------.

------~------

ERRATA SHEET

Page Line

8 3 Q.; = :...lscnarge at time t-~t => in cfs;

2
8 4 A"' = :;ross-sectional area at time t-~t; in ft ; and

8 27
two terms of the variable energy slope of EQ.(~).

10 2~32 ~:::i:J.ted. with Q The energy slope is determined from EQ. (14) in which
12 m
 U.S. DEPARTMENT OF COMMERCE
National Oceanic and Atmospheric Administration
National Weather Service

NOAA Technical Memorandum NWS HYDR0-16

A DYNAMIC MODEL OF STAGE-DISCHARGE RELATIONS

AFFECTED BY CHANGING DISCHARGE

D. L. Fread

Office of Hydrology

WASHINGTON, D.C.
November 1973
 CONTENTS

Abstract . 1

Section 1. Introduction . 1

Section 2. Model Development 4

Section 3. Application.of Model 10

3.1 Data Requirements . . 10

3.2 Test Applications . . 12

Section 4. Graphical Estimation of Dynamic Loop 30

Section 5. Summary . 36

Acknowledgments 37

References . . 38

Appendix A: Solution by Newton Iteration A-1

Appendix B: Computer Program (DYNMOD) . . . B-1

ii
 ABSTRACT

A mathematical model is developed to simulate the dynamic relationship

which exists betv:een stage and discharge when the energy slope is
variable due to the effects of changing discharge. The model enables
either stage or discharge to be computed if the other is specified
(measured or predicted) and the channel slope, cross-sectional area and
roughness properties are known. The model is tested on several stage-
discharge relations, which are influenced significantly by changing
discharge, and found to provide excellent results. Also presented is a
simple and easily-applied graphical procedure to estimate the magnitude
of the changing discharge effect on stage-discharge ratings.

SECTION 1. INTRODUCTION
The hydrologist is frequently concerned with the conversion of flood
discharges at a given location along a stream channel into corresponding
stages or vice versa. This is accomplished via a relation between stage
and discharge which applies to that particular location. Such a relation
or "rating curve" is usually developed empirically from a number of
previous streamflow measurements and the corresponding stages.

Unfortunately, the observed measurements of stage and discharge will

not usually form a unique relation; i.e., a single value of stage does
not correspond to a single value of discharge. This is illustrated in
Figure 1. Deviations of the measurements from a single-value rating curve
can be the result of such factors as:

1) Discharge measurement errors (Carter and Anderson, 1963);

2) Shifting control (Linsley, et ~., 1949; Corbett, 1943) due to:

a. scour and/or fill (Simons, et al., 1973);

b. alluvial bed form changes (Simons and Richardson, 1961;
Dawdy, 1961);
c. seasonal changes in vegetative growth along the channel; and
d. ice formation; .

3) Variable energy slope (Linsley, et ~., 1949; Corbett, 1943) due

to:

a. variable backwater from a downstream tributary, reservoir, or

tidal estuary;
b. conversion of discharge into or out of channel storage;
c. return of overbank flow; and
d. flow accelerations of unsteady, nonuniform flow (Henderson,
1966).

This study is concerned with only the effect of the last of the above
factors; i.e., a variable energy slope caused by flow accelerations of
unsteady, nonuniform flow (changing discharge).
 X- Observed
X

- ~~
X

~Single Value Curve

I
..r::.

Q- Discharge (cfs)-+

Figure 1. Stage- Discharge Relation.

Several authors; e.g., Linsley, et al. (1949),Corbett (1943), and

Henderson (1966), have dealt with the stage-discharge relation influenced
by a variable energy slope. It is well known that changing discharge can
produce a so-called "hysteresis loop" in the stage-discharge rating
curve such as the one shown in Fig. 2. For this type of rating curve,
there exist two different stages for each discharge. The lesser of
the two stage values is associated with the rising limb of the discharge
hydrograph and the greater value occurs during the recession of the
discharge. A hysteresis loop, which is caused by a variable energy slope
due to changing discharge, is termed a "dynamic loop" herein since it
occurs because of the changing or dynamic nature of the flood discharge.

2
 Recession Limb~
~~
~

Curve

--+
....
'+-
Limb

Q)
0')

....0
CJ)
I
~

Q-Discharge (cfs)-+

Figure 2. Stage- Discharge Relation with Hysteresis Loop

Due to Changing Discharge.

3
 SECTION 2. MODEL DEVELOPMENT

A unique, dynamic stage-discharge relation for a particular location along

a channel can be determined via a mathematical model based on the complete
one-dimensional equations of unsteady flow and the Manning equation, which
accounts for energy losses due to the resistance of the channel boundary.
These equations are derived in several references; e.g., Chow (1959) and
Henderson (1966), and are simply stated herein as:

A av + v aA + B ah = o
ax ax at (1)

av + v av + g (2.x_ + s - s ) =o
'if ax ax . 0 (2)

and

Q = 1.486 A R2 / 3 s112
(3)
n
in which x = distance
along the channel, in ft;
t = time, in sec; 2
A = channel cross-sectional area, in ft ;
B =width of the channel at the water surface, in ft;
h = water surface elevation above a datum plane, in ft;
y = depth of flow, in ft;
s = energy slope, in ft/ft;
v =mean velocity of flow across the section, in ft/sec;
So =effective bottom slope of the channel ~in ft/ft;
2
g = acceleration due to gravity, in ft/sec ;
n =Manning's coefficient, in sec/ftl/3;
Q = discharge, in ft3/sec; and
R = hydraulic radius, in ft.

In the development that follows, the following assumptions are made for a
short section of channel containing the gaging station or forecast point:

1) Lateral inflow or outflow is negligible;

2) The channel width is essentially constant; i.e., 3B/3x. ~o;
3) Energy losses from channel friction and turbulence are
described by the Manning equation; ,
4) The geometry of the section is essentially permanent; i.e.,
any scour or fill is negligible;
5) The bulk of the flood wave is moving approximately as a
kinematic wave which implies that the energy slope is
approximately equal to the channel bottom slope; and
6) The flow at the section is controlled by the channel
geometry, friction, and bottom slope and by the shape
of the flood wave.

4
 An expression for the energy slope is obtained by rearranging Eq {2)
in the following fonn:

s = {4)

The four terms on the right side of Eq (4) represent the component slopes
which produce the variable energy slope S due to changing discharge.
From left to right respectively, the four slopes are attributed to:
gravity force, pressure force, convective (spatial) acceleration, and
local {temporal) acceleration.

Using assumption (2), Eq (1) may be expressed as:

av = - .Y! !1. - ! ah {5)

ax A ax A at

Upon subst~tuting Eq (5) and V=Q/A in Eq (4), the following expression

for the variable energy slope S is obtained:

s = s0 + 1 a(O/A) (6)
g dt

Information concerning the characteristics of either the stage or discharge

hydrograph generally is available only at the location for which the
rating curve is required. Such lack of spatial resolution of the hydro-
graph requires that the derivative terms with respect to x in Eqs {4) and
(5) be replaced by equivalent expressions which can be evaluated from
the available information. Henderson {1966) shows that, if the bulk of
the flood wave is moving approximately as a kinematic wave, the following
expression may be used to eliminate the need for spatial resolution of
the specified hydrograph:

{7)

in which c is the kinematic wave velocity and r is the ratio of the

channel bottom slope to an average wave slopeo

The kinematic wave velocity c may be determined from observations of

the time interval between equal rises in stage, h, at gaging stations
along the channel. Also, c may be computed from a relationship given by
Henderson (1966) and Chow (1959):

5
 c = 1 dQ,
(8)
s dh

where dQ/dh is the slope of the single-value rating curve. If the channel
is assumed to be prismatic, the kinematic wave velocity can be computed
directly by substituting Eq (3) in Eq (~). After differentiation, the
following is obtained:

c = KV = K Q/A, (9)

where

K = 5- 2A dB/dh, (10)
) 3B2

and the hydraulic radius R is approximated by the hydraulic depth D; ioe.,

R t~~t D = A/B. (11)

This is a good approximation of the hydraulic radius for large channels.

If the hydraulic radius is used in lieu of the hydraulic dept~, the term
dB/dh in Eq (10) would be replaced by dP/dh, where P is the w.e!;:ted perimeter
of the channel cross-section; and, the term B2 would be replaced by the
product PB. From an inspection of Eq (10) it is evident that K has an upper
limit of about 1.7 when dB/dh is negligible and a lower limit of about 1.3
for a triangular-shaped channel. It has been observed that K can be approx-
imated as 1.3 for many natural channels (Corbett, 1943; Linsley, et al. , 1949) .
Although there are a number of methods for determining the kinematic-;ave
velocity, 'Eq (9) is used in this study.

It should be noted that the wave velocity is frequently greater than that
computed by Eq (9); however, this may result from the fact that the wave is
more nearly a dynamic wave than a kinematic wave. The dynamic wave velocity
as given by Henderson (1966) is:

(12)

A comparison of Eq (9) with Eq (12) indicates that the dynamic wave

velocity can be considerably larger than the kinematic wave velocity,
particularly for large A/B ratios. The dynamic wave predominates over
the kinematic wave when the channel flow is pooled such as behind a dam
or other constriction in the channel. Under this condition the flow is not
controlled by channel geometry, friction, bottom slope, and the shape of
the flood wave; therefore,flow in pooled areas where dynamic waves are
formed is not treated herein. In some instances when pooling
occurs only during the lower stages, the wave velocity changes

6
from that of a dynamic wave to more nearly that of a kinematic
wave as the stage increases. The portion of the rating curve, associated
with the higher stages when the kinematic wave approximation is more
applicable, could be determined approximately by the method developed
herein.

