Surface wash on a low-angled slope near Bloemfontein


Le Roux, J.S.; Roos, Z.N.

South African Geographical Journal 64(2): 114-124

1982


Surface wash was measured for three consecutive years on a fairly low-angled slope by means of 6 pairs of wash-traps. The erosion rate was found to be more or less constant over the whole slope. This implies that, for the period under observation, the slope retreated parallel to itself. Multiple regression analysis of erosion (dependent variable), rainfall, rainfall intensity and the product of rainfall and rainfall intensity indicated that the only truly significant variable was the product of rainfall and rainfall intensity. The three variables together accounted for 58 percent of the variance in erosion. The rate of soil loss calculated from the wash-trap data for 37 runoff events was 46 g m-2 per year or, calculated as ground lowering, 18 mm 1000 yr-1 at rock density, or 34 mm 1000 yr-1 at soil density.

SURFACE
WASH
ON
A
LOW
-ANGLED
SLOPE
NEAR
BLOEMFONTEIN
J.
S.
LE
ROUX
and
Z.
N.
ROOS
University
of
the
Orange
Free
State,
Bloemfontein
(Manuscript
received
18th
June,
1982;
in
revised
form
16th
August,
1982)
ABSTRACT
Surface
wash
was
measured
for
three
consecutive
years
on
a
fairly
low
-angled
slope
by
means
of
6
pairs
of
wash
-traps.
The
erosion
rate
was
found
to
be
more
or
less
constant
over
the
whole
slope.
This
implies
that,
for
the
period
under
observation,
the
slope
retreated
parallel
to
itself.
Multiple
regression
analysis
of
erosion
(dependent
variable),
rainfall,
rain-
fall
intensity
and
the
product
of
rainfall
and
rainfall
intensity
indicated
that
the
only
truly
significant
variable
was
the
product
of
rainfall
and
rain-
fall
intensity.
The
three
variables
together
accounted
for
58
percent
of
the
variance
in
erosion.
The
rate
of
soil
loss
calculated
from
the
wash
-trap
data
for
37
runoff
events
was
46
g
m-
2
per
year
or,
calculated
as
ground
lowering,
18
mm
1000
yr-
1
at
rock
density,
or
34
mm
1000
yr-
1
at
soil
density.
Introduction
This
study
is
an
attempt
to
analyze
aspects
and
rates
of
surface
wash
on
a
natural
slope
assumed
to
be
representative
of
slope
con-
figuration
in
the
southern
Orange
Free
State
(Fig.
1).
Surface
wash
refers
to
the
downslope
transport
of
weathered
material
(regolith)
by
sheetwash
(a
thin
film
of
moving
water)
and
rillwash
(water
flowing
in
shallow,
ill-defined
micro
-channels).
The
effect
of
raindrop
impact
is
considered
to
play
an
important
part
in
the
total
rate
of
transport
of
surface
wash,
as
it
substantially
aids
wash
transport
by
detaching
regolith
particles,
and,
because
of
the
slope,
has
a
net
downslope
com-
ponent.
Wash
-traps
were
used
to
determine
the
rate
of
surface
wash.
The
wash
-trap
used
in
this
study
(Fig.
2)
is
a
modification
of
the
trap
used
by
Young
(1972),
the
modification
being
the
attachment
of
a
piece
of
rubber
hose
to
the
bottom
of
the
trough,
in
order
to
collect
both
water
and
sediment
in
one
or
two
251
plastic
containers.
Water
and
sediment
in
the
containers
were
removed
weekly
for
laboratory
measurement
of
water
volume
and
mass
of
the
trapped
sediment.
The
wash
volume
and
chemical
load
were
not
taken
into
account
for
this
report.
A
continously
recording
rain
gauge
was
installed
to
measure
rainfall
intensity,
and
an
ordinary
rain
gauge
was
also
installed
as
a
control
to
measure
the
amount
of
rainfall.
Rainfall
intensity
was
calculated
only
for
rainfall
events
which
resulted
in
actual
runoff.
During
the
period
observed
no
rainfall
event
of
less
than
7
mm
produced
runoff.
Study
Area
The
slope
studied
is
situated
approximately
10
km
north
of
Redders-
burg.
Bedrock
is
mostly
shale
of
the
Beaufort
Group,
with
some
dolerite
at
the
top
of
the
hillock.
The
area
is
well
covered
by
perennial
grasses
and
isolated
shrublets
which
provide
good
natural
grazing.
