TR 8
TR 8
TR 8
SA TISH CFIANDRA
DIRECTOR
STUDY GROUP
S K JAIN
ROORKEE
I 986-87
CONTENTS
PAGE
Abstract
List of Figures ii
INTRODUCTION 1
1.0
1.1 Importance of Water Balance of Reservoirs 1
Diagramatic representation of 7
1
components of water balance of
a reservoir
level in a reservoir
Computation of fluctuation of water 40
1
considered.
3
Spillway discharge tables showing discharge at different
reservoir levels for various gate openings(if spillway
is gated),
Discharge through undersluices at various reservoir
levels,
needed:
Details of turbines
power generated
after-bay levels
efficiency of turbines and generators
head loss in penstocks and turbines
can be expressed as :
where,
IS Surface water inflow into the reservoir,
computation
6 = error term
The water enters in the reservoir through surface inflow
and direct precipitation; the water that leaves reservoir
comprises of releases through outlets and spillways, evaporation
and losses due to seepage. The reservoir storage increases if
the inflow exceeds outflow and decreases if the outflow exceeds
inflow. All the components of water balance equation should be
independently estimated. The term6 in the above equation (1)
represents the net effect of errors involved in the estimation
5
of different components. In practice it is quite likely that
of the reservoir,
Q. = temporary water losses by the ice left on the
FIG.1 Diagramatic representation of components of water
balance of a reservoir (adapted from Ferguson & Znamensky)
7
shores after the fall of level of the reservoir.
where,
Aw = water surface area of the reservoir in km 2
611 = error of mean level estimation(m)
I s = discharge into the reservoir
8
The site at which streamflow measurements are carried out
is called a gauging site. The location of the gauging site
may be little away from the reservoir to avoid the back-water
effect and in such cases, the contribution of the area lying
between the gauging site and the rim of the reservoir has to
be considered to arrive at the correct figure. Most commonly
the variables measured at the gauging site are river stage
and discharge. If only river state is measured at the gauging
site, discharge can be estimated using the rating curve at
the site. A large number of methods are available to determine
streamflow at a particular site. These methods include the
velocity area method, slope area method moving boat method,
dilution methods and ultrasonic method. These methods are
described in great detail by Herschy(1978). Selection of a
particular method largely depends upon the site and flow
conditions, equipment available and accuracy requirements.
9
sub-basin can be estimated using the unit discharges from the
gauged subbasins. The procedure is repeated for all such sub-
basins which lack measurements and discharge in the reservoir
and the summation of discharges of all such subbasins gives the
inflow to the reservoir. In the situations where no sub-basin
with adequate measuring stations is available, it may still be
possible to use this method if a gauged sub-basin with similar
hydromet and topographic characteristics is available in a
neighbouring watershed. The indices which are important in
comparing the subbasins are drainage density,mean slope soil
type and land use. The formulae which could be used to determine
water
For the long term water balance computations, the storage
10
change is not very important parameter and in mean water balance
computations, it may as well be assumed zero. In case the
measurements of P and Eu are not available, it may be possible
u
to use this method by correlating with physiographic character-
istics such as slope,elevation etc. A number of studies have
been cited in Ferguson and Znamensky(1981) where this has been
successfully achieved.
11
between precipitation and evaporation. The normal annual
precipitation may be obtained from climatic maps and the normal
annual evaporation may be computed by heat balance or by
empirical formulae. It may be mentioned that for small basins,
this method may give quite incorrect results.
12
ending precipitation are also available (Linsley,et al.1975).
The measurements of precipitation are expressed in terms of
vertical depth of water which would accumulate on a level
surface if the precipitation remained where it fell. The gauges
which are commonly used to measure rainfall include weighing-
type, Siphon-type and Tipping bucket raingauge. Of late, radar
is also being increasingly used for estimation of precipitation.
orifice.
