Geology Reference
In-Depth Information
other problem issues identified by Nachtergaele
et al
. (2001a,b), although not specifically related
to EGEM but typical of many erosion models, is
the faulty calculation of channel erodibility
(Torri
et al
. 1987) and critical flow shear stress.
The latter is calculated according to Smerdon and
Beasly (1961):
are recurring phenomena and their position can
be derived from earlier observations. If this infor-
mation is not available, for example for a model-
ling study of an unknown area, a separate
analysis using local slope and contributing area
thresholds could identify potential initiation
points in a more objective manner. This analy-
sis then has to be done outside the REGEM
framework.
A first important change in the hydrological
module is that it has been adapted to simulate
unsteady flow. REGEM still remains based on
TR-55 methods, using the empirical curve
number method to estimate runoff processes
(USDA Soil Conservation Service, 1986), but now
the calculated runoff volume and peak runoff rate
are used to construct a triangular hydrograph for
each event. Since the model now considers
unsteady and spatially varied flow, these are able
to influence the headcut migration rate, channel
width and depth, and the rates of sediment
entrainment, transport and deposition.
Two sources of sediment are considered in
REGEM: headcut erosion and channel widening.
The gully incision processes start with a localized
incision, which is considered of little relevance
since its eroded volume is not taken into account.
Once the headcut is formed, this is one of the main
processes of sediment production. The processes of
plunge-pool erosion and headcut migration are rep-
resented by the fully analytical model of Alonso
et al
. (2002), discussed further on in this chapter.
These headcut migration processes dynamic-
ally interact with the hydrological model compo-
nent. After every time step, during which the
gully head can migrate, the contributing area and
resulting runoff discharge are modified accord-
ingly. As the gully grows, the actual drainage area
is reduced, so that the discharge at the gully head
diminishes accordingly. Since the real shape of
the catchment in which the modelled gully is
located has been approximated by one single
amorphous cell in AnnAGNPS, some approxima-
tions need to be introduced. The drainage area
contributing to the runoff at each timestep
j
is
expressed as a function of the actual gully length,
L
j
, and the maximum gully length possible,
L
max
:
0.0128
Pc
t
=
0.311*10
(19.3)
c
where
Pc
is clay content (%) of the soil.
Channel erodibility (
KC
) is then assumed to
be inversely linearly related to
t
c
. However, an
extensive literature review has recently shown
the limitations of this simple approach (Knapen
et al
., 2007). Furthermore, the flow's transport
capacity is calculated according to Yalin's method.
This implies that sediment deposition cannot be
simulated and that particle-size characteristics
are simplified to some representative grain diam-
eter, so sediment sorting cannot be represented.
With the objective of overcoming some of the
limitations of EGEM, the physical basis and pre-
dictive capabilities of EGEM were further elabo-
rated. This resulted in an improved version of
EGEM, called the Revised Ephemeral Gully
Erosion Model (REGEM) (Gordon
et al
., 2007).
Four major changes can be emphasized: (1) dynamic
gully length by explicitly modelling headcut
migration processes; (2) spatially and temporally
varied runoff discharge during events; (3) improved
gully width estimates, based on discharge alone; (4)
five different particle sizes accounted for in the
sediment transport calculations.
The computational framework has been
designed in order to incorporate the model in the
Annualized Agricultural Nonpoint Source
Pollution Model (AnnAGNPS). AnnAGNPS uses
amorphous areas or subwatersheds, which
implies that some of the sub-pixel gully erosion
processes need to be simplified. One disadvan-
tage of this approach is that local topography is
simplified to some average subwatershed charac-
teristics. Therefore, gully initiation points need
to be defined by the user, hereby introducing
some degree of subjectivity. However, usually
this is not problematic, since ephemeral gullies