The
paddock
enclosing
the
study
area
is
well
managed
and
no
signs
of
accelerated
erosion
were
noted.
South
African
Geographical
Journal,
Vol.
64
No.
2,
1982
Surface
Wash
on
a
Low
-Angled
Slope
115
near
Bloemfontein
1
26
°
10
°
Petrusburg
BLOEMFONTEIN
Tierpoort
Dam
...
K
11i
P
e
I'
Thaba
Nch
29
°
34
Rif
,p.
v
.e
-STUDY
AREA
REDDERSBUR
FourieSpruil
Dam
0
5
10km
1
Fig.
1:
The
location
of
the
slope
studied.
The
average
annual
rainfall
for
the
area
is
484
mm,
of
which
approximately
eighty
percent
falls
during
the
summer
months
from
October
to
April.
Daily
maximum
temperatures
vary
between
30°
and
33
°
C
in
January
and
between
17°
and
21°
C
in
July.
The
average
minimum
temperature
for
July
is
about
C,
but
temperatures
may
drop
to
as
low
as
minus
8°C.
Frost
occurs
fairly
regularly
during
the
nights
of
June,
July
and
August.
116
The
South
African
Geographical
Journal
00mm
____
300mm
WASH
-TRAP
To
Container
Fig.
2:
The
wash
-trap
used
in
this
study.
A
length
of
hosepipe
leads
the
collected
wash
into
one
or
two
251
containers.
Field
and
Laboratory
Techniques
Installation
of
Wash
-traps
A
total
of
twelve
wash
-traps
was
installed
in
such
a
way
as
to
exclude
or
at
least
minimize
the
possible
influence
of
the
traps
on
each
other
(Fig.
3
and
4).
The
troughs
were
carefully
installed
in
such
a
way
as
to
mimimize
soil
surface
disturbance
on
the
upslope
side.
The
upslope
side
of
the
trough
has
a
lip
of
approximately
30
mm
which
was
very
carefully
pushed
into
the
soil
15
mm
below
the
surface.
Water
and
sedi-
ment
can
therefore
flow
freely
into
the
trough.
As
some
disturbance
of
the
soil
above
the
lip
was
unavoidable,
the
surface
was
slightly
stabilized
by
spraying
with
an
aerosol
paint.
The
wash
collected
from
initial
runoff
events
was
discarded
to
allow
for
settled
conditions
to
be
attained.
Delimitation
of
Trap
Drainage
Area
It
is
difficult
to
assess
the
catchment
area
of
each
trap
accurately.
The
only
way
to
reach
a
completely
accurate
estimate
of
the
absolute
size
of
the
area
would
have
been
to
enclose
the
area
above
the
traps,
but
this
would
have
introduced
insurmountable
difficulties
on
a
long
slope,
as
well
as
other
undesirable
side
effects,
such
as
a
disturbed
surface.
The
catchment
of
each
'trough
was
determined
by
careful
surveying
and
contour
analysis.
Orthogonals
were
drawn
from
the
traps
to
the
watershed
of
each
trap
and
the
area
determined.
The
wash
-traps
were
installed
in
November
1977,
while
the
observation
period
was
started
in
March
1978
and
ended
in
April
1981.
During
this
period
the
annual
rainfall
figures
differed
significantly
(Table
1).
Surface
Wash
on
a
Low
-Angled
Slope
117
near
Bloemfontein
Fig.
3:
S.
3
4
I
?
t
,0
7
.2
i
io
5,0
4,0
3,0
X
7,0
11
12
•••
........"'"....'
—1,o
44„
1g
9
10
0
30m
r
---
1
Stream
.-
1
Contours(m)
r
---
1
Wash
trap
Ell
—",
h-
Doterite
Limit
of
doterite
debris
Small
scarp
Position
of
the
wash
-traps
on
the
slope.
D.i.,,,,.
,
1.2
(Poor
I)
ri
Wosh-Irops
4.
0
2.4
(Pror
2)
5.4
(Pool)
No
(Poo•o)
40
2.10
(Poo
5)
1102
(pc
..
s)
00
120
11O
-
Fig.
4:
Profile
of
the
slope
with
the
positions
of
the
traps
indicated.
118
The
South
African
Geographical
Journal
TABLE
1
:
Rainfall
Figures
for
the
Observation
Period.