The value of r in Eq (7) may be taken as a constant for a particular

channel. Typical values of r range from 10 to lOOG It is used in Eq (7)
as part of a small correction which accounts for the fact that a typical
flood wave is not exactly a kinematic wave. To arrive at a value for r,
the wave slope is approximated from the characteristics of a typical flood,
event for a particular channel location. The wave slope is determined
by dividing the height of the wave by its half-length, the latter obtained
by assuming that the wave travels as a kinematic wave during the interval
of time from the initiation of the wave to the occurrence of the wave
peak at the location of concern. The half-length is determined from the
product of the average kinematic wave velocity and the time to peak stage.
Eq (9) is used to determine the average kinematic velocity, with Q and A
taken as the average values during the flood event and K assumed equal
to 1G3. Hence, the following expression is obtained for evaluating r:

r = 56200 (Qp + Qo) T S0 (13)

(hp - ho) A

where:

Q0 = discharge at beginning of typical flood, in cfs;

~ = peak discharge for typical flood, in cfs;

h0 stage at beginning of typical flood, in ft;

~ peak stage of typical flood, in ft;

A = cross-sectional ar~a associated with the average stage,

(h + h )/2, in ft ; and
p 0

T interval of time from beginning of rise in stage until the

occurrence of the peak stage, in days.

Since c and r are defined by Eqs (9) and (13), Eq {7) can be substituted
in Eq (6), with the partial derivatives in the latter replaced by finite
difference notation. After some rearrangement, the following equation is
obtained:

S = S0 +[ A + (1 - 1) (14)
XQ K

7
where:

At = small interval of time, in sees;

Q discharge at time t - ~t, in cfs;

2
A cross-sectional area at time t- 6t, in ft ; and

ohs= change in water surface elevation during the time

interval 6t,in ft/sec ( hs = h-h' , where h' is the
stage at time t- 6t) o 6t

Eq (14) is the expression for the variable energy slope S which is

caused by varying dischargeo All the terms on the right side of the
equation except S0 account for the effect of the dynamic characteristic
of the flowo If the flow is steady (unchanging with time) the energy
slope is constant and equivalent to the bottom slope, S 0 o This is
evident from Eq (14) since all terms on the right side of the equation
except the first term vanish when the flow is steady; i.e., Q' = Q,
ohs=O, and r is infinitely large since the wave slope vanishes for
steady uniform flowo

An expression may be derived for the dynamic relation between stage

and discharge when the energy slope is variable due to changing discharge.
This can be obtained by substituting the hydraulic depth for the hydraulic
radius in Eq (3) and using Eq (14) for the variable energy slope; i.e.,

Q-1.486AD
213
(s 0 +[~+( oh
s
+ Q~/A~ - Q/A
n KQ g6t
+ 2So ( 1 - BQ2 ) Jl/2 = 0 (15)
3r2 gA3

This equation differs from others given in the literature; e.g., Linsley,
et al. (1949), Corbett (1943), which include only terms equivalent to
the~irst two terms of the variable energy slope of Eq (6).

Eq (15) forms the basis of a model that can be used to determine

either. discharge when the rate of change of stage is known (as in stream
gaging) or stage when the rate of change of discharge is known (as in
stream forecasting)o These alternative conditions are denoted respectively
as Case A and Case B. A brief description of each follows:

Case A The discharge hydrograph is determined from a

specified (observed) stage hydrograph. In this
case, Q is unknown in Eq (15); the known quantities
consist of constants (S 0 , r, g, 6 t), known
functions (A, B, n, K) of the specified stage h,

8
 and known quantities (Q~,A~ , h~) associated
with the time t-~t.

Case B The stage hydrograph is determined from a

specified (predicted) discharge hydrograph.
In this situation, h is the unknown in Eq (15)
and the terms (A, B, n, K.) are functions of h.
The known quantities consist of constants
(S 0 , r, g, ~t), the specified discharge Q,
and the quantities (Q~, A', h~)associated
with the time t-~to

The unknowns Q and h in Case A and Case B respectively, are not

expressed in an explicit manner; therefore, an iterative solution via
a digital computer is required. A very efficient method for obtaining
a solution is by Newton Iteration (Issacson and Keller, 1966)e Details
of the application of the Newton Iteration technique to the solution of
Eq (15) for Case A and Case B are presented in Appendix Ao

A Fortran IV computer program for modeling the dynamic relationship

between stage and discharge for both Case A and Case B is presented in
Appendix B. The program is written for execution on a CDC 6600 computer.

9
 SECTION 3. APPLICATION Of MODEL
3.1 DATA REQUIREMENTS

In the application of Eq (15) to determine the dynamic relation of stage

and discharge, the following information is required:

1)' the effective bottom slope (S 0 ) ;

2) the cross-sectional area (A)and the surface width

(B) as functions of the stage;

3) the ~~nnings coefficient (n) as a function of the

stage; and

4) the specified (observed or predicted) stage or dis-

charge hydrograph.

3.1.1 EFFECTIVE BOTTOM SLOPE

In natural channels, the bottom slope is often quite irregular due to the
presence of deep bends, pools, crossings, shallow riffles and other irregu-
larities that occur along the length of the channel. The bottom slope (S 0 ),
used in Eq (15), is the slope that is effective in controlling the flow. A
good measure of the effective bottom slope is the water surface profile
associated with either the condition of low flow or peak flow. The low flow
profile is obtained from the minimum recorded stages at gaging stations both
up and downstream from the location for-which the bottom slope is required.
The peak flow profile is obtained from the maximum recorded stages. The low
flow profile is preferred unless it is noticeably affected by the irregu-
larity qf the channel bottom, in which case the peak flow profile should be
used. The effective channel bottom slope is computed by subtracting the
downstream stage from the upstream stage, both referenced to the same datum,
and dividing by the distance (in feet) separating the two gage locations.

3.1.2 CROSS-SECTIONAL AREA AND SURFACE WIDTH

These are geometric properties of the channel location and can be obtained
from either hydrographic surveys or stream gaging records. ,Each is a func-
tion of the stage. By plotting values of A and B versus the stage, the
extent of the variation can be determined. Approximately, three to ten
values of A, B, and h will describe adequately the variation of area and
width with stage, and linear interpolation will provide any intermediate
values required during the computation of the stage-discharge relation.

It is assumed that the properties of the cross section are known throughout
the duration of the specified stage or discharge hydrograph. Thus, if there
is significant scour or fill at the cross section during this period, the
computed stage~discharge relationship will be inaccurate because of the
unknown variations of the cross-sectional properties. The primary effect
of scour or fill on the stage-discharge relation is a change in the
position of the mean single-value rating curve. The position

10
is lowered when scour occurs and raised when fill occurs. Scour and fill have
only a secondary effect on the shape and magnitude of the dynamic loop~ since
the two most dominant terms (S 0 .and ohs) of the variable energy slope are not
affected substantially by relatively small changes in the cross-sectional area
brought about by sc'our and fill. Gradual long term changes in the cross-
sectional properties which are monitored by periodic streamflow measurements
should not prevent the use of the dynamic model. Significant scour and fill
which produces rapid changes in the cross section will present difficulties in
obtaining accurate stage-discharge relations; however~ the dynamic model can
provide a good approximation of the shape and magnitude of the dynamic loop.

3.1.3 MANNING'S COEFFICIENT, n

Stage-discharge measurements can be used to compute Manning's n via the

following ~earrangement of Eq (3); viz.,

213 112 (16)

n = 1.486 AD S0
Q

where Q is the discharge associated with the mean single-value rating curve.
It should be noted in Eq (16) that s 0 , the effective channel bottom slope, is
assumed to be equal to the energy slope. This curve is simply "sketched in"
between (h,Q) points which have been obtained from discharge measurements.
There should be points associated with both rising and falling discharges in
order to obtain a representative mean rating curve.

Generally, n will vary with stage since the roughness properties of the
channel change with stage due to (a) the variation of vegetation and channel
bank irregularities with elevation and (b) the partial_correlation of allu-
vial bed form changes with stage.

The following ld.near relation between n and stage is used herein:

n = nLo + (nL1 - nLo) (h - hLo), (17)

(hL1 - hLo)

where nLo is the n value associated with the stage hLO and nLl is the value
associated with the stage hLl" Equation (17) applies throughout the range in
stage~ hLo<h<hLl: . Som7times the n,h relation varies sufficiently to require
two l1near relat1ons; 1.e., Eq {17) for a lower range of stage hLo<h<hLl and
a similar equation for an upperrange in stage huo<h<hul

If the n,h relation is subject to change due to seasonal changes in

vegetation, man-made changes in the roughness properties of the channel

11
bank or over-bank, or changes in alluvial bed forms, the relation may be
updated using the most recent discharge measurements. The value of Q
used to compute the updated value of n via Eq. (16} should be corrected
to eliminate the dynamic effect. The corrected value of the discharge
is determined from the following:

Q = Qm- 1.486 A D2/3 ( s1/2- so1/2) (18)

n

where Qm is the ~easured discharge. The values of A, D, and n are obtained

from their currently known relationships with stage for the particular stage
associated with Qm The energy slope S is determined from Eq. (13} in which
Q is replaced by Qm

3.1.4 SPECIFIED HYDROGRAPH

The specified (known} stage hydrograph is obtained from a measured stage

versus time record. The record may be daily, hourly, or have a smaller
temporal resolution. The degree of resolution should be such as to pro-
vide an adequate definition of the variation of stage with time.

