Professional Documents
Culture Documents
4, NO.
WATER
RESOURCES
RESEARCH
OCTOBER
1968
A Least-Squares
HydrographAnalysiso[ ComplexStorms
on SmallAgriculturalWatersheds
D.
E.
OVERTON
There
INTRODUCTION
ad-
vances objectively by operating on basic hydrologicdata in an attempt to define the precise nature of the variability of hydrologic
response.
956
). E. OVERTON
served storms by a finite convolutionof rainfall excessr, with a T-hour unit hydrograph
TUH. The matrix technique incorporates a
least-squares
bestfit procedure.
The procedurereported by Snyder therefore
is amenable to analysis of all storm hydrographs and provides for a 'best-fit' TUH that
may be used to solve for the instantaneous
unit hydrograph IUH, such as by an S-curve
technique [Nash, 1958]. If underlying hydrologic assumptionsare acceptable,the degreeof
fit of the convolutedTUH will provide much
information on the nature of the system. If
the fit is good,the system could be said to be
approximately linear. Recently, Newton and
lation.
cess in continuing application of the leastWith the ordinates of the unit hydrograph
squaresmethod to TVA watersheds.
U(T, t -- ) and the increments of rainfall
Another method of linear systems analysis excessr(t), the calculatedstorm dischargeQ(t)
was reported by O'Donnell [1960]. He utilized at the end of the first time interval T is
an orthogonal function (sine-cosine Fourier
series)
for fittingobserved
rainexcess-time
and discharge
hydrograph-time
data.The IUH
((T)= r(T)U(1)
(2)
andat t = 2T
the seriescoefficients
of the IUH weresolved
(3)
explicitly
in termsof theseries
coefficients
of andfort = 3T
the rain excessand dischargehydrographs.
Muchhydrologic
information
is needed
on ((3T)= r(T)U(3)
velopment
of hydrologic
design.
In thisreport,
q- r(2T)U(2) + r(3T)U(1)
(4)
cultural
watersheds
gaged
bytheAgricultural
andforthenthdischarge
Research Service [U.S.
1965]
was
analyzed
toevaluate
thefeasibility
)(nT)
= r(T)U(nT)
q- ... q-r(nT)U(1)
(5)
of detecting
hydrologic
response
by theleast- Generalizing,
atanytimeiT willbe
sheds
have
characteristically
steep
rises;
thereforeit is morerepresentative
to combine
the
(iT)= [r(jT)U(i
-- j + 1)] (6)
i
ideasof Snyder[1955]andO'Donnell
[1960], To insureadequaterepresentation
of the
representing
theunithydrograph
asa Fourier steeprisinghydrographs
of smallagricultural
series,
andto solvefortheseries
coefficients
by watersheds
of the Agricultural
Research
Serthe matrixinversion
least-squares
technique.vice,the unit hydrograph
wasrepresented
by
PROBLEM
FORMULATION
U(T, t)
theunithydrograph
wasintitially
considered= aoq- [akcos(kwt)q-bksin(kwt)]
(7)
to be linear,time-invariant
within a storm.
A group of storms was analyzed for each
watershed,and a best-fit unit hydrographwas
deleted. It
was allowed to
Complex Storms
hydrographshape.Then
9.57
U(T, t) =
[bksin(kwt)]
- /
(.
B--- LL
is the
time
(18)
wherein
where
(8)
k=l
T(N-
1)
base of the
(10)
observed
O/Ob
[Q(t)- )(t)]2 = 0
(19)
(11)
(12)
(3)
+ b7sin (7wT)]
form
O/Ob
= o
(20)
or
O(i') = (i)
+ (i')
+ ...
q- b7B7(iT)
(15)
where
(i)
r Bl()2
BI(t)B7(' 51
= () h [- + )]
( )
(t)(t)
...
()
B(t)Q(t)
If there are n positive ordinateson the observed hydrographs, then the observational
( t)Q(t)
matrix becomes
and in matrix notational
_B(nT) ...
form
(22)
ib
B7(nT)J
(21)
__
__
Q(nT):
()
Ibl---IBB[ - IBm[
(23)
958
D.E.
OVERTON
FROM SMALLAGRICULTURAL
WATERSI-IEDS tions at Danville, Vermont; Coshocton,Ohio;
Storm
records
published
bytheAgricultural
Hastings,
Nebraska;
Riesel,
Texas;andSafResearch
Service
[U.S.Dept.Agr.,1963,
1965,]ford,Arizona.
Thewatersheds
range
in size
wereselected
to investigate
thefeasibility
of from122to 579acres
andrepresent
a cross
deriving
T-hourunithydrographs.
