Geology Reference
In-Depth Information
satisfactory performance in these studies was
the inability to model the evolution of a signifi-
cant gully on the mining landform. This was
due to an inability to resolve a mine road, and
consequent impacts on overland flow concen-
tration, in the digital elevation model, rather
than any inherent deficiency in the LEM or the
database of parameters.
Using the Bowen Basin database, Martinez
et al
. (2009) showed that both EAMS-SIBERIA
and RUSLE gave good matches to the observed
erosion rate (since 1950) estimated by using cae-
sium-137 for a natural catchment. The catchment
where the comparison was done was a grazing
paddock of 7 hectares and slopes ranging from
5-30%. Both models allowed for the vegetation
cover by the use of USLE cover factors. In this
application there was little change in the land-
form, so the comparison is more a test of the
validity of the Bowen Basin database than of
the LEM.
computing load significantly. Some of the LEMs
are quite efficient, but they will always use more
computer time than a traditional model. Small
areas (a few square kms at a resolution of tens of
metres, or a million computational nodes in
space) can be easily run on a desktop computer
with modest run times. Long simulations for
large areas can still require significant compute
times (CPU days).
The CPU time issue leads to one of the most
common criticisms of LEMs by erosion modellers
from a traditional background. This criticism is
that the physics of LEMs is simpler than in their
current model. This is generally in reference to
the fluvial erosion model. Much of this criticism
is, in our view, misguided. As noted above, the
fundamental physics in the best LEMS and the
most recent process-based traditional models are
very similar. Many details such as spatial distri-
bution of soil cover, practice and soil properties
are commonly ignored by LEMs, but this is a
practical issue reflecting their research heritage
rather than a fundamental flaw. For instance, the
EAMS erosion assessment package built around
the first author's SIBERIA and TelluSim LEMs
allows the input of these management factors,
although at the current time it is more an excep-
tion to the rule in that regard (see Hancock
et al
.
(2008b) for an example of using spatially variable
soil erodibility in a LEM).
A limitation of the current fluvial erosion
models in many LEMs that is only now being
addressed (Coulthard
et al
., 2000) is that they
generally do not track the full soil grading of the
eroded material. In the original science applica-
tions this was not seen as a critical failing, and
ignoring the grading simplifies the computations
considerably. Several LEMs now include sedi-
ment grading as part of their calculations. Some
motivations for current efforts to include eroded
grading are: (1) to model multilayer caps and the
long-term behaviour of constructed rock armours
on waste containment structures; (2) to simulate
better any depositional behaviour; and (3) to allow
the modelling of water quality parameters such
as sorbed pollutants and radionuclides. This area
is evolving rapidly.
18.5
Landform Evolution Model Limitations
The discussion above highlights the new insights
that can be made using LEMs. However, they are
not without their limitations at the current time.
At the science level, an evolving landform imposes
limitations, which may or may not be important
for any particular application. Many things that
can be assumed to be constant in a traditional
model will evolve in concert with the landform.
Soils change as the soil armours in the bottoms
of gullies. This gully and armour development
changes the spatial distribution of the soil erodibil-
ity in a way that is intimately linked with the evo-
lution of the landform. Similarly, the soil moisture
distribution may also change with the evolving
landform. This will change the distribution of veg-
etation density (particularly in arid regions), and
therefore the cover factor across the landform.
At a practical level, an LEM must solve the
erosion equations many millions of times as the
landform evolves over time, whereas a tradi-
tional model only has to solve the erosion equa-
tions once. For an LEM this increases the
