at 3000 m, of 0.99 shows
quite good agreement between model and measurements. Increasing the number
of measurements to a density of 15, it can be seen that the accuracy of the
individually determined source strengths does not significantly increase. The use
of 15 measurements shows no significant further improvement in the results. A
much stronger variability in the results can be seen by changing the location of
the four measurements inside the plume. From all model experiments it can be
summarized that the inverse modelling procedure reaches a minimum difference
between modelled and theoretical source strengths of less than 5% by the using
only four measurements. Therefore, relative to the source configuration at the
BSL Werk Schkopau, four measurements of gaseous mercury at a distance of
1500 m appeared to be satisfactory. With further increasing distances, the
accuracy of the results for individual emission sources decreases. An important
result for all test runs was that the total source strength at BSL Werk Schkopau
was almost reproduced by this procedure with uncertainties less than 1%.
Following these findings, a second series of test runs was performed in order to
reproduce the results using four measurements at a distance of 1500 m for the
other meteorological conditions indicated in Table 2. Now all cases were
reproduced by the inverse modelling procedure within the same low uncertainties as shown in Table 2. The differences between modelled and theoretical
concentrations in nearly all runs were lower than 1 ng/m3. The results above
confirmed the good performance of the inverse modelling technique and suggest
an optimal distance of 1000 m for obtaining in situ measurements.
6
Emissions of Gaseous Mercury
For applying the inverse modelling procedure described above, data from several
measurement campaigns in the period from December 1993 to April 1994 were
used to estimate the range of source strengths i"n a first guess, which were used as
start values for the minimizing procedure. These estimates are based on vertical
wind- and temperature fields interpolated from neighbouring meteorological
stations. The measurements which were used to determine source strengths were
detected with peak values of 235 ng/m3 within a distance of 1000 m from the
sources. The total source strength of the BSL Werk Schkopau including the three
closed chlor-alkali factories, the one chlor-alkali factory still in operation and the
partly removed acetaldehyde production plant was roughly estimated to be lower
than 10 kg/day. In this estimation of the source strength an error of a factor of 2
was assumed.
In June 1994 a 3-day measurement campaign was undertaken in order to
estimate the source strengths at the BSL Werk Schkopau with higher accuracy.
The meteorological data of this dry episode were simultaneously measured by
radio-sonde ascents and theodolite observations to obtain more reliable data at
the BSL Werk Schkopau for the dispersion modelling.