The specified discharge hydrograph is obtained from a record of the

measured discharges or from a forecast discharge hydrograph. The latter
may be the product of some flow routing technique which does not provide
both discharge and stage; e.g., the Muskingham method, the Tatum method.

~t in Eq. (15) should not be greater than the temporal resolution of the
specified stage or discharge hydrograph. If the specified hydrograph has
irregularities, a ~t which is less than the resolution of the hydrograph
allows the resulting irregular stage-discharge rating curve to be defined
more accur~tely. In this study, a ~t of 3 hours is used with specified
hydrographs that have a 24-hour resolution.

3.2 TEST APPLICATIONS

The dynamic relation between stage and discharge is modeled herein for
locations on the lower Mississippi, Atchafalaya, and Red Rivers. In each
of these locations, the stage-discharge relation is influenced by a variable
energy slope due to changing discharge.
The necessary information (bottom slope, cross-sectional area and surface
width as functions of stage, and Manning's n} for each of the locations is
given in Table 1. The cross-sectional properties represent time-average
values. The departures of the cross-sectional properties for any 1 year
from the average values shown in the table are less than approximately 3
percent. If the departures were considerably larger, cross-sectional proper
ties applicable to each year would improve the accuracy of the computed
stage-discharge relations. At each location, the floods, which were selected
for testing the model, caused minimal scour and fill. The computed values

12
 TABLE 1.--CROSS-SECTION AND HYDRAULIC DATA FOR LOCATIONS
ON THE LOWER MISSISSIPPI, RED, AND ATCHAFALAYA RIVERS

Location s0 Year h h ! h A B
nLO nLl LO Ll
11uo I nUl huo hUl

Mississippi River 1966 0.0148 0.01354 5. 50. 16.0 72500. 3000.

Tarbert Landing, 0.0000143 1967 0.01560 0.01542 5. 50. 34.0 134000. 3540.
Louisiana
1969 0.0159 0.01392 5 50. 41.2 164000. 3630.
48.0 200000. 3690.
15.0 55300. 3020.
Mississippi River 16.0 58000. 3060.
18.0 64400. 3160.
1-' Red River Landing 0.0000143 1963 0.0108 0.0113 ~5. 50.
w 19.0 68000. 3360.
Louisiana
30.0 104000. 3420.
37.0 133500. 4310.
44.0 163300. 4400.

Red River
1964 0.031 0.018 ~5. 55. 0.0180 0.0112 55. 75. 45.0 4800. 463.
Alexandria, La. 0.0000233
1966 0.031 0.018 ~5. 55. 0.0180 0.0130 55. 80. 52.0 8300. 528.
60.0 13000. 576.
80.0 26600. 696.

Atchafalaya River 1.0 20000. 960.

Simmisport, La. 0.0000211 1964 0.0395 0.0196 5. 16.1 0.0196 0.0143 16.1 40. 20.0 42100. 1250.
40.0 66000. 1425.
 of Manning's n vary somewhat for each flood. These variations are due appar-
ently to the changing alluvial bed forms being not only a function of stage
but also dependent on water temperature, discharge, and sediment load. Also,
changes in bank vegetation due to seasonal effects may be responsible for
some of the variability.

Using the information shown in Table 1 and observed stage-hydrographs,

stage-discharge rating curves and discharge hydrographs were computed for
selected floods at the locations discussed below.

3.2.1 MISSISSIPPI RIVER, TARBERT LANDING, LA.

Figures 3-8 illustrate computed stage-discharge relationships and discharge

hydrographs for floods which occurred during the years 1966, 1967, and 1969.

In Fig. 3, the solid line representing the computed rating curve for the
flood of 1966 was obtained by 1) using Eq (15) to compute the discharge
hydrograph shown in Fig. 4 from the specified stage hydrograph shown in
the insert of Fig. 3 and 2) plotting the observed stage against the computed
discharge. Measured values of stage and discharge are shown in Fig. 3 also.
The computed rating curve has a substantial dynamic loop; the discharge value
of 550,000 cfs has two associated stages, one on the rising limb and one on
the recession limb of the specified stage hydrograph, which differ by approx-
, imately 7 feet. The loop rating curve has irregularities which reflect each
variation in the rate of change of the stage hydrograph. A comparison of the
computed discharges shown in Fig. 4 with the observed discharges is quite
good, the root mean square (rms) error being less than 4 percent.

The 1967 multiple-peaked stage hydrograph, shown in the insert of Fig. 5,

yields the complex computed stage-discharge relation shown in the figure.
Each peak of the specified stage hydrograph produces a different loop in the
rating curv~. As shown in Fig. 6 the computed discharge hydrograph agrees
quite well with the discharge observations, the rms error being only about
3 perc~nt.

The 1969 observed stage hydrograph shown in the insert of figure 7, yields
the single-looped rating curve of the :tigure. The uniformity of the stage
hydrograph produces the relatively smoo.th computed rating curve . . A comparison
of computed and observed. discharges is shown in figure 8. The rms error is
about 2 percent.

3.2.2 MISSISSIPPI RIVER, RED RIVER LANDING, LA .

.The 1963 observed stage hydrograph, shown in the insert of Fig. 9, yields
the computed stage-discharge rating curve of the figure. The dynamic loop is
rather significant; for a discharge of 450,000 cfs, the maximunL difference
between stages of the rising and falling limbs of the hydrograph is about
9 feet. This is a result of the rather severe rate of change of stage
which exists for both the rising and the recession limbs of the stage hydro-
graph. Also, noticeable variations in the rate of change of the stage
hydrograph are reflected as irregularities superimposed on an otherwise

14
 G> OBSERVED
-COMPUTED

45

--+35
' +-
'Q;
0)

-0
en
L__
~--
~
I
..c: 30
~
-"-
" +-
_. 35
Q)
0)

25 -0
en
..c:
I
25

20
20 40 60
t- Time( days)+-
151 I I I I I I I I I I

- 200 300 400 500 600 700 800 900 1000 1100
Q- Discharge(1000cfs)+-
F i g u r e 3. Stage - Di s c h a r g e ReI at ion for M i s s i s s i p p i Ri ve r 1 Ta r be r t l a n d i n g 1 l a..
( 2 I 9 I 66- 4 I 11 I 66).
 0 OBSERVED
-COMPUTED
1200 Comparison of Computed and
Observed Discharges:
Root mean square error __ 3.8'ro

+ 1000
en
'"I-
u
0
0
0
-
...........
Q)
0)
800

...... '-
0
0\
..s::::: 600
u
en
-
a
I
(J 400

200

OT-----+-~--+-----r-----r-----~----~--------------------~
0 10 20 30 40 50 60
t-Time(days) ~
Figure 4. Discharge Hydrograph for Mississippi River, Tarbert landing, La.
( 2 I 9 I 66- 4 I 11/.66).
 e OBSERVED
- COMPUTED

e
40

-
....,
f-< --
~ 35
Q)
0>

- 0
t J)
I
...s::::: 25

e 40 80 120
t- Time (days) +-

15'+---~--~~--~---+----~--~---+------------~
200 300 400 500 600 700 800 900
Q- Discnarge ( lOOOcfs) +-
Figure 5. Stage - 0 i s c h a r g e R e Iat ion for M iss i s s i p pi R i v e r Ta r bert
1
Land i n g 1 La.
( 2 I l I 67- aI 3 I 6 7).
 00BSERVED
-COMPUTED
Compo rison of Computed and
800. Observed Discharges:
Root mean square error __ 3.2 o/o

+ 700
-.....
en
u
g 600
--
0
Cl)
en
~ ~

OJ 0
..s:::. 500
u
en
-
0
I
0 400

300

2oo1L-+--+--+--+---j-=-~~~~-~--:c;--
0 20 40 60 80 100 120 140 160 180
t -Time(days) +
F i gu r e 6. D i s c h a rg e Hy d r o g rap h for M i s s i s s i p pi R i v e r 1
T a r bert La n d i n g 1 La .
(2/ 1 /67-8/3/67).
 50 0 OBSERVED
-COMPUTED

40
+

-
( /')

~ 1:301 ~7 I I i.......
-
~ 40
C)
0

254-' / 1 en
I
..s:::: 30

I ~ I
20

15 I I I I I I I I I I I

200 300 400 500 600 700 800 900 1000 1100
Q- Discharge (lOOOcfs) ~
f ig ure 7. 5 tag e - 0 i s c h a r g e ReI at i o n for M i s si s s i p p i R i v e r , Ta r be r t Land in g , La .
( 1 I 23 I 69- 3 I 26 I 69).
 1100~------------~=-~-----------------------------------------

0 OBSERVED
\ -COMPUTED

\ Comparison of Computed and

Observed Discharges:
Root mean square error--2.3%

~
t
...-
800 \
VI
......
u
0
0

--
0

G)
C)
"-
0
N
0 ..s:::. 600
u
VI
-
0
I
c

0 10 20 30 40 50 60 70
t-Time (days) +-
Fjgure 8. Discharge Hydrograph for Mississippi River, Tarbert Landing, La.
( 1 I 23 I 69- 3 I 26 I 69 ).
 00BSERVED
-COMPUTED

t
-.-....
- Q)
C)

-
0
( /)
I
J~t
~ I 45
..r::.
/,/"
I
J
I
N
I-' 30-J_ ~cv /
-+-
1.:::=. 35
Q)

-
C)
0
( /)
I
..r::. 25

15i~~L__L~L_~~~~~;
0 20 40 60
t-Time(days)+-

300 400 500 600 700 800 900 1000

Q- Discharge (1000cfs}~
Figure 9. Stage- Discharge Relation for Mississippi Rive~ Red River Landing, La.
( 3 I 11 I 63- 5 I 13 I 63 ) .
smooth single-looped rating curve. The computed discharge hydrograph, shown
in Fig. 10, compares with measured discharges quite well, the rms error being
1.6 percent.