Seventeen
section
ofthenational
datapopulation
ofsmall
watershedspublished by the Agricultural Research Service.
TABLE
1. Storms
Analyzed
byLeast-Squares
Fifteen
of thestorms
werecomplex.
Two
Unit Hydrograph Techniques
storms were observed 5-minute unit hydro-
4/25/61
9/23/45
6/12/57
1/21/59
7/28/61
0.40
0.94
2.64
0.09
3.52
8/07/61
W-1
8/03/59
519 acres
7/19/57
Danville, Vt. 11/28/59
1.70
*
*
in.
0.10
0.83
0.71
0.46
0.64
0.27
0.65
0.17
0.07
0.41
1.55
3.12
1.35
0.93
1.00
0.23
0.20
1.62
0.84
0.25
0.37
0.44
0.94
1.24
0.98
0.43
W-2
146 acres
Hastings,
Nebr. a
W-3
481 acres
5/15/60
8/11/61
4/22/57
5/02/57
6/15/57
Riesel, Texase 4/24/57
W-C
5/13/57
579 acres
7/16/57
' Coshocton ....
b Safford .......
Danville ......
10-minute UH
5-minute UH
1-hour
UH
a Hastings .....
10-minute
e Riesel ........
10-minute
storm
UH
Southwest
United
States
* Observed5-minuteunit hydrographs.
Location
Coshocton
W-10
Safford
W-1
Danville
W-2
Hastings
W-3
Riesel
W-C
Storm
Efficiency
4/25/61
.68
.23
9/23/45
6/12/57
1/21/59
7/28/61
8/07/61
.87
.75
.11
.68
.82
1.09
.36
.23
.74
--.59
8/03/59
7/19/57
ba
b4
b5
b6
.26
.35
.28
.42
1.16
.21
.11
.34
.94
-.10
.02
--.05
--.10
--.92
.71
--.06
--.12
--.16
.66
-4.23
.03
.21
.00
--.35
3.25
.10
.03
.05
.13
-1.52
.06
--.02
--.03
--2.8
.02
.00
.00
.04
.11
23
-- 07
-- 08
-- 18
-- 01
-- 04
--.03
.15
.11
--.05
.04
--.23
.03
.01
.00
-.78
b7
.37
(5-minuteunit hydrograph)
(5-minuteunit hydrograph)
11/28/59
.52
.06
.06
.04
5/15/60
8/11/61
4/22/57
5/02/57
6/15/57
4/24/57
5/13/57
7/16/57
.74
.77
.76
.82
.65
.78
.75
.76
.65
1.32
.38
.50
.40
.44
.30
.17
.47
.95
.36
.37
.40
.20
.19
.10
.04
.26
.19
.02
.22
02
05
03
--.22
.06
.02
--.17
.05
--.01
--.04
--.02
--.15
.18
--.05
--.17
--.10
--.01
--.09
--.03
Complex Storms
959
RESULTS
i.o
-o 5
--I .o
40
80
120
160
2OO
24O
280
TIME (MINUTES)
W-10.
OF
ANALYSIS
E- (So- Se)/So
cared.
The results from the Safford watershedare
much better, as shown in Figure 2. Although
(24) theTUHshape
varies
considerably,
all cal-
2.5
SAFFORD, ARIZONA
5-MINUTE
UNIT
W- I
519 ACRES
HYDROGRAPHS
-19 -57
2.0
.5
-59
8-7-61
20
40
I
60
80
100
120
140
TIME (MINUTES)
160
960
D.E.
OVERTON
DANVILLE,
I HOUR
VT.
UNIT
W-2
HYDROGRAPH
--
TIME
(HOURS)
I0
II
12
20I I I I I
IO-MINUTE
UNIT
HYDROGRAPH Examination
,.
of
the
efficiencies indicates
-0.5
iO
80
1210
TIME
(MINUTES)
'60 200 240 the TUH, as notedin Figure1. An example
Fig.4. Calculated
unithydrographs
forHastingsof degree
of fit is shown
in Figure6. Thisis
W-3.
Complex Storms
0
961
J
R1ESEL TEXAS
IO-MINUTE
07'-
06
"
W-C
UNIT
HYDROGRAPHS
'-24
-57
0.4
0.3
02
/ //
O.I
0
20
/' '
40
60
.....""o,
'X"'-X X ''.............
.... 80
I00
120
140
IGO
180
200
220
TIME(MINUTES)
240
260
280
300
320
340
360
April25t 1961
Efficiency
:0 68
08-
xObserved
--
0.4-
Calculated(fitted)
?/
/'/
",
xx
I
962
D.r.