Year
Rainfall
(mm)
Runoff
events
'79-'80
400,5
9
'78-'79
443.2
13
'80-'81
537,1
15
A
pair
of
traps
instead
of
a
single
trap
was
used
in
order
to
reduce
the
possibility
of
large
deviations
in
the
estimated
area
of
the
catchment.
The
slope
and
catchment
area
of
the
wash
-traps
are
as
follows
(Table
2):
TABLE
2:
Mean
Slope
and
Catchment
Area
of
Wash
-traps
Pair
of
wash
-traps
1
2
3
4
5
6
Mean
slope
(degrees)
9,67
8,03
5,57
3,94
4,04
2,64
Mean
catchment
area
(m
2
)
3,71
12,58
22,76
30,11
39,90
46,51
Laboratory
Technique
The
plastic
cans
containing
sediment
and
water
were
taken
to
the
laboratory
once
a
week
and
the
sediment
filtered
off.
The
dry
weight
of
the
sediment
was
determined
after
oven
drying
at
105°C
for
24
hours.
Soil
Loss
Data
and
Statistical
Analysis
Soil
Loss
Figures
Tested
for
Significance
The
more
important
descriptive
statistics
are
presented
in
Table
3.
TABLE
3:
Mean
Soil
Loss
per
Runoff
Event
Pair
of
wash
-traps
1
2
3
4
5
6
Mean
soil
loss
(g
m-
2
)
4,34
3,38
4,55
2.55
3,58
3,90
Standard
deviation
5,76
5,53
5,64
2,90
4,10
5,05
Standard
error
of
the
mean
0,95
0,91
0,93
0,48
0,67
0,83
The
soil
loss
figures
were
tested
for
Snedecor's
F
and
Student's
t.
Only
for
pair
4
do
the
variances
differ
significantly.
The
F
test
for
the
group
of
6
pairs
as
a
whole
produced
an
F
value
of
0,77
(df
=
216,
and
df
=
5
for
rainfall
events
and
the
six
pairs
respectively).
Student's
t
may
therefore
be
applied
with
some
confidence
and
in
fact
indicated
no
significant
difference
among
the
mean
soil
loss
figures
for
the
six
pairs
of
wash
-traps
for
the
37
runoff
events.
Snedecor's
F
and
Student's
t
tests
were
also
applied
to
the
body
of
soil
loss
data
to
determine
whether
the
soil
loss
figures
differ
significantly
between
the
different
years
(Table
4).
The
F
and
t
tests
indicate
that
the
variances
and
the
averages
of
the
erosion
figures
differ
significantly
between
the
first
and
second
years,
the
second
and
third
years,
and
between
the
first
and
third
years.
TABLE
4:
Surface
Wash
on
a
Low
-Angled
Slope
119
near
Bloemfontein
Snedecor's
F
and
Student's
t
Statistics
of
Soil
Loss
Data
for
all
Runoff
Events
1978/'79-1979/'80
1978/'79-1980/'81
1979/T0-1980/
.
81
F
3.92
(p
=
0,01%)
8,99
(p
=
0,01%)
2,29
(p
=
0,01%)
2,35
(p
=
2,0%)
5.52
(p
=
0,01%)
3,31
(p
=
0,1%)
df
(F)
77
&
53
89
&
53
89
&
77
df
(t)
130
142
166
df
=
degrees
of
feedom
p
=
probability
Correlation
and
Regression
Analysis
In
order
to
determine
whether
erosion
on
the
slope
investigated
is
significantly
influenced
by
rainfall
and
rainfall
intensity,
the
data
were
analysed
by
means
of
simple,
partial,
semipartial
and
multiple
correlation
and
by
multiple
regression
analysis.
This
analysis
was
based
on
34
run-
off
events,
since
three
intensity
figures
were
not
recorded
due
to
a
faulty
recording
rain
gauge.
The
data
for
the
different
variables
are
presented
in
Table
5.