3.2.3 RED RIVER, ALEXANDRIA, LOUISIANA

Figures 11 to 14 illustrate stage-discharge relation and discharge hydro-

graphs for floods occurring during the years of 1964 and 1966.
The computed 1964 rating curve in Fig. 11 consists of several dynamic loops
each of which'correspond to individual peaks of the multiple-peaked stage
hydrograph shown in the insert. The largest peak corresponds to a dynamic
loop having approximately a 3-ft. difference in the rising and falling stages.
The computed discharge hydrograph is shown in Fig. 12. The rms error of the
computed versus observed discharges is 8.4 percent.

The single-peak and uniform variation of the observed stage hydrograph of

1966, shown in the insert of Fig. 13, yields the smooth single-loop rating
curve shown in Fig. 13 and the computed discharge hydrograph of Fig. 14.
The rms error of the computed versus observed discharges is 4.6 percent.

3.2.4 ATCHAFALAYA RIVER, SIMMESPORT, LOUISIANA

The 1964 observed stage hydrograph, shown in the insert of Fig. 15, results
in the computed rating curve of Fig. 15. The irregularity of the stage
hydrograph causes the computed rating curve to contain several small loops
and other irregularities. The dynamic loop of the rating curve has a maxi-
mum 8-ft. difference in rising and falling stages corresponding to the single
discharge value of 200,000 cfs. The irregularities observed in the stage
hydrograph are reflected in the computed discharge hydrograph of Fig. 16.
The computed and observed discharges have an rms error of 7.5 percent.

3.2.5 CONCLUDING REMARKS

, In each of the previous applications, the dynamic loop of the computed
stage-discharge rating curve is quite significant. This occurs, even though
the maximum rate of change of stage is of the order of a few feet per day,
because the effective bottom slope at each location is quite small. The
importance of the bottom slope and the rate of change of stage will be dis-
cussed further in section 4. Equation (15) provides computed discharges
which agree closely with the measured values,, the average rms error being
approximately 4 percent.
In each of the examples, the discharge hydrograph was computed from a
specified or given stage hydrograph. This condition corresponds to the
Case A condition identified previously in Chapter 2. This condition was
used to test the dynamic model since the observed stage hydrograph is
usually better defined than the discharge hydrograph. Thus, Case A allows
a more meaningful test than would Case B. The Case B alternative function
of the dynamic model was tested using the computed discharge hydrographs of
a few of the previous applications as the specified discharge hydrograph.
None of the computed stage hydrographs are presented since each was found
to be essentially identical with the corresponding observed stage hydrograph.

22
 0 OBSERVED
-COMPUTED
Comparison of Computed and
+ Observed Discharges:
-...
en
u
Root mean sq-uare error __ 1.6%

0
0

--
0

Q)
Ol
'-
0
.s:::. 600
u
Cl)
N
w -
0
I
0 500

10 20 30 40 50 60
t- Time (days)+-
Figure 10. Discharge Hydrograph for Mississippi River, Red River Landing, La.
( 3/11/63-5/ 13/63).
 -so_,...------------------------------,

75 _J._ 0 OBSERVED
-COMPUTED

70

+ 65
--....
.......

~ 60
0
N ,._
..!:'- V)

..:: 55

50

45

40 I I I I I I I I I I I I I

0 10 20 30 40 50 60 70 80 90 100 110
Q- Discharge (1000 cfs) +
Figure 11. Stage-Discharge Relation for Red River, Alexandria, la.(2/5/64-6/l7/6.4).
 0 OBSERVED
-COMPUTED
Comparison of Computed and
Observed Discharges:
+ 120 Root mean square error--8.4%
-
.....
V)

u
0
0
0 100'
-
Q)
0>
1.-
0
...c
u
V)

,..,o
lJ1
I
0

40

20

20 40 60 80 100 120 140

t - Time (days) +
figure 12. Discharge Hydrograph for Red River, Alexandria, La. (2/5/64- 6/17/64).
 80 ~

e OBSERVED
75
-COMPUTED

70

65r / / I 85

-+- 60 -t-
~ 1
Cl>
/ / 75

N-
0')

"'.V>
~
0
55
1 // 1~.,
- Cl>

-
0')
0
( f)

50 _j_ // I ..l: 55

45
20 40 60 so
t-Time (days)-+

40 I 1 1 1 I I . _I I I I !
0 20 40 60 80 100 120 140 160 180 200
Q- Discharge (lOOOcfs) +-
Figure 13. Stage- Discharge Relation for Red River, Alexandria, La.(4/13/66-7/l /66).

~~3- '~--"~~
 ~ .-.~

160

140 -4- I \ 0 OBSERVED

-COMPUTED
Comparison of Computed and
Observed Discharges:
I I . Root mean square error__ 4.6%
+ 120

....-
u
Cl'l

0
0 100
o-
-
,.....
(I)
C>
"-
0 80
..s::.
N
u
Cl'l
'-1
-
0
I
() 60

40

20

o~---------~----------+----------+------~--1---------~r---------~--------+-------+------------------
0 10 20 30 40 50 60 70 80
t-Time (days)+
Figure 14. Discharge Hydrograph for Red River, Alexandria,la. (4/13/66-7/ 1 /66).
 ,''40-,---------------------------.

0 OBSERVED
-COMPUTED

+
~ C
20
-l #/
-
V)
0

I
N .S:::.
00

0 .40 so 120 160

t- Time (days)..-
0
0+----r---+----r---+---~---+----~--+-------~
0 50 100 150 200 250 300 350 400
Q- Discharge (lOOOcfs) ~
Figure 15. Stage- Discharge Relation for Atchafalaya River, Simmesport, La.
(1/7/64- 6/16/64).
 0 OBSERVED
-COMPUTED
Comparison of Computed and
Observed Discharges:
Root mean square error__ 7.5%

+ 400
N
\.0
-
II)
~
u
0
0
0 300
-
,..-

Q)
C)
~

0
..s:::: 200
u
II)

0
I
() 100

0 I I I I I I I I I I

0 20 40 60 80 100 120 140

t -Time (days) +
Figure 16. Discharge Hydrograph for Atchafalaya River, Simmesport, La.
( 1 I 7 I 6 4 - 6 I 16 I 64 ).
 SECTION 4. GRAJ?HICAL ESTIMATION OF DYNAMIC LOOP

When stage-discharge measurements do not plot as a single-value rating

curve, this is often attributed to some combination of the effects of scour
and fill, bed form changes, or measurement errors. A simple and easily-
applied graphical procedure is presented in this section to detennine the
approximate magnitude of the deviations from the single-value rating curve
which are due only to the effects of changing discharge. If this estimate
is considered to be significant, the use of the dynamic model presented
previously in sections 3 and 4 is warranted.

In order to estimate the magnitude of the effect of changing discharge on

the stage-discharge relation, some simple measure of this effect is
required. Referring to Fig. 2, such a measure is the difference between the
stage associated with the single-value curve and that associated with either
the rising or recession limb of the hydrograph. This measure of the dynamic
loop is denoted as 6h. The following development will provide a means of
estimating the magnitude of 6h for a particular channel location and flood
event.

By using the hydraulic depth in lieu of the hydraulic radius, Eq(3) can
be rearranged to yield:

D =[ Q n
1.486 B
J0.6 s-0.3
(19)

Eq(l9) can be used to obtain an approximation for Llh by:

1) substituting Eq(3), in which R is replaced by D and S is replaced

by S0 , for Q in Eq(l9);

2) assuming the channel width B is constant throughout the .Llh change

'in stage; i.e., dB/dh = O; and

3) assuming a small change in hydraulic depth is equivalent to a small

change in stage; i.e. , .Llh = LlD.

The approximation is given by the following expression:

(20)

where the energy slope S is approximated by:

s ~ s0 + ~h
s
(21)

30
 Eq(21) was obtained by:

1) using only tile first two terms, which are the most significant,
of the right si.de of Eq(l4};

2) assuming K is constant and equal to 1.3; and

3) assuming Q in Eq (l4} may be approximated by Eq (3} in which S0

is used in lieu of S.

An inspection of Eqs(20) and (21) indicates that the independent parameters

necessary to approximate the
.
magnitude of ~h in the' dynamic loop are s 0 ' oh s '
D, and n. Thus, b y allowlng these parameters to assume values which
encompass the practical range of each, Eqs(20) and (21) may be used to deter-
mine the ~h associated with various parameter values. The results of these
computations are summarized by the family of graphical relationships shown
in Fig. 17.

The following steps summarize the use of Fig. 17 to obtain an estimate of

the ~h magnitude of the dynamic loop:

1) compute the value of Q2/3 ;

n
2
2) use Graph A and the values of D / 3 and oh to obtain Ko;
n
3) use Graph B and the values of S0 and K to obtainS;
0

4) compute the value of S0 /S; and

5) use Graph C and the values of S /S and D to obtain ~h.

0

The following example illustrates the use of Fig. 17 to estimate ~h:

The approximate value of 6h is to be determined when S0 = 0.00008~

D = 20.0 ft, n = 0.020, and oh = 1.0 ft/hr. First, the parameter JJ..z/3 is
computed to be 368. With this value and the given value of oh, n Graph A
is used to obtain a value of 0.82 for K0 Then, with the value of Ko and
the given value of S0 , Graph B is used to obtain a value of 0.000137 for S.
Finally, the ratio S0 /S is computed to be 0.584 and this value, along with
the given value of D, is used in Graph C to obtain the ~h value of 2.9 ft.