13
collection of precipitation data with due consideration given
to the needs as well as the economy. Typically purposes
behind setting up a network are management of reservoirs for
irrigation water supply and hydro-electric power generation,
planning of water resources systems, agricultural planning
and meteorology. The optimum density of network which is the
number of gauges per unit area is determined based upon the
purposes of measuring data. A relatively thinner network may
be required for estimation of seasonal figures while a dense
network will be needed for flood forecasting purposes. The
World Meteorological Organisation,WM0(1974) has provided
detailed recommendations for the optimum network for various
climatic zones and type of terrain for general hydrologic
values may be missing for one or more periods. One method which is
14
popularly used to fill short data gaps is normal ratio method.
In this method, the precipitation at the station x is estimated
from the observations at three stations which are as close to
station x as possible. The precipitation at station x is
estimated by
N N
P
x
- 1 14x
P + X + X—P (6)
N A NB B N C
A
where
N represents the normal annual precipitation, and P represents
precipitation. Inconsistency in the observed data may be
present due to change in the location of the station. Checks
are available to test this type of inconsistency. A graphical
method, called double mass analysis is in use since a long
time. In this method, the consistency of the record at a station
is tested by comparing its accumulated annual or seasonal
precipitation with the concurrent accummulated values of mean
precipitation for a group of surrounding stations. A change
in slope of the line indicates a change in the precipitation/
regime at the base station. This type of change can not be
attributed to meteorological causes because in that case, all
the stations would be similarly affected.
15
precipitation is the Thiessen Polygon method. The method is
lb
3.3.4 Polynomial Interpolation
= E i=1,2....n (8)
1 x=1af(xiY
kk'i)
where m is the number of monomials and m
Here the aim is to determine parameters ak ,k=1....m such that
the sum of square of errors is minimized,i.e.,
Min Z = E 2
(pi- pt) (9)
1=1
yields.
m a. n n
E 1 /
fk(x.,3y.)f.(x. y.)= E .
p.f (x1.,Y) (10)
i=1 j=1 3 1 3' j . i
1-1 1 k
Solution of the system of equations(10) gives
n
ak = J E, Tki pi k = 1, m (11)
17
where
T. = H .f.(x.,y.) k=1...m
kj 1=1 Ia.1 3 i=1...m
(8) is written as
pt
1 = k=1 akf k(x.y
,.) j = 1,2.... (14)
3 3
which upon solving yields
ak = Bki pi (15)
w(d) = 1/(d+1)
w(d) = 1/d 2
w(d) =
w(d) = e-ad
where d represents distance and a is a constant.
19
n (18)
z a.d..
P- = i.1 j = 1....n
J 1 31
Solvillgecluati°11gOthe"efficielitsa.are
1 determined
a. =E c..p. (19)
1 j.1 13 j
P* E c p di (20)
i=i j=1 1J J o
defining = Var
(23)
iCovIo) = Cov(po pj) j = 1....n
il w
Making use of homogeneity in the variances, the covariance
20
where p( ) is spatial correlation coefficient. A spatial
weights wi ,i = 1....n.
The estimation of the variable will be unbiased if the
w. = 1 (26)
i=1
Incorporating equation(26) into equation(22) using
Lagrange multiplier,
n n
2 2 w Cov(p o p-) 1:41 1 E 1 ww 1 Cov(n 11)
a = o - 2 1 i
11
- wi p(d ii) + X .),j=1...n
= P (doj (25a)
i=1
ki (29b)
i-1.1 wi ' 1
Upon solving (n+1) simultaneous equations(eqns.293 and 29b) the
21
. The choices of correlation assuming homogeneity and isotropicity
are :
The reciporocal model p(d) = q/(1+d/c0),
The square-root modelp(d) =1/1/ l+d/co, and
The exponential modelp (d) =
quantity p!,
s which is a linear function of variables. Three
types of problems may arise here:
To estimate the value of the variable at a point,
To estimate the value of the variable over a mesh of
given area centered at a known point, and
To estimate the value of variable over a specified
domain.
The first type of problems are called point Kriging;
or under estimation,
optimal, i.e., the variance between the observed
+ 2 X wi -1] (30)
i1
=
The theory of kriging assumes the increments of the
variables to follow the weak stationarity of second order.
Under this assumption, a random function is said to be stationary
if the first two moments of its joint probability distribution
at k arbitrary points are invariant under simultaneous trans
lation of all the points. Now the semivariogram is defined
as
= 4 Var[p.1
- p.]
3.
