Geology Reference
In-Depth Information
mean annual rainfall, but also that an increase in
soil erosion can be expected under lower storm
frequency, especially in humid catchments, when
considering lower rainfall rates. The relationship
between climate and soil loss appears to be con-
trolled by soil texture characteristics in shape and
magnitude. This is one of the few studies focus-
ing specifically on the impacts of drought fre-
quency and length on soil erosion processes
within the context of climate change.
Finally, Lane
et al
. (2007) used a more detailed
modelling approach to study the feedback
between impacts of climate change on flood fre-
quency and sedimentation in Britain. The model-
ling framework included a detailed inundation
model coupled with an estimate of channel aggra-
dation from suspended sediment deposition; the
model was applied using the A2 climate change
scenario for the 2050s and 2080s using the
HADRM3 RCM. Results point to an increase in
inundated area, due to rainfall changes alone, of
12.2% to 14.7% during relatively frequent floods
(1-in-0.5 and 1-in-2 year events). When consider-
ing also the impact of sedimentation in the river
bed, the inundated area increases by 38.2% to
52.1%. The results indicate that in-channel sedi-
mentation increases the sensitivity of flood inun-
dation to climate change, and measures to prevent
streambank erosion might aggravate this problem
as the river would require enlargement to com-
pensate for the rising channel bed. This study
highlights one possible off-site effect of increased
soil erosion rates, a subject which has not received
much attention in recent research.
Nevertheless, there are a number of examples
in the literature. One case study is the Soil
Erosion Network's model intercomparison exer-
cise (Nearing
et al
., 2005), aimed at investigating
the response of different soil erosion models, with
different methods and levels of process represen-
tation, to key variables expected to be impacted
by climate change: precipitation and vegetation.
Seven different models were applied to two water-
sheds, one humid (in Belgium) and one semi-arid
(in Arizona, US); they utilize different approaches
to erosion process description, and temporal and
spatial discretization, and include continuous as
well as event-based models. This modelling
approach was used to study the response of three
storms per catchment to hypothetical changes in
storm rainfall, vegetation cover and ground cover,
from −20% to 20%.
The model response to these changes was
coherent, with all models responding more
strongly to changes in rainfall. The median ratio
of sensitivity of sediment yield to rainfall changes
was around 8 in the humid catchment, and around
5 in the semi-arid catchment, with models
responding more strongly to changes in rainfall
amount and intensity than to changes in rainfall
amount alone. For vegetation and ground cover,
the sensitivity was around −2 in both catchments.
The coefficients of variation between models are
significant, but most models responded within a
similar range, especially for stronger storms;
these results indicate that the tested models give,
in relative terms, similar responses to climate
forcings, which increases the credibility of the
different modelling approaches.
Another extreme event study was made by
Nunes (2007), who applied the MEFIDIS model
(Nunes
et al
., 2005) to one humid and one semi-
arid catchment in Portugal. MEFIDIS is a model
optimized for extreme events, with a high proc-
ess discretization in space and time. The author
performed a first approach using hypothetical
scenarios of rainfall and vegetation cover change,
similar to the one presented above. The results
for sediment yield sensitivity were similar, but
the author also analysed differences in erosion
response with spatial scale. Catchment-scale
15.4.5
Modelling extreme events
The previous examples focused on long-term
continuous modelling; there are fewer case stud-
ies specifically focusing on individual extreme
events. This can be attributed in part to the com-
plexity of the processes involved, particularly
when compared with the low spatial and tempo-
ral resolution of current climate prediction
approaches. Furthermore, studies at the extreme
event scale are dependent on longer-term predic-
tions for vegetation cover and land use.