The curves of Fig. 17 enable ~ to be determined for any combination of the

relevant parameters (S 0 , oh, D, and n). Of these, S0 and oh are the more
dominant; therefore, by fixing the values of D and nat particular values,
valuable insight concerning the effect of S0 and oh on the magnitude of ~h
may be attained. Thus, if D and n are assumed constant at 10 ft and 0.015,
respectively, the curves shown in Fig. 18 are obtained. If the values of D
and n are assumed constant at 30 ft and 0.015, respectively, the curves
shown in Fig. 19 are obtained. The following conclusions may be drawn from
an examination of Figs. 18 and 19:

1) ~ increases as oh increases;
31
 100

1.o

+. 0.1
0
~-

o.o1

0.001 10.
100 200 300 400 500 600 700
02/3 +- 9.
n
a.

7.
0.001

6.
t
-
..c:::
5.

<l
4.

CJ) 0.0001
3.

2.
IGRAPHBI
L: _____ _j
1.

o.
0.00001 0.0001 0. 001 o. .2 .4 .6 ' .8 1.0
_g_.
So 5
Figure 17. Generalized Graphical Estimation of Magnitude ( ah)
of the Dynamic Loop. 32
 D= .1o. ft
n = .015

1.0

0.1

0.01
0.00001

Figure 18. Graphical Estimation of Magnitude ( ah) of the Dynamic

loop for Constant D of lOft. and n of 0.015.
33
 D= 30. ft
n= 0.015

Sh ( ft /hr)
jl
)I
)I

1.0

0.1

0.00001 0.0001 0.001

!s ft /mi s! ft /mi

s0 ...

Figure 19. Graphical Estimation of Magnitude (~h) of the Dynamic

loop for Constant D of 30. ft. and n of 0.015.
34
 2) ~ decreases as S0 increases;

3) tJt increases as D increas.es;

4) Ah is ins.ignificant (say less than about 0.1 ft) when S0 > 0.001
for most .naturally occurring values of ISh; i.e., oh < 4 ft/hr

5) ~ can be significant when 0.0001 < so < 0.001 for 0.1 < oh
< 3 ft/hr; and

6) ~ is significant when so < 0.0001 for oh < 0.05 ft/hr.

The relation of Manning's n to ~ is determined by selecting constant

values for S0 , oh, and D and using Fig. 17 to determine Llli for different
values of n. In this manner, it is found that ~ increases as n increases.

35
 SECTION 5o SUMMARY

From the equations of unsteady flow, a mathematical model has been

developed which computes either stage or discharge if the other is
specified along with the channel slope, cross-sectional properties, and
Manning's n. The model simulates the dynamic relation which exists
between stage and discharge due to the effect of a variable energy slope
caused by changing discharge. This effect, which is often observed as a
loop in stage-discharge rating curves, was accurately modeled in several
test applications; however, caution must be exercised when applying the
model to locations where significant scour, fill and/or bed form changes
occur since the model is only as accurate as the specified data.

The model can be used in forecasting to convert the forecast discharge

hydrograph into a stage hydrograph which properly reflects the dynamic
relation that exists between stage and discharge due to a variable -
energy slope. Also, the model can be used in stream gaging to convert
an observed stage hydrograph into a discharge hydrograph when the effect
of changing discharge is significant.

A convenient graphical procedure has been presented to estimate the

magnitude of the changing discharge effect on stage-discharge ratings,_
This is useful in determining if the magnitude of the dynamic loop
warrants the use of the mathematical model. The magnitude of the dynamic
loop has been found to be related inversely to the channel bottom slope,
and directly to the ~ate of change of stage, the hydraulic depth, and the
Manning's n. As a general rule, the dynamic loop may be significant if
the channel bottom slope is less than 0.001 ft/ft (about 5 ft/mile) and
the rate of change of stage is greater than about 0.10 ft/hr. This rate
of change of stage may decrease with decreasing values of the bottom slope.

36
 ACKNOWLEDGMENTS

The author wishes to acknowledge the assistance given by Dr. E. 1. Peck,

Dr. R. A. Clark, and Mr. W. T. Sittner in reviewing the report and
Mr. R. Tubella in drafting the figures.

37
 REFERENCES

Carter, R. Wo, and I.E. Anderson, "Accuracy of Current Meter Measurements,"

Journal of the Hydraulics Division, ASCE, Volo 89, No. HY4, Proc. Paper
3572, pp. 105-115, July, 1963.

Chow, V.To, Open-Channel Hydraulics, McGraw-Hill Book Company, New York,

1959.

Corbett, D.M. , "Stream Gaging Procedure," Geological Survey Water Supply

Paper 888, u.s. Government Printing Office, Washington, 1943.

Dawdy, D.R., "Depth-Discharge Relations of Alluvial Streams--Discnntinuous

Rating Curves," Geological Survey Water Supply Paper 1498-C, 'U.s.
Government Printing Office, Washington, 1961.

Hende~son, F.M., Open Channel Flow, MacMillan Company, New York, 1966.

Isaacson, E. and H.B. Keller, Analysis of Numerical Methods, John Wiley

& Sons, Inca, New York, 1966.

Linsley, R.K., M.A. Kohler, and J.L.H. Paulhus, Applied Hydrology,

McGraw-Hill Book Company, New York, 1949.

Simons, D.B., and E.V. Richardson, "The Effect of Bed Roughness on Depth-
Discharge Relations in Alluvial Channels," Geological Survey t-ater Supply
Paper 1498-E, u.s. Government Printing Office, Washington, 1961.

Simons, D.B., M.A. Stevens, and J.H. Duke, "Predicting Stages on Sand-Bed
Rivers," Journal of the Waterways, Harbors and Coastal Engr. Division,
ASCE, Vol. 99, No. WW2, Proc. Paper 9731, pp. 231-243, May, 1973.

38
 APPENDIX A

SOLUTION BY NEWTON ITERATION

A nonlinear equation may be solved by a functional iterative technique

such as Newton iterationo Consider the following equation expressed in
functional form:

f(x) =0 (A-1)

The solution of Eq (A-1) is obtained in an iterative manner, proceeding

from a first solution estimate xk towards succeeding improved estimates
xk+l, which tend to converge toward an acceptable solution. The orderly
procedure by which the improved solution estimate xk+l is obtained so
that it converges to an acceptable solution is known as Newton iteration
and is described as followso

A nonlinear equation such as Eq (A-1) may be linearized by using only

the first two terms of its Taylor series expansion at xk; ioe.,

df(xk)
k
f(x) = f(xk) + - - - (x-x ) (A-2)
dx"

The right sid~ of Eq (A-2) is the linear function of xk. that best
approximates the nonlinear function f(x) ~hich is evaluated at xk. An
iterative procedure, which will cause f(x ) to approach zero as the quantity
(x-xk) approaches zero, can be obtained from Eq (A-2) by setting f(x)
equal to zero and replacing x with xk+l, which will be an improved solution
estimate for x if the iterative procedure is convergent. Hence Eq (A-2)
takes the form:

xk+l = xk - f(xk) (A-3)

df(xk)/dxk

where the k superscript denotes the number of iteration.

'

Eq (A-3), the general iteration algorithm of Newton, is repeated until

the difference (xk+l - xk) is less thane which is a suitable error
tolerance for the solutio~ of ~ (A-1). When this occurs, the iteration
process has converged; i.e., xk l has approached x to within the
prescribed error tolerance E.

A-1
 The convergence of the iteration process depends on a good first
solution estimate xk=l. If the estimate is sufficiently close to x,
convergence is attained; and it is at a quadratic rate; i.e. second order,
since the iterative procedure involves the first derivative. The
nonlinear equation which is solved by the Newton iterative algorithm in
this report is a time dependent finite-difference equation. A first
estimate of the solution is obtained by using the solution associated with
the time t-~t. In this study the iteration process always converged.
The convergence process can be hastened when the first solution estimate
xk=l is made closer to the acceptable solution. A simple linear extra-
polation is used to provide better first solution estimates. Thus,

k=l
x. = x. + (x. - x. 2 )/2 (A-4)
J J- 1 J- 1 J-

where the j subscript denotes the solution at time t and j-1 denotes
the solution at time t-~t, etc.

In the following, the Newton iteration algorithm is applied to Eq (15)~

which is presented here for convenient reference:

Q - 1.486 ADZ/J [ S
0
+ [ ~_ + ( 1 - _! ) BQ ] oh +Q,#f/A' - Q/A
s
n ~ K g~ g.llt
+ 2S 0 ( l _ BQ2 ) ] l/2 = O (A-5)
3r2 gA3

First, the Case A condition is treated where the discharge Q is the

unknown. in Eq (A-5) and then the Case B condition is presented in which
the stage h is the unknown.