TABLE
5:
Data
used
for
Correlation
and
Regression
Analysis
Runoff
event
E
(Y)
R
(X.,)
I
(X
9
)
RI
(X,)
1
0,36
20,5
4,0
82,0
2
0,45
30.0
4.6
138,0
3
1,06
59,5
4,1
244.0
4
0,36
59,1
3,0
177,3
5
0,63
7,4
30,0
222,0
6
4,26
73,0
2,3
167,9
7
3,14
43,3
7,7
336,5
8
3,71
41,5
5,0
207,5
9
1,13
50,7
3,8
192,7
10
5,79
44,0
8,1
356.4
11
0,30
34.8
3.8
132.2
12
0,24
42,0
2,9
121,8
13
0,39
7,0
9,0
63,0
14
1,63
76,4
8,4
641,8
15
4,28
13,5
13,5
182,3
16
0,11
31,5
3,2
100,8
17
2,15
11,0
6;
6.8,2
18
2,23
20
10,7
299,6
19
1,46
31,0
5,1
158,1
20
12,98
21,0
17,0
357,0
21
1,60
47,9
2,0
95,8
22
0,1,4
24,0
2,0
48,0
23
0,82
10
i
32.0
336.0
24
3,66
30,0
4,2
126,0
25
11,33
24,0
12,0
288.0
26
4,99
22,2
12,0
266,0
27
4,63
16,1
10,5
169,1
28
16,99
136,0
9,7
1319,2
29
18,34
34,6
21,7
750,8
30
7,88
33,8
8,4
283,9
31
3,40
37,3
14,0
522,2
32
5,64
65,0
6,7
435,5
33
3,55
27,6
17,0
469,2
34
0,96
33,3
3,8
126,5
Mean
(X)
3,84
37,29
9,07
278,98
St.
dev.
(6)
4,65
_
24,90
7,38
246,95
E
=
Y
=
sot
loss
in
g
m-
2
;
R
=
X
1
=
rainfall
in
mm;
I
=
X,
=
rainfall
intensity
in
mm
h
-I;
R
I
=
X
q
=
product
of
rainfall
and
intensity
in
mm
2
h-1
120
The
South
African
Geographical
Journal
The
simple
correlation
analysis
for
soil
loss
against
rainfall
and
rainfall
intensity
shows
poor
correlations
(r
=
0,33
and
r
=
0,32
re-
spectively)
compared
to
the
correlation
between
soil
loss
and
the
product
of
rainfall
and
rainfall
intensity
(r
=
0,72).
As
can
be
expected
the
three
tndependent
variables
showed
rather
serious
collinearity
(Table
6).
TABLE
6:
Simple
Correlation
Matrix
for
Sail
Loss,
Rainfall,
Rainfall
Intensity,
and
the
Product
of
Rainfall
and
Rainfall
Intensity.
Y
X,
X,
X
R
Y
1,00
X,
0,33
1,00
X,
0,32
—0,33
1,00
X,
0,72
0,66
0,35
1,00
Y
=
Soil
Loss
X,
=
Rainfall
X,
=
Rainfall
Intensity
X„
=Product
of
Rainfall
and
Rainfall
Intensity
The
more
important
multiple
partial
coefficients
of
determination
were
found
to
be
/
4
,
2
,
2
,
=
0,05
and
r
2
,3,12
=
0,39.
The
former
indicates
that
the
explanatory
power
of
rainfall
intensity
is
only
5
percent
when
the
effects
of
rainfall
and
the
product
of
rainfall
and
rainfall
intensity
are
kept
constant
but
that
it
increases
the
percentage
explained
to
39
percent
if
it
is
weighted
by
rainfall.
Only
three
semi
-partial
coefficients
of
determination
proved
to
be
of
value.
Rainfall
(r
2
,
(
1,2,
)
=
0,02)
explains
2
percent
of
the
total
variance
in
erosion
which
is
not
explained
by
intensity
and
the
product
of
rainfall
and
rainfall
intensity.
Similarly
rainfall
intensity
(ri11(2,1S)
0,015)
ex-
plains
only
1,5
percent
of
the
variance,
and
the
product
of
rainfall
and
rainfall
intensity
(r
2
,
(
,,,,
)
=0,25)
25
percent
of
the
variance
when
em-
ployed
last
in
the
regression
equation.
These
semi
-partial
coefficients
of
determination
show
the
very
important
effect
of
the
third
independent
variable
(product
of
rainfall
and
rainfall
intensity)
in
the
explanation
of
soil
loss.
The
multiple
correlation
analysis
of
the
data
using
rainfall,
rainfall
intensity
and
the
product
of
rainfall
and
rainfall
intensity
as
independent
variables
produced
a
value
(R)
of
0,76.
