next up previous
Next: 4 Application and Experimental Up: 3 Validation Previous: 3.3 Simulation of a

3.4 Chemical Box calculation

 


  
Figure 7: The concentrations of the reference case were calculated using the Gong & Cho solver and the Livermore-ODE-Solver (lsode), using an integration step of 60 sec. The upper panel shows the absolute ozone concentrations. The curves resulting from the Gong & Cho solver and from the lsode solver can not be distinguished, because they are almost identical. The plot also shows the results of case 19 a published by Stockwell et al. The lower panel shows the relative deviation in percents ( 100 $\cdot$ ${\frac{c_{\rm gongcho} - c_{\rm lsode}}{c_{\rm lsode}}}$ ) for Ozone and NO 2 between Gong & Cho and lsode. The maximum deviation found for NO 2 was 0.12 %.
\begin{figure}
 \begin{center}
 \leavevmode
 \epsfxsize=\textwidth
 \epsffile{chemcomp.eps} \end{center} \end{figure}

To check the accuracy of the chemical solver several test calculations were carried out. The case presented here refers to case 19 a in Stockwell et al. (1997), which closely resembles the ``PLUME/2'' case in Kuhn et al. (1998). The box model results of our implementation of the Gong & Cho-solver were compared to results obtained using the Livermore ODE-Solver (Hindmarsh; 1983) which is technically based on the Gear-solver (Gear; 1971) and widely accepted as a reference. Furthermore, we compared the results with those obtained by Stockwell et al. (1997). Figure 7 shows an excellent agreement between the results of the Gong & Cho-Solver, and the Livermore solver. Only slight differences to the data of Stockwell et al. (1997) were found. They are mainly due to the different procedures of calculating photodissociation constants.


next up previous
Next: 4 Application and Experimental Up: 3 Validation Previous: 3.3 Simulation of a
Silvan Perego
1/21/1999