Case A:

Eq (A-5) can be solved for Q at time t as follows:

Qk+l = Qk _
f(Qk) (A-6)
df(Qk)/dQk
where the superscript k denotes the number of iteration; f(Qk) is
Eq (A-5) evaluated with the unknown Q replaced by the approximation
qk; and the term df(Qk)/dQk is the derivative of f(Qk) with respect to
qk. Thus,

= qk _ L L l/2 (A-7)
2 0 '

A-2
where:

13 + 14 + 15 Qk + 16(Qk)2 ' (A-8)

qk
213 ,
1.486 AD (A-9)
n

13 = so + 2 so + Q"' (A-10)
3 r2 g A"' /i.t

1 = A oh8 (A-ll)
4
K

(1 - l) B oh 8 - 1 , and (A-12)
K g A2 .g A At

L6 = - 2 S0 B , (A-13)
3 r2 gA3
in which

D = A , (A-14)
B

oh (h - h"') (A-15)
s
/1t

dB/dh.
K=2_l_A__ _ (A-16)
3 3 B2

dB/dh (B-B ... ) (A-17)

(h-h ... )

(A-18)

Also,

1 ""' 0.51211 (A-19)

1 172
0
where:

(A-20)

A-3
 In the above equations, A and B are known functions o~ the stage and are
evaluated.at h; B.. , A~, h~ are known from the time period previous; i.e.,
t~ or t-ilt; S , r, n , n
11 , hLo, h 11 are constants.
0 10

Case B:

Eq(A-5) can be solved for h at time t as follows:

hk+l = hk - f (hk) ~--='----.,-

(A-21)
df(hk)/dhk

where the superscript k denotes the number of iteration; f(hk) is Eq(A-5)

evaluated with the unknown h, which is also implicitly contained in the
terms D, t'
B, n, K, and oh , replaced bk the approximation ~k; and the
term df(h )/dhk is. the deri~ative of f(h ) with respect to h. Thus,

f(hk) = Q- 1.486 AD 2/ 3J 0 1 / 2 (A-22)

n

where:

J 2 + (J 3 A+J 4 ~ ) (h k -h . . ) + J 5 + J 6 ~ (A-23)
A2 1\ A3

in which
J = + 2 S0 + Q ... (A-24)
2 80 3 r2 A... ~t
g

1 (A-25)
K Q Lit
1 (A-26)
(1-rz) _Q_
g Lit

J5 _.Q__, and (A-27)

g Llt
. -2 s q2
J6 = 3 r~ g
(A-28)

In the above,
5 2 .A dB/dhk
K = 3 - 3 --2 (evaluated at hk) , (A-29)
B
k (Bk .--B~} (A-30)
dB/ dh = (hk -h-,.)

and

A-4
 -1.486[J 1/2 [ ~ d(D2/3) + D2/3 B - _A_D_2_/_3 dn/dhk ]
o n dhk n n2

2
+ 0 . 5A D I 3 J 1 ] , (A-31)
n J 1/2
0

(A-32)

+ (nLl -nLo)
n = niD (hk-h )
Lo
(A-33)
(hLl -hLo) '
n -nLo)
dn/dhk = ( Ll -h ) , and (A-34)
(hLl Lo

:d(D2/3) 2
= -3
D2/3. (~ - dB/dhk)
A B
. (A-35)
dhk

In the above equations A and B are specified functions of the stage and
are evaluated at hk; B... , A. . , h . . are known from the previous time t-~t; and
g, ~t, S 0 , r, nLo, nL 1 ,hLo' hLl are constants.

For either Case A or Case B!l the solution of Eq(A-5) via Newton iteration
requires only about two iterations when the following convergence criteria
are used:

(A-36)

and

(A-37)

where:

1.0 cfs' (A-38)

and.

0.001 ft. (A-39)

A-5
 APPENDIX B

COMPUTER PROGRAM (DYNMOD)

The. Fortran IV computer program presented in this section will compute

either the Case A or Case B conditiono An index labeled (IQH) functions
as a decision variable so that the appropriate branches of the program are
followed for either Case A or Case B. If IQH is assigned a value of
one (integer), the Case A condition is performed in which discrete values
of the discharge hydrograph are computed using an input of discrete
values of the stage hydrograph. If IQH is assigned a value of two, the
Case B condition is performed in which discrete values of the stage
hydrograph are computed using an input of discrete values of the discharge
hydro graph.

The program (DYNMOD) is designed with a (Main Program) and several

subroutines for performing repetitive computations. Input/output
information is handled by the Main Program. It also contains the basic
program logic. Subroutine (QSOLVE) solves Eq (15) for the discharge Q
using Newton Iteration. Subroutine (HSOLVE) solves Eq (15) for the stage
(H) using Newton iteration. 9.lbroutine (SECT) computes the geometrical
properties of the cross-section for a particular elevation of the water
9..1rface. Rubroutine (FRICT) computes Manning's n for a particular eleva-
tion of the water surface.

The output furnished by the program includes all the input information,
as well as the following:

Time The time (in hours) associated with the

computations being printed-out; the time
corresponds to the temporal resolution of
the specified (input) hydrograph.

Stage The actual stage (in ft above mean sea level

or above a datum plane) which is read-in as
the specified stage in the Case A condition or
which is computed as in the Case B condition.

Discharge The actual discharge (in cfs) which is read-in

as the specified discharge in the Case B
~condition or which is computed as in the Case A
condition.

Normal discharge __ A fictitious discharge (in cfs) which would occur

simultaneously with the actual stage if there
were no dynamic effect due to a variable energy
slope caused by changing discharge; this
discharge would occur if the energy slope were
equivalent to the effective channel bottom slope.

B-1
 Dynamic effect ____ A value (in cfs) which is the algebraic
difference between Discharge and Normal
Discharge; this represents the effect of the
flow dynamics on the discharge.

Normal stage --.-- A fictitious stage (in ft above MSL) which would
occur simultaneously with the actual discharge
if there were no dynamic effect.

Dynamic effect ---- Also used as a value (in ft) which is the
algebraic difference bet\~een Stage and Normal
Stage; this represents the effect of the flow
dynamics on the stage.

Comment statements, inserted within the program, are provided as

additional clarification.

The Fortran IV computer program (DYNMOD) and a typical INPUT/OUTPUT

listing follows:

B-2
 PROGRAM DYNMOD CINPUTtOUTPUTl
DIMENSION HU<200ltT1C200)tQUC200)
*** THIS PROGRAM IS A DYNAMIC MODEL OF THE STAGE-DISCHARGE RELATION.
*** THE MODEL CONSIDERS 'THE VARIABLE ENERGY SLOPE DUE TO CHANGING
*** DISCHARGE WHICH CAUSES A LOOP IN THE.STAGE-DISCHARGE RATING CURVE.
COMMON /All CMLltCML2tCMU2tCMHLltCMHL2tCMHU2
COMMON /A2/ HSClOltBSClQ),AS<lOltNCS
*** GRAVITY ACCELERATION CONSTANT
G=32.172
*** ITERATIVE CONVERGENCE CRITERION
EPH=O.OOl
EPQ=J.O
*** INPUT DATA
*** IF IQH=lt THE STAGE HYDROGRAPH IS REQUIRED AS INPUT DATA AND THE
*** DISCHARGE HYDROGRAPH IS COMPUTED BY THE PROGRAM.
*** IF IQH=2t THE DISCHARGE HYDROGRAPH IS REQUIRED AS INPUT DATA
*** AND THE STAGE HYDROGRAPH~ IS COMPUTED BY THE PR-OGRAM.
*** NU IS T~E NUMBER OF POINT VALUES OF THE SPECIFIED HYDROGRAPH.
*** NCS IS THE NUMBER OF VALUES OF HS.
*** GZ IS THE ELEVATION <FT> OF THE GAGE ZERO ABOVE MSL.
*** DT IS THE DELTA TIME STEP CHRS) AT WHICH THE COMPUTATIONS PROCEED
*** DTHU IS THE TIME INTERVAL CHRS> BETWEEN VALUES OF SPECIFIED
*** HYDROGRAPH.
*** HS IS THE WATER SURFACE ELEVATION <ABOVE MSLl FOR KNOWN VALUES OF
*** CROSS-SECTIONAL AREA AND WIDTH.
*** BS IS THE KNOWN VALUE OF THE SURFACE WIDTH FOR THE ELEVATION HS.
*** AS IS THE KNOWN VALUE OF THE CROSS-SECTIONAL AREA FOR ELEV. HS.
*** HU IS SPECIFIED STAGE HYDROGRAPH VALUE FOR EACH DTHU CHRS1
*** TIME INCREMENT.
*** QU IS THE SPECIFIED DISCHARGE HYDROGRAPH VALUE FOR EACH DTHU
*** CHRS> TIME INCREMENT.
READ 51, IQH
PRINT 50, IQH
READ ~l,NUtNCS,GZ,DTtDTHU
PRINT 55, NU,NCS,GZ,DTtDTHU
READ 52t<HSCKltK=ltNCS>
PRINT 60
PRINT 52tCHSCKltK=ltNCS)
READ 52tCBSCKltK=ltNCS>
PRINT 61
PRINT 52tCBSCKltK=ltNCS)
READ 52t<AS<KltK=ltNCS>
PRINT 62
PRINT 52tCASCKltK=l,NCS)
IF CI QH-1 > 8, 8, 9
8 READ 52tCHUCKltK=ltNU>
PRINT 63
PRINT 52tCHU<KltK=ltNUl
*** CORRECT SPECIFIED STAGES TO MEAN SEA LEVEL DATUM
DO 12 K=ltNU
12 HU<K>=HU<K>+GZ
GO TO 10
9 READ 52, (QUCKltK=ltNUl
PRINT 58
B-3
 PRINT 52' (QU(Kl,K=l,NU>
. *** INPUT FRICTION COEFFICIENTS AND EFFECTIVE BOTTOM SLOPE
10 READ 53, CMLJ,CML2,CMU2,CMHL1,CMHL2,CMHU2,SO
PRINT 59,CML1,CML2,CMU2,CMHL1,CMHL2,CMHU2,SO
*** TYPICAL FLOOD DATA FOR COMPUTING CONSTANT <FR)
READ 54, TP,QMAX,QMIN,HMAX,HMIN
PRINT 67,TP,QMAX,QMIN,HMAX,HMIN
*** CONVERT DT TO SECONDS
DTS=DT*3600.
*** COMPUTE HOUR ASSOCIATED WITH POINT VALUES OF SPECIFIED HYDROGRAPH.
Tl<l>=O
DO 11 K=2,NU
11 Tl<K>=Tl<K-1)+DTHU
*** COMPUTE FR CONSTANT
HMAX=HMAX+GZ
HMIN=HMIN+GZ
HA=0.5*<HMAX+HMIN>
CALL SECT<HA,A'B'R,DB,DR>
FR=56200.*(0MAX+QMIN>*TP*SO/A/(HMAX-HMIN)
PRINT 56, FR
CONR=2e/3e*SO/FR/FR
KT=l
TT=O.
IF<IQH-1> 30,30,31
*** COMPUTE INITIAL DISCHARGE
30 HP=HU <1)
CALL SECT<HP,A,s,R,DBtDR>
CALL FRICT<HP,CM,DCM>
QP= le486/CM*SORT<SO>*A*R
PRINT 57, QP
GO TO 14
31 Yl=(QU(l)*CML1/(J.486*SQRT<SO)*BS<l>) >**<3.15.)
DO 32 K=l,20
CALL SECT<Yl'A'B'R~DB,DRl
CALL FRICT<Yl,CMtDCM1
F=QU(ll-1.4~6*SORT<SO)/CM*A*R
DF=-1.486*SORT<SO>*<B*R/CM+A*DR/CM-A*R*DCM/CM/CM)
Y=Yl-F/DF
IF<ARS<Y-Yll-EPHl 33,32,32
32 Yl=Y
33 HP=Y
QP=QU(l)
14 QF=QP
HF=HP
DHN=Oe
PRINT 68
GO TO 43
15 TT=TT+DT
*** INTERPOLATE AT TIME TT FROM SPECIFIED HYDROGRAPH
DO 16 K=f,NU
IF <TT-T1{K)) 17,16,16
16 CONTINUE
GO TO 26
17 KK=K-1
B-4
 IFCIQH-1) 40,40t41
*** INTERPOLATlON OF STAGE HYDROGRAPH
40 HF=HU<KK>+<)T-Tl<KKll/DTHU*<HUCK>-HU<KK))
CALL QSOLVE-<HFtHP,QP,EPQ,SO,CONR,DTStDO,GtOF)
GO ~0 43
*** INTERPOLATION OF DISCHARGE HYDROGRAPH
41 OF=QUCKK>+<TT-Tl<KK)l/DTHU*(QUCK>-OUCKK))
CALL HSOLVE (QF,QP,HPtEPH,SO,CONRtDTStDH,GtHF>
43 DQ=QF-QP
DH=HF-HP
*** COMPUTE NORMAL DISCHARGE
CALL SECT CHFtAtBtRtDBtDR>
QN=1.486/CM*SORTCSO)*A*R
DQN=QF-QN
*** CO~PUTE NORMAL STAGE
HNK=HF+DHN/2.
DO 22 K=lt20
CALL SECT<HNKtAtBtRtDBtDR>
CALL FRICT <HNK,CM,DCM>
F=QF-1.486*SQRTCS0l*A*R/CM
DF=-1.486*SQRTCSO>*<A*DR/CM+R*R/CM-A*R*DCM/CM/CM>
HNKK=HNK-F/DF
IFCABS<HNKK-HNK>-EPHl23t22t22
?2 HNK=HNKK
23 HN=HNKK
*** COMPUTE DIFFERENCES IN NORMAL AND DYNAMIC STAGES
DHN=HF-HN
*** CHECK TO SEE IF COMPUTATIONS ARE TO BE PRINTED
IF<TT-Tl<KT>> 25t24t24
24 KT=KT+l
*** PRINT COMPUTATIONS
PRINT 69,TT,HF,QF,QN,DQN,HN,DHN
25 QP=QF
HP=HF
GO TO 15
26 CONTINUE
STOP
50 FORMAT<5Xt*IQH=*I2>
51 FORMATC2Il0t3Fl0.2)
52 FORMAT<8Fl0e2l
53 FORMAT<3F10.5t3Fl0.2tF10.6)
54 FOPMATC5F10.2>
55 FORMAT C5Xt* NO. OF HYDROGRAPH PTS=*;I5t5Xt
1* NO. OF CROSS SECTION PTS=*ti3//5Xt*GAGE DATUM ELEVATIONCFT>=*t
2FA.2//t5Xt* DELTA TIME INCREMENT<HRS>=*tF6elt5X,
3* RESOLUTION OF SPECIFIED HYDROGRAPH<HRS>=*tF6el)
56 FORMATC5Xt*FR CONSTANT=*tF10.2)
~7 FORMAT<5Xt*INITIAL STEADY DISCHARGE<CFS>=*tFlOeO)
59 FORMAT<2X*<CML1) IS N VALUE AT LOWEST STAGE OF LOWER RANGE OF STAG
1E=*tFl0e5ti2Xt*CCML2> IS N VALUE AT HIGHEST STAGE OF LOWER RANGE
20F STAGE =*tFl0.5,/2Xt*CCMU2) IS N VALUE AT HIGHEST STAGE OF UPPER
3 RANGE OF STAGE =*tF10.5t/2Xt*CCMHLll IS STAGE ASSOCIATED WITH CML
41 =*tFl0e2t/2Xt*CCMHL2) IS STAGE ASSOCIATED WITH CML2 =*tFlOe2t/2X
5t*CCMHU2) IS STAGE ASSOCIATED WITH CMU2 =*tF10.2t/2Xt*EFFECTIVE B
 60TTOM SLOPE (50) =*tF12.8)
60 FORMATC25Xt* CROSS-SECTION FLEVATIONSCFT ABOVE MSL>*>
61 FORMATC25Xt* CROSS-SECTION WIDTHSCFT>*>
62 FORMATC25Xt* CROSS-SECTION AREAS (SQ. FT.>*>
58 FORMATC25Xt*SPECIFIED DISCHARGE HYDROGRAPHCCFS>*>
63 FORMATC25Xt* SPECIFIED STAGE HYDROGRAPHCFT>*>
67 FORMATC5Xt* TIME TO PEAK<DAYS>=*tF6.1t5Xt* MAX. ANNUAL DISCHARGE<C
1FS>=*tFlO.Ot/lt5X,*MIN. DISCHARGECCFS>=*tFlO.Ot5Xt* MAX. ANNUAL
2STAGE<FT>=*tF8e2t5Xt* MIN STAGE<FT>=*tf8.2)
68 FORMATClXt*TIMECHRSl*tlXt*STAGE<FT MSL>*tlXt*DISCHARGE<CFS>*tlXt*N
lORMAL DISCHCCFS> *tlXt*DYNAMIC EFFEtTCCFSl*tlX,*NORMAL STAGE<FT>*'
21Xt*DYNAMIC EFFECT<FT>*>
69 FORMATCF7.lt7XtF6.2,7XtFlOe0t8X,FlO.Ot8X.,FlO.OtlOX,F6e2tl3XtF6.2)
END
SUBROUTI,NE QSOLVE CHF,HP,QP,EPQ,SOtCONR,DTS,DQ,G,QF)
*** THIS SUBROUTINF LJSES NEWTON ITERATION TO SOLVE EQC15> FOR THE
*** UNKNOWN DISCHARGE WHEN THE STAGE HYDROGRAPH IS GIVEN.
*** COMPUTE CONSTANTS FOR EQC15)
CALL SECTCHP,AtBtRtDBtDR>
FL3=SO+CONR+QP/G/A/DTS
CALL SECT CHFtAtBtRtDBtDR>
CALL FRICT CHFtCMtDCM>
FK=5.13.-2.13.*A*DB/B/B
FL2=1.486*A*R/CM
DHS=<HF-HPl/DTS
FL4=A*DHS/FK
FL5=<1.-1.1FK>*B*DHS/G/A/A-le/G/A/DTS
FL6=-CONR/G*B/A/A/A
*** COMPUTE STARTING VALUE FOR ITERATIVE SOLUTION OF EQC15)
QK=QP+DQ/2.
*** SOLVE EQC15) BY NEWTON ITERATION
DO 20 K=lt20.
FLO=FL3+FL4/QK+FL5*0K+FL6*GK*QK
FL1=-FL4/QK/QK+FL5+2e*FL6*QK
F=QK-FL2*SGRTCFL0)
DF=1.~0.5*FL2*FL1/SQRTCFL0)
. QKK=QK-F/DF
IFCABSCQKK-QK>-EPQ) 2lt20t20
20 QK=QKK
21 QF=QK
RETURN
END
SUBROUTINE HSOLVE CQF,QP,HP,EPH,SOtCONR,DTStDH,G,HF>
*** THIS SUBROUTINE USES NEWTON ITERATION TO SOLVE EQ(15> FOR T~E
*** UNKNOWN STAGE WHEN THE DISCHARGE HYDROGRAPH IS GIVEN.
*** COMPUTES CONSTANTS FOR fQ(15>
CALL SECTCHP,AtBtRtDBtDR>
FJ2=SO+CONR+QP/G/A/DTS
FK=5.13.-2.13.*A/BIB*DB
*** COMPUTE STARTING VALUE FOR ITERATIVE SOLUTION OF EQC15>
HK=HP+DH/2e
*** SOLVF EQC15) RY NEWTON ITERATION
DO 20 K=1t20
CALL SECTCHK,AtBtRtDBtDR>
B-6
 CALL FRICTCHKtCM,DCM>
DH=HK-HP
FJ3=le/QF/FK/DTS
FJ4=Cl.-l.IFK>*GF/G/DTS
FJ5=-QF/G/DTS
FJ6=-CONR*GF*QF/G
FJO=FJ2+CFJ3*A+FJ4*B/A/Al*DH+FJ5/A+FJ6*B/A/A/A
FJ1=FJ3*A+FJ4*B/A/A+DH*<FJ3*B+FJ4/A/A*<DB-2.*B*B/A))-FJ5*B/A/A+FJ6
l/A/A/A*CDB-3.*B*B/Al
F=QF-le486*A*RICM*SQRTCFJO>
DF=-1e486*CSQRTCFJO)*CA/CM*DR+R*B/CM-A*R*DCM/CM/CM>+0.5*A*R/CM*FJ1
1/SQRTCFJO))
HKK=HK-F/DF
IF CABSCHKK-HK>-EPH)21,20t20
20 HK=HKK
21 HF=HKK
RETURN
ENO
SUBROUTINE SECT CY,A,BtR,DB,DR>
COMMON IA21 HS(lO),BSC10J,AS<lOJ,NCS
*** THIS SUBROUTINE COMPUTES THE GEOMETRICAL PROPERTIES OF THE
*** CROSS-SECTION AT A SPECIFIED WATER SURFACE ELEVATION.
DO 10 K=2,NCS
IF <Y-HSCKJ)5,5tl0
5 KT=K
GO TO 15
10 CONTINUE
15 KL=KT-1
DB=<BSCKTJ-~SCKL)J/CHSCKT>-HSCKLJ)
B=BSCKLl+DB*CY-HS<KLJ)
A=AS<KLJ+(AS<KTJ-ASCKL) )/(HS<KTJ-HSCKLJ J*CY-HSCKL))
R=CA/Bl**C2./3.)
DR=2.13.*R*CB/A-D8/8)
RETURN
END
SUBROUTINE FRICTCY,CM,DCM>
COMMON /All CMLl,CML2,CMU2,CMHLltCMHL2,CMHU2
*** THIS SUBROUTINE COMPUTES MANNINGS N COEFFICIENT FOR A
*** SPECIFIED WATER SURFACE ELEVATION.
CMU1=CML2
CMHU1=CMHL2
IF CY-CMHL2ll0,10,12
*** LOWER RANGE OF STAGE
10 DCM=<CML2-CML1)/CCMHL2-CMHL1l
CM=CMLl+DCM*<Y-CMHLll
GO TO 20
*** UPPER RANGE OF STAGE
12 DCM=<CMU2-CMU1)/CCMHU2-CMHU1)
CM=CMU1+DCM*(Y-CMHU1>
20 RETURN
END