The
explanatory
power
of
the
independent
variables
if
employed
cumulatively
produced
the
following
values
:
r
2
,,1
=
11
percent,
R
2
y,
12
31
percent
and
R
2
,,
12
,
=
58
percent
(where
y
=
soil
loss
in
g
m
-2
;
1
=
rainfall
in
mm;
2
=
rainfall
intensity
in
mm
h
-1
and
3
=
product
of
1
and
2).
The
Multiple
Regression
Equation
The
relationship
between
soil
loss
(Y
in
g
m
-2
)
as
a
dependent
variable
and
rainfall
(X
i
in
mm),
rainfall
intensity
(X,
in
mm
h
-1
)
and
rainfall
multiplied
by
rainfall
intensity
(X,
in
mm
2
h
-
')
is
given
by
Y
=
3,00
0,10
X
i
0,17
X,
+
0,02
X,
±
3,16
(R,„
12
,
=
0,76)
t
=
2,03
1,29
4,38
p
=
0,05
0,21
0,00
e
=
0.05
0,13
0,01
where
t
is
Student's
t,
p
probability
and
e
the
standard
error
of
the
co-
efficient
concerned.
Surface
Wash
on
a
Low
-Angled
Slope
121
near
Bloemfontein
The
coefficients
of
the
independent
variables
differ
markedly
in
degree
of
significance.
The
Student's
t
test
indicates
that
the
negative
correlation
between
rainfall
and
soil
loss,
as
found
in
this
study,
is
significant
at
a
5,1
percent
level
of
confidence
but
that
the
negative
correlation
with
rainfall
intensity
is
uncertain
at
a
20,7
percent
level
of
confidence.
The
positive
correlation
between
erosion
and
the
product
of
rainfall
intensity
is
highly
significant.
The
F
test
for
significance
has
also
been
carried
out
for
R
2
y,123
(0,58)
and
was
found
to
be
significant
at
the
0,0001
percent
level
of
confidence,
and
the
correlation
with
erosion
in
this
area
of
all
independent
variables
taken
together
probably
holds
good
for
sign
and
may
hold
good
for
magnitude
as
well.
The
Rate
of
Erosion
Despite
the
fact
that
the
study
has
been
conducted
for
only
three
consecutive
years,
the
high
significance
of
the
multiple
correlation
co-
efficient
indicates
that
the
rate
of
erosion
found
in
this
study
approxi-
mates
the
true
rate
of
erosion.
Fortunately
the
three
years
have
covered
an
average
rainfall
more
or
less
similar
to
the
long-term
average.
The
average
for
this
period
is
460,3
mm,
while
the
long-term
average
is
484
mm
per
annum.
The
soil
loss
figures
and
standard
errors
of
the
mean
have
been
calculated
for
37
runoff
events.
Mean
soil
loss
(X)
=
3,72
g
m
-2
per
runoff
event
Standard
error
of
X
=
0,80
g
m
-2
per
runoff
event
Net
soil
loss
per
annum
=
3.72
X
37
g
m
2
=
45,88
g
3
Standard
error
of
the
mean
=0,80
X
37
=
9,91
g
m
-2
3
The
rate
of
erosion
if
calculated
as
surface
lowering
in
terms
of
rock
removal,
taking
2
500
kg
m
-3
as
the
average
density
of
rock
(Le
Roux
and
Roos,
1981)
in
mm
1000
yr"
is
:
Erosion
rate
at
rock
density
=18,4
±
4,0
mm
(68%
probabality)
1000
yr
1
.
And
in
terms
of
soil
density
(determined
from
samples
along
the
slope
as
1
340
kg
m
-2
);
Erosion
rate
at
soil
density
±
34
mm
1
000
yr
-1
.
Discussion
The
factors
controlling
soil
erosion
fall
into
four
main
groups
(Kirkby,
1975):
(i)
depth,
permeability
and
other
properties
of
the
soil;
(ii)
gradient
of
the
land;
(iii)
Frequency
distribution
of
rainstorms
able
to
produce
overland
flow;
and
(iv)
land
use,
especially
the
amount
of
vegetation
cover
during
the
rainstorm
periods
of
the
year.
122
The
South
African
Geographical
Journal
The
fact
that
Snedecor's
F
and
Student's
t
indicated
no
significant
differences
in
soil
loss
for
the
six
pairs
of
wash
-traps
is
important.
On
this
rather
short
slope
the
first
group
of
controlling
factors
are
fairly
constant
for
all
except
the
first
pair
of
wash
-traps,
while
group
(iii)
and
(iv)
are,
of
course,
the
same
for
all
the
wash
-traps.
