Geology Reference
In-Depth Information
Table 12.9
Average observed rill erosion rates compared with LISEM simulations.
Erosion rate
(tonnes km
−2
)
Class boundaries
(tonnes km
−2
)
Correctly predicted by LISEM
(% of observed class area)
Erosion class
No rill erosion
251
0-800
51
Slight rill erosion
1450
800-3000
25
Moderate rill erosion
4822
3000-8000
21
Severe rill erosion
17,195
Over 8000
14
on the sediment plot. Nevertheless, the sediment
concentration as determined from the barrels at
the bottom of the plot was in excess of 700 g l
−1
,
resulting in a sheet erosion rate of 836 t km
−2
(8.4 t
ha
−1
). The sediment delivery ratio of the fields
was assumed to be 0.3. This gives erosion rates
(on event basis) of 251 t km
−2
for croplands that do
not show evidence of rill erosion.
Table 12.9 shows the average erosion rates
obtained from the erosion plot data as well as the
rill measurements on the fields. The erosion rate
for the 'no rill erosion' class is thus a single event
estimate based on sediment plot data, while the
other rates are based on rill mapping. Since only
one event produced rills in 1999, all these rills
must have formed on 20 July. The resulting 1999
rill erosion map is shown in Fig. 12.6a.
LISEM produces maps of erosion and deposi-
tion rates in t ha
−1
. Since the field mapping only
involved erosion and not deposition, it would
appear logical to use the LISEM erosion map only.
Erosion and deposition, however, cannot be treated
as separate entities in LISEM simulations; deposi-
tion and re-entrainment can occur during the
simulation. Thus, the same sediment can be
eroded several times and be deposited several
times. Therefore, the net erosion map should be
used to assess the performance of the LISEM
model in a spatial way. The result should be judged
more on patterns than on amounts, because dur-
ing mapping deposition was ignored. The net ero-
sion map was also based on the classification given
in Table 12.9 and is shown in Fig. 12.6b. Since the
rill erosion map (Fig. 12.6a) only shows erosion on
fields, only the cropland areas were used for clas-
sification of the LISEM net erosion map.
Comparing the maps in Fig. 12.6, it is obvious
that both maps have the highest erosion rates in
the southeastern part of the catchment. This is
also in agreement with the observed distribution
of rainfall on 990720 (Section 12.6.1). The lack of
heavy rain in the western part of the catchment is
probably the reason why the LISEM prediction for
the 'no rill erosion' class is much better than for
the other classes (Table 12.9). Closer inspection of
Fig. 12.6 shows that although the overall pattern
was similar, the pattern in detail was very differ-
ent. This is also reflected by the data given in Table
12.9, which show that for the classes with more
severe rill erosion, only about 20% of the observed
area for that particular class is predicted correctly.
(v) Land use scenarios
Figure 12.7 shows a clas-
sified version of the present land use erosion map
produced by LISEM. Without classification the
map appearance would be totally dominated by a
few very high values. For a large area in the south-
ern part of the catchment, no serious erosion was
predicted. This was caused by the fact that, accord-
ing to the measured rainfall data, less rain fell in
this area during the 980801 event. Comparison of
Figs 12.7 and 12.1 indicates that other areas with
negligible erosion rates for the 980801 storm were
mainly under woodland. A zone along both sides
of the main valleys also had little erosion since
this area is underlain by bedrock, which has a
much higher cohesion, both in reality and in the
model. The hilltop areas generally had slight or