13
(31)
= a 2 - Cov(d..) i,j = 1,....n
13
where y(d.. ) is the semivariogram which is function of the
13
distance between i and j points. Making substitution in (30)
yields
23
E w.y(d..) + = Y (d03 ) = 2,....n
1=1 13 j
2
1=1 w.y(d
ae ) +
1 oi
(34)
As mentioned above, the knowledge of variogram is required
d> a
Exponential 1(d) = g {(1-exp(-fdll /a).]
Gaussian Y(d) = g[(1-exp(- idt 2 /a 2j
3.4 Evaporation
24
process. The most important factor in the process is radiation
followed by wind speed and vapour pressure of the air overlying
the surface. The amount of evaporation also varies with
latitude, season, time of day and condition of sky. It is
difficult to categorically express the relative effect of the
controlling meteorological factors, if radiation exchange and
.25
on or above the ground surface experience little higher evaporation since extra
26
•
and it should be refilled when the depth of water falls to
18 cm. The water level can be measured using a hook gauge.
Rn - R h - R r + R V = 0 (35)
where
Rn Net radiation absorbed by the reservoir,
27
R = energy stored in the reservoir
r
Rv = net energy content of inflowing and outflowing water
(36)
E = (1(11+R), - lc)/ "v(l+R)
where
E = evaporation in centimeters
n = density of water.
The Bowen ratio can be computed by the following equation
wnere
p = atmospheric pressure,
Rn = Rs - Rr + Ra - Rar - Ro (38)
28
where
Rs = sun and sky short wave radiation incident
upon the water surface
sections.
31
3.5 Estimation of reservoir Outflow
Q. = C 1.5
sp sp Cq b1/2gh
sp sp (46)
where
C
sp = submergence coefficient,
C = discharge coefficient for spillway,
= width of the spillway
sp
32
Where C is the coefficient of the weir, L is the clear
low le/el outlets for releasing water when the reservoir water
Q Or = or e A i2qh. (48)
33
through turbines. Each of these is discussed Delow.
interval.
34
The term n in equation(5U) represents combined efficiency
individual efficiencies.
turbines are generally on the lower side and the error in daily
AS + AS +AS (52)
TM bs
where,
AS = change in storage in reservoir.
w
A SYM = change in channel storage of all those streams
which directly debouch in the reservoir between the
gauging site which lies just upstream of reservoir
back in summer.
36
High level
— -
5, •
It •h
h>2
Lii .
--ze.
."-*: "Ih. • d `,
d
>1/4. <
37
Out of these four, the first component is most
.38
reservoir influences the stage of the river unless the storage
in this zone is less than 5-10% of the total accumulation.
When wind blows over a reservoir, it applies shear stress
on the water surface and thereby it tries to carry water along
with it. This leads to a redistribution of water in reservoir,
there will be greater storage in the down wind direction and
xis the angle between wind direction and the line joining
39
(a)
40
reservoir crosses the equilibrium axis perpendicular to a given
frequency.
41
methods are available for this purpose which could be used
depending upon the slope of the water surface and degree of
knowledge of morphometric characteristics of the reservoir.
If the water surface of the entire reservoir is more-or-less
horizontal, the mean water level can be used to determine
the storage. Otherwise volume may have to be determined
for each subarea but this requires individual elevation-
storage curves for each subarea.
(1981).
43
4.0 ERRORS IN WATER BALANCE COMPUTATION
j„ 2 ÷ 62 ... (55)
6<
If this criterion is not satisfied then it is required
to reevaluate the estimation procedure and measurements of
individual components. Since the magnitude of different
comporents vary widely, percentage errors in them will also
44
upon the time period of computation. As this period increases
the magnitude of error in various terms which represent inflow
and outflow to and from the reservoir also increases. However,
the error in the term representing change of storage tends to
available.
45
5.0 CONCLUSIONS
46
REFERENCES
Baumgartner,A., and E.Reichel(1975), World Water Balance,
Elsevier Publishing Company,Amsterdam.
Ferguson,H.L., and V.A.Znamensky,' Methods of Computation
of the water balance of large lakes and reservoirs',
Studies & Reports in Hydrology461,Unesco 81.
47