The
decrease
in
gradient
down
slope
did
not
lead
to
a
decrease
in
soil
loss.
This
implies
that
some
other
factor
must
balance
the
effect
of
gradient
and
the
most
obvious
factor
in
this
case
must
be
slope
length.
Factors
such
as
soil
erodibility
and
ground
cover
by
vegetation
could
hardly
be
expected
to
effect
such
a
uniform
influence
as
shown
by
the
statistics.
A
preliminary
survey
of
vegetal
ground
cover
percentages
shows
figures
of
30,6,
25,8,
26,9, 27,5,
26,land
25,7
for
pairs
1,
2,
3,
4,
5
and
6
respectively.
Energy
lost
by
retardation
of
the
surface
wash
on
the
lower
gradient
is
com-
pensated
for
by
an
increase
in
volume
of
overland
flow.
If
the
data
for
the
three
years
prove
to
be
true
for
the
long
term,
it
would
imply
that
the
rate
of
soil
loss
on
this
slope
is
more
or
less
the
same
from
the
crest
to
the
base
and
that
the
slope
is
retreating
parallel
to
itself.
Further-
more
this
also
suggests,
since
only
a
small
part
of
the
crest
of
the
hillock
shows
some
dolerite
outcrop,
that
the
rate
of
erosion
equals
the
rate
of
weathering.
The
slope
is
therefore
for
the
most
part
a
transportation
slope
subject
to
control
by
erosion.
The
slightly
concave
profile
of
the
slope
(Fig.
4)
may
be
due
to
three
factors
:
(a)
the
restricting
effect
of
the
fixed
slope
base;
(b)
the
increase
in
volume
and
the
potential
increase
in
the
erosive
capacity
of
the
surface
wash.
This
may
be
an
absolute
increase
in
erosivity,
or
(c)
the
diminishing
size
of
the
particles
of
the
regolith
downslope.
In
this
study
no
absolute
downslope
increase
in
the
erosive
power
was
found.
Transporting
power
increased,
as
shown
by
the
fact
that
the
erosion
rate
per
unit
area
remained
more
or
less
constant.
This
would
either
mean
that
the
concavity
has
been
inherited
from
a
former
cycle,
or
that
the
increase
in
erosivity
at
the
lower
end
could
not
be
detected,
as
a
result
of
possible
small
experimental
errors.
The
restricting
effect
of
the
slope
base
is
necessary,
since
active
undercutting
would
lead
to
a
convexity.
In
the
case
of
this
slope
a
small
scarp
has
actually
been
moving
away
from
the
streamlet
and
forming
a
slight
convexity.
This
may
have
been
initiated
by
a
particularly
heavy
storm
flushing
out
large
quantities
of
sediment
in
1972
(Le
Roux
and
Roos,
1979).
The
three
consecutive
years
of
this
investigation
have
coincided
with
an
increase
in
annual
rainfall
as
well
as
an
increasing
product
of
rainfall
and
intensity,
averaging
196,0,
211,5
and
432,5
mm
2
h
-1
respectively.
The
rate
of
erosion
varied
sympathetically
and
positively
with
the
increase
in
rainfall
and
the
product
of
rainfall
and
rainfall
intensity.
The
erosion
figures
of
1,68,
2,95
and
5,60
g
m
-2
for
the
three
years
indicate
that
increasing
rainfall
and
accompanying
factors
may
be
the
cause
of
the
increase
in
the
intensity
of
erosion.
The
multiple
coefficient
of
determination
(R'1,123
0,58)
indicates
that
there
are
probably
more
independent
variables
that
contribute
to
the
rate
of
erosion.
Possible
additional
factors
may
be
grazing,
which
removes
vegetal
cover
and
could
increase
erosion
by
raindrop
impact,
Surface
Wash
on
a
Low
-Angled
Slope
123
near
Woemfontein
disaggregation
of
topsoil
particles
from
trampling
by
grazing
animals,
crusting
as
a
result
of
raindrop
impact,
disintergration
of
soil
peds
as
a
result
of
freeze
-thaw
action,
the
influence
of
burrowing
insects,
reptiles
and
small
mammals,
and
experimental
errors
as
well
as
other
possible
variables.
In
the
partial
correlation
analysis
it
has
been
shown
that
rainfall
intensity
explains
little
of
the
erosion.
