Geoscience Reference
In-Depth Information
Figure 6.5. Forecasted water levels and total uncertainty (standard deviation) in
the forecasted water levels due to uncertain parameters: (a) during
rising flow, and (b) during subsiding flow situations.
(Fig. 6.6). Triangular PDFs of the forecasted water levels are assumed to compute the
confidence bounds. It can be clearly seen that the observed water levels are well within the
95% confidence limits in the case of the subsiding flow, whereas they are mostly outside
the limits in the rising flow case. By limiting the analysis to the five uncertain parameters
considered in this exercise and assuming that the estimate of uncertainty in the propagation time
is reasonably correct, the uncertainty assumed in the rating curve conversion is not large
enough to keep the observed water levels within 95% confidence limits in the rising flow
example. This conclusion is in agreement with the assumption that a large uncertainty in
the rating curve conversion can be expected for the higher water levels. This is due to the
fact that as the data for the extreme events (very high or very low flow) are naturally sparser, the
rating curves in these regions are produced either by extrapolation or interpolation with
very limited data. Consequently, this leads to a larger uncertainty in the estimated flows.
Search WWH ::




Custom Search