I
OWRTON
SUMMARY
./-MAY
27,1948
X=475in/hr
3
APRIL
17,
1941
X=
I95
in/hr
I/1
OCT.
2,
194,
X=
15Z
,n/hr
0
20
40
60
80
I00
120
140
rain
AND
CONCLUSIONS
The feasibility of deriving linear, time-invariant unit hydrographs from small agricultural watersheds from complex storms was
shown using the least-squares techniques of
Snyder [19'55]. The TUH was expressedas a
seven-term sine series to assure good representation of the steep rises characteristic of
hydrographs of small agricultural watersheds.
On four of the five watershedsstudied, the
derived unit hydrographs appeared sensible.
However, the TUH varied considerably from
storm to storm. 'Goodfits were obtainedupon
synthesisof the complex storm hydrographs
a smallagricultural
watershed(from Minshall). by summation
of the TUH.
The TUH was found to vary significantly
Nonlinear
Approximation
well-known
Minshall
[1960]
hydrographs
by
Overton[1967].Thisrelationis
Lag-
tz/pe
(27)
uponby Amorocho
[9,67].
An attempt
to Here./isthewatershed
lagmodulus
andexexplain
thevariability
of theMinshall
hydro- presses
theinteraction
of theinputpeand
graphs
wasmade
by Overton
[96.7].
It was Lag,which
istheresponse
ofthesystem.
Equashown
therethatthelagtimeforeach
storm tion25 thereby
denotes
nonlinearity
of the
wasrelated
to therainfall
excess
ratein the system
fromstormto storm.
Thisrelation
form
may suggest
a similarnonlinearity
within
Lag= ta/pe
(25) storms.
However,
it seems
thatthemathematical
finesse involved in a nonlinear approach is
where./is thewatershed
lagmodulus
in inches notwarranted
at thistimebecause
of thelack
andpe is theaverage
rainfall
excess
ratein of basicinformation
needed
for calculating
inches/hour.
Thissametypeof generality
was rainfallexcess.
noticed here. Figure 8 shows the relationship
of lag to average rain excessrate for the Hastings hydrographs. This relation follows the
same general form of equation 25.
The problem of nonlinearity centers upon
defining the IUH and TUH for convolution
with rain excess so that
'zo
Q(t) =
r(r) U[r(r), t-
r dr]
(26)
< IO _
LAS
TIME
HASTINGS,
v.
SUPPLY
NEBRASKA
RATE
W$
out.
.7
.B
.9
1.0
I.I
( ( inches/ hour)
excess
rate for I-Iastings
W-3.
Complex Storms
963
546-557, 1960.
Overton, D, E., Analytical simulation of watershed hydrographs from rainfall, Proc. Intern.
Hydrol. Symp., Fort Collins, Colo., 9-17, September6-8, 1967.
Reich, B. R., Design hydrographs from rainfall
for very small watersheds, Colorado State University, CER 60, BMR 52, 1962.
Sherman, L. K., Streamflow frommrainfall by the
unit graph method, Eng. News-Record, 103,
Amorocho,
J., Thenonlinear
prediction
problem 501-505,
1932.
in thestudyof therunoffcycle,WaterRe- Snyder,
W. M., Hydrograph
analysis
by method
sources
Res.,3(3),861-880,
1967.
of least-squares,
Proc.Am.Soc.CivilEngrs.,
Brakensiek,
D. L.,Automated
system
foranalysis 793,September
1955.
of runoff
hydrographs,
U. $. Dept.Agr.,Agr. U.S. Department
of Agriculture,
Agricultural
Res.$e.rv.,
41-120,
March
1966.
Research
Service,
Hydrologic
datafor experiGray,D. M., Synthetic
unit hydrographs
for mentalagricultural
watersheds
in the United
small
watersheds,
Proc.
Am.Soc.CivilEngrs., States,1956-1959,
compiled
by HaroldW.
J. Hydraulics
Div.,88(HY4),33-54,
July1961. Hobbs,
U. $. Dept.Agr.Misc.Publ.No.945,
Minshall,
N. E.,Predicting
storm
runoff
onsmall 672pp.,November
1963.
experimental
watersheds,
Proc.Am.$oc.Civil U.S. Department
of Agriculture,
Agricultural
Engrs.,
J.
Hydraulics
Div.,
86(HY8),
August
Research
Service,
Hydrologic
data
for experi1960.
Nash, J. E., Determining runoff from rainfall,
Proc. Inst. Civil Engrs. (London), 10, 163-184,
1958.
Newton, D. W., and J. W. inyard, Unit hydrographs from complex sto.rmsusing a large computer, Proc. Am. Soc. Civil Engrs., J. Hydraulics
W.