Geoscience Reference
In-Depth Information
This is equivalent to assuming that the cumulative volume of retention
F
(
t
) can be predicted
as
F
(
t
)
S
max
=
Q
P
−
I
a
(B6.3.5)
F
(
t
) is often interpreted as a cumulative volume of infiltration, but it is not necessary to assume
that the predicted stormflow is all overland flow, since it may not have been in the origi-
nal small catchment data on which the method is based (application of the method to one
of the permeable, forested, Coweeta catchments (Hjelmfelt
et al.
, 1982) is a good example
of this).
A further assumption is usually made in the SCS-CNmethod that
I
a
=
S
max
with
commonly
assumed to be
0
.
2. Thus, with this assumption, the volume of storm runoff may be predicted
from a general form of the SCS-CN equation:
≈
(
P
−
S
max
)
2
P
+
Q
=
(B6.3.6)
(1
−
)
S
max
B6.3.2 The Curve Number
The SCS-CN method can be applied by specifying a single parameter called the curve number
(CN) and the popularity of the method arises from the tabulation of CN values by the USDA for
a wide variety of soil types and conditions (see Table B6.3.1). In the standard SCS-CN method,
the value of
S
max
(in mm) for a given soil is related to the curve number as
1000
CN
−
10
S
max
2
.
54
=
(B6.3.7)
where the constant 2.54mm converts from the original units of inches to millimetres. Ponce
(1989) and others suggest that this is better written as
100
CN
−
1
S
max
C
=
(B6.3.8)
where
S
max
now varies between 0 and
C
. The original equation implies a value of
C
of 10 inches
(254mm). It is easily seen from Equation (B6.3.8) that the range 0 to
C
is covered by curve
numbers in the range from 50 to 100. Although values less than 50 are sometimes quoted in
the literature, this implies that they are not physically meaningful for the original
C
of 254mm.
Curve number range tables are provided for the estimation of the curve number for different
circumstances of soil, vegetation and antecedent conditions (e.g. Table B6.3.1). For areas of
complex land use, it was normally suggested in the past that a linear combination of the
component curve numbers, weighted by the area to which they apply, should be used to
determine an effective curve number for the area (e.g. USDA-SCS, 1985). However, because of
the nonlinearity of the runoff prediction equation, this will not give the same results as applying
the equation to each individual area and weighting the resulting runoff by the components'
fractional areas. Grove
et al.
(1998) show how using a distributed curve number calculation
increases the volume of predicted runoff, by as much as 100% in their examples, relative to
using the composite method. Now that computer limitations are much less restrictive, the latter
procedure should be adopted and, in fact, arises naturally when the SCS-CN method is applied
to an HRU classification of a catchment based on GIS overlays.
