Geology Reference
In-Depth Information
Table 12.7
Nash-Sutcliffe coefficients for LISEM 163 and LISEM LP.
980801
980823
990720
000811
000829
Average
LISEM 163
0.74
−0.28
0.64
−0.50
0.94
0.31
LISEM LP
0.86
0.24
0.91
0.33
0.85
0.64
Table 12.8
Simulated peak discharge (l s
-1
) using the calibrated datasets
for the different events.
Calibration dataset
Event
980801
980823
990720
000811
000829
Measured
980801
4967
12,004
3617
4346
3873
5125
980823
35
591
13
13
24
701
990720
4649
10,536
3330
4151
3301
3589
000811
612
4573
83
189
95
214
000829
9608
21,329
14,331
15,471
8886
8757
between the rise of the measured and simulated
hydrographs decreased somewhat.
For the 990720 storm the 'calibration' channel
length was much shorter than for most other
events. This can be explained by the fact that this
storm only produced high intensity rain in the
areas close to the catchment outlet. The most
striking difference between the simulation with
LISEM 163 and LISEM LP is that LISEM LP no
longer simulated a double peak for this event.
The double peak was probably caused by water
from different parts of the catchment arriving at
the weir slightly out of phase. The hydrograph
predicted with LISEM LP suggests that the first
peak has been retarded by the adaptations to
LISEM, so that both peaks are now in phase.
Calibration of the 000829 event did not show
an early peak (Fig. 12.5). The rising limb of the
hydrograph was reproduced almost perfectly, espe-
cially by LISEM 163, but the falling limb of the
hydrograph went down a little too slowly. Three
possible causes for the early peaks have been men-
tioned earlier. Because of the very abrupt nature of
the 000829 storm, where very high-intensity rain
suddenly occurred, it is possible that the aforemen-
tioned effects of infiltration in macropores (such as
fissures and pipes) and the channel bed did not play
a large role in this case. The effect of overland flow
distance should still occur. In contrast to the other
events, the result for LISEM 163 was slightly bet-
ter than for LISEM LP, as was also indicated by the
Nash-Sutcliffe coefficient (Table 12.7).
(iii) Results for peak discharge validation
To vali-
date LISEM LP, the calibrated datasets for each
storm were applied to the other four events. Using
the calibration dataset of one event for the other
four events usually gave worse fits than those pre-
sented in Fig. 12.5. The results of this validation are
shown in Plate 10 for the 990720 event. Table 12.8
gives a summary of the results for peak discharge
for all events. Plate 10 shows that applying calibra-
tion settings from another event almost always
gave results that differed much from those obtained
by calibrating on that particular event. Table 12.8
shows that this was the case for all events. It is
therefore concluded that a separate calibration is
necessary for events of different magnitudes, and
probably even for each event separately.
(iv) Spatial patterns
A minor event that occurred
at the sediment plot on 21 July 1999 was used to
derive a sheet erosion rate on an event basis. This
minor storm did not result in any rill formation
