APPLICATION: FLOOD FORECASTING MODEL FOR KLODZKO CATCHMENT (POLAND) - Modelling Uncertainty in Flood Forecasting Systems - page 99

in Figure 5.9. The figure also shows the basin average precipitation on the negative
y -axis. It is to be noticed that the simulation was carried out taking the accumulated
precipitation in each subbasin, and not with the basin average precipitation.
Secondly, the methodology is applied to estimate the uncertainty in the forecasted
discharges due to the uncertainty in the forecasted precipitation. The results are
characterised as “reconstructed” and “uniform”. Whereas the “reconstructed” (i.e. with
disaggregation) considers uncertainty due to the unknown temporal distribution of the
precipitation, the “uniform” (i.e. without disaggregation) implies that average or
uniformly distributed (throughout the subperiods) precipitation is used. The results
produced by normal and micro GAs are very similar. Compared results for some cases are
presented in Subsection 5.4.3. Since the results from both GAs are very close the results
presented in Subsections 5.4.1 and 5.4.2 are all from the normal GA. The com-
parison of the results from the two versions of the GA is presented in Subsection \5.4.3.
5.4.1 Results with 3 subperiods
Figure 5.10(a) shows the upper and lower bounds of the forecasted discharges using
reconstructed precipitation with 3 subperiods. Two cases with ! =0 and ! =1 are
presented. In this graph, the lower bounds of the two cases are almost overlapping,
whereas some deviations can be seen in the upper bounds.
Figure 5.11 is presented to further illustrate the differences in these two cases. It plots
the uncertainty bounds in the forecasted discharges, i.e. Q UB ! Q LB , for ! =0 and ! =
1 against the forecast hours. The upper and lower bounds of the forecasted discharges
with uniform precipitation are also presented in Fig. 5.10(b). For ! =0 the upper and
lower bounds show very little differences. For ! =1 with the uniform precipitation, there
exist no upper and lower bounds as there is only one value of precipitation, i.e. P i, mc (see
Fig.5.5).
Figure 5.12(a) presents uncertainty bounds, Q UB - Q LB , for ! =0 with reconstructed and
uniform precipitations. The uncertainty bound for ! =1 with reconstructed precipitation is
presented in Figure 5.12(b). As there are no upper and lower bounds, there is no
uncertainty bound with uniform precipitation for ! =1. It is clearly seen that in all fore-
casts the output uncertainty is dominated by the case with a reconstructed
precipitation. This suggests that the uncertainty due to the unknown temporal distribution
can be more significant than the uncertainty in the quantity of the precipitation.