In
many
multiple
correlation
analyses
such
a
variable
may
be
ignored,
but
in
the
case
of
this
study
it
would
clearly
be
inadvisable,
as
this
factor
is
in
any
case
necessary
in
order
to
calculate
the
third
variable
(X
3
).
The
multiple
regression
equation
provides
a
model,
albeit
tentative
at
present,
for
the
prediction
of
erosion
when
rainfall
and
rainfall
intensity
figures
are
available.
This
model
needs
to
be
improved
by
additional
experiments
on
other
slopes
and
in
other
areas.
The
rather
low
significance
of
the
coefficients
of
rainfall
and
rainfall
intensity
needs
to
be
studied
further.
The
rate
of
erosion
of
±
18
mm
1
000
yr
-
'
is
in
the
same
order
of
magnitude
as
the
results
obtained
from
the
rate
of
erosion
found
by
calculations
from
the
amount
of
sediment
accumulated
in
a
small
farm
impoundment
(29
mm
1
000
yr':
catchment
area
±
5
km
2
)
and
in
the
Fouriespruit
Dam
(23
mm
1
000
yr
-1
.
catchment
area
±
700
km
2
)
(Le
Roux
and
Roos,
1979,
1981).
The
slope
studied
is
situated
in
the
catchment
of
both
these
dams.
It
should
be
noted
that
the
chemical
load
has
not
been
taken
into
account
in
any
of
these
studies.
Since
only
a
minor
percentage
of
the
area
con-
cerned
is
covered
by
primary
rocks,
and
most
of
the
ions
removed
in
solution
have
probably
been
stored
in
the
interstices
of
the
sediment
particles,
the
figure
for
ground
lowering
is
not
expected
to
be
affected
much
by
the
inclusion
of
solution
loss.
This
aspect
however
needs
further
investigation.
The
rate
of
weathering
on
this
slope
is
not
known.
In
fact,
very
little
is
known
about
the
rate
of
weathering
in
any
part
of
South
Africa.
Since
the
slope
investigated
in
this
study
is
only
used
for
grazing
and
is
situated
in
a
paddock
which
is
fairly
well
managed.
it
is
assumed
that
the
rate
of
erosion
equals
the
rate
of
feathering.
This
means
that
weathering
is
probably
occuring
at
a
rate
of
18
mm
for
0.4
tons
of
rock
ha
-1
yr
-1
)
per
1
000
years.
An
erosion
figure
exceeding
this
amount
in
these
regions
possibly
indicates
bad
farming
practices.
It
is
interesting
to
note
that
Stocking
(1978)
has
found
that
natural
soil
renewal
is
probably
less
than
1
ton
ha
-
]
yr
-1
.
A
ton
ha
-
'
yr
-1
is
equal
to
40
mm
1
000
yr'
at
rock
density
(2
500
kg
m
-2
).
Some
aspects
needing
further
research
are
the
influences
of
changing
ground
cover
due
to
grazing
and
drought,
soil
erodibility,
chemical
erosion
(solution
loss)
and.
possibly,
the
affect
of
soil
depth
and
infiltra-
tion
capacity
on
the
rate
of
erosion.
Acknowledgements
The
authors
wish
to
acknowledge
funding
from
the
Department
of
Agriculture
and
Fisheries.
The
authors
also
wish
to
thank
Professor
Tony
Young
whose
"Slopes"
and
subsequent
personal
communication
originally
inspired
this
research.
124
The
South
African
Geographical
Journal
REFERENCES
Kirkby,
M.
J.,
1975:
Erosion
by
water
on
hillslopes.
In
Chorley,
R.
J.,
(ed),
Introduction
to
fluvial
processes.
Methuen,
London,
98-107.
Le
Roux,
J.
S.
and
Roos,
Z.
N.,
1979:
Rate
of
erosion
in
the
catchment
of
the
Bulbergfontein
dam
near
Reddersburg
in
the
Orange
Free
State,
Journal
of
the
Limnological
Society
of
South
Africa.
5(2),
89-93.
Le
Roux,
J.
S.
and
Roos,
Z.
N.,
1982:
The
rate
of
soil
erosion
in
the
Wuras
dam
catchment
calculated
from
sediment
trapped
in
the
dam.
Annals
of
geomorphology,
26(3),
315-329.
Stocking,
M.,
1978:
A
dilemma
for
soil
conservation.
Area,
10,
(4),
306-308.
Young,
A.,
1972:
Slopes,
Longman
Inc.,
New
York.