B-7
 IOH= I
NOo Of HYOROGRAPH PTSz 64 NO. Of CROSS SECTION PTS 4
GAGE DATUM ELEVATION(fTl 2 3.49
DELTA TIME INCREMENT(HRS1= 3.0 RESOLUTION Of SPECIFIED HYDROGRAPHCHRSl 24.0
CROSS-SECTION ELEVATIONSCFT ABOVE MSLl
16.00 34.00 41.20 48.00
CROSS-SECTION WlDTHSCFTl
3000.00 3540.00 3630.00 3690~00
CROSS-SECTION AREAS CSQ. FT.l
72500.00 134000.00 164000.00 200000.00
SPECIFIED STAGE HYDROGRAPH(FTI
18.29 18.59 19.56 21.27 23.22 25.11 26.78 28.02
29.01 29.84 31.01 3?.54 33.79 34.51 35.74 36.63
37.32 3B.02 38.56 39.00 39.54 40.10 40.67 41.10
41.40 41.68 41.86 42.11 42.40 42.50 42.80 42.74
42.38 41.89 41.29 40.58 39.82 38.81 37.70 36.53
35.11 33.88 32.97 32.07 31.10 30.38 29.82 29.30
28.77 28.26 27.75 27.28 26.90 26.81 26.64 26.59
26.20 25.80 25.45 25.02 25.11 24.72 24.02 23.99
CCML11 IS N VALUE AT LOWEST STAGE Of LOWER RANGE Of STAGE .01590
(CML2l IS N VALUE AT HIGHEST STAGE Of LOWER RANGE Of STAGE .01392
CCMU21 IS N VALUE AT HIGHEST STAGE Of UPPER RANGE Of STAGE .01392
CCMHL1l IS STAGE ASSOCIATED WITH CML1 z 5.00
CCMHL2l IS STAGE ASSOCIATED WITH CML2 = 50.00
CCMHU2) IS STAGE ASSOCIATED WITH CMU2 50.00
EffECTIVE BOTTOM SLOPE CSOI = .00001430
TIME TO PEAKCDAYSI= 30.0 MAX. ANNUAL DISCHARGECCFSI 1064000
MIN. DISCHARGECCfS): 319000 MAX. ANNUAL STAGECFT) 42o74 HIN STAGECFT>= 18.29
FR CONSTANT= 10.18
INITIAL STEADY OISCHARGECCfSl= 323237
TIMECHRSI STAGE!FT MSLI DISCHARGECCFS) NORMAL DISCH(CfSl DYNAMIC EffECT!CfSl NORMAL STAGE ( fT) OYNA~IC EffECT(fTl
o.o 21.78 323237 323237 0 21.78 o.oo
24.0 22.08 337255 329293 7962 22.52 -.44
48.0 23.05 371583 348613 22970 24.28 -1.23
72.0 24.76 423051 383157 39894 26.82 -2.06
96.0 26.71 471073 423371 47702 29.10 -2.39
120.0 28.60 512768 463363 49405 31.01 -2.41
144.0 30.27 546285 499490 46795 32.51 -2.24
168.0 31.51 563946 526434 37512 33.28 -J-77
192.0 32.50 580051 548379 31672 33.98 -1.48
216.0 33.33 594817 566860 27957 34.47 -}.14
240.0 34.50 634415 597768 36648 35.74 -1.24
264.0 36.03 695029 646434 48595 37.62 -Io59
288.0 37.28 728821 686503 42318 38.63 -}.35
312.0 38.00 735959 709323 26636 38.85 -.85
336.0 39.23 795864 751720 44144 40.59 -}.36
360.0 40.12 815691 781637 34054 41.16 -1.04
384.0 40.81 833019 805060 27959 41.55 -.74
408.0 41.51 861131 832588 28543 42.16 -.65
432. 0 42. 05 880282 857069 23213 42.58 -.53
456.0 42.49 897078 877289 19788 42.94 -.45
480.0 43.03 926800 902825 23975 43.56 -.53
504.0 43.59 954667 929480 25186 44.14 -.55
528.0 44.16 982978 956941 26038 44.73 -.57
552. 0 44.59 998337 9775.62 20775 45.04 -.45
576.0 44.89 1007599 991974 15625 45.23 -~34
600.0 45.17 1020669 1005743 14926 45.49 -.32
624.0 45.35 1025197 1Q14426 10771 45.58 -.23
648.0 45.60 1040906 1027052 13854 45.89 -.29
672.0 45.89 1057379 1041684 15695 46.22 -.33
696.0 45.99 1053738 1046235 7503 46.15 -.16
720.0 46.29 1078225 1061904 16321 46.63 -.34
744.0 46.23 1058347 1057919 428 46.24 -.01
768.0 45.87 1025673 1038963 -13290 45.59 .28
792.0 45.38 994973 1014161 -19i88 44.97 o41
816.0 44. 78 9.60255 984291 -24036 44.26 .52
840.0 44.07 920788 949520 -28732 43.44 .63
864.0 43.31 882614 913100 -30486 42.63 .68
888.0 42.30 823985 865211 -41226 41.35 .95
912.0 41.19 769111 813793 -44681 39.82 }o37
936.0 40.02 725974 772823 -46849 38.55 lo47
960.0 38.60 666914 724001 -57087 36.75 1o85
984.0 37.37 637330 684064 -46734 35.83 lo54
1008.0 36.46 623426 655551 -32125 35.39 1o07
1032.0 35.56 596052 627045 -30993 34.51 }.05
1056.0 34.59 563779 596497 -32718 33.27 1o32
1080.0 33.87 551059 575686 -24627 32.72 1o15
1104.0 33.31 544904 563256 -18353 32.44 .87
1128.0 32.79 534895 551537 -16642 32.00 .79
1152.0 32.26 522738 539541 -16802 31.46 .so
1176.0 31.75 512287 528162 -15874 30.99 .76
1200.0 31.24 501137 516816 15678 30.48 .76
1224.0 30.77 492438 506533 -14095 30.08 o69
1248.o 30.39 487519 498401 -10882 29.86 .53
1272.0 30.30 495700 497065 -1365 30.23 .07
1296.0 30.13 4R9268 493178 -3910 29.94 .19
1320.0
1344.0
30.08
29.69
492234
472112
492337 -103 30.08 .oo
483136 -11024 29.15 .54
1368.0 29.29 463237 474479 -11242 28.73 .56
1392.0 28.94 457558 467071 -9513 28.47 .47
1416.0 28.51 445748 457736 -11988 27.91 o60
1440.0 28.60 464668 460653 4015 28.80 -.20
1464.0 28.21 440852 451458 -10606 27.68 .53
1488.0 27.51 415605 436120 -20515 26.46 leOS
B-8