Runoff

Runoff
Draining or flowing off of precipitation from a catchment area
through a surface channel

The output from the catchment in a given unit of time

• For a given precipitation,

– evapotranspiration,
– initial loss,
– infiltration and
– detention-storage requirements

have to be first satisfied before the commencement of

Runoff
• Overland flow - excess precipitation
moves over the land surfaces to reach
smaller channels

• Surface runoff - Flows from several

small channels join bigger channels and
flows from these in turn combine to form a
larger stream, and so on, reaches the
catchment outlet
 Interflow/seepage/subsurface/storm
flow or Quick return flow

A part of the precipitation that infiltered moves laterally

through upper crusts of the soil and returns to the
surface at some location away from the point of entry
into the soil
– Depends on geological conditions of the
catchment (physiographic feature)

• A fairly pervious soil overlying a hard impermeable surface

is conducive to large interflows

• Depending upon time delay between the infiltration and the

outflow, the interflow is classified
– Prompt Interflow
– Delayed interflow
 Groundwater Runoff/flow
• Another route for the infiltered water is to
undergo deep percolation and reach the
groundwater storage in the soil

• The groundwater follows

– a complicated and long path of travel, and ultimately reaches the
surface
– large time lag, being of the order of months and years

• provides the dry-weather flow in perennial

streams
 Classification of Runoff
• Based on the time delay between the precipitation
and the runoff

– Direct Runoff: part of runoff which enters the stream

immediately after the precipitation
• Surface runoff
• Prompt interflow
• Precipitation on the channel surface

– Base Flow: The delayed flow that reaches a stream

essentially as groundwater flow
 Streamflow
• A stream can be defined as a flow channel into which
surface runoff from a specified basin drains.

• Stream-flow represent the runoff phase of the hydrologic

cycle.

• Stream-flow is amenable to fairly accurate measurement

in contrast to evaporation, precipitation,
evapotranspiration etc.

• There is considerable exchange of water between a

stream and underground water.
 Stream-flow measurement
techniques (Hydrometry)
• Direct measurement
– Area-velocity method
– Dilution technique
– Electromagnetic method
– Ultrasonic method

• Indirect measurement
– Hydraulic structures, such as weirs, flumes and gated
structures
– Slope area method
 General
• Stream-flow is measured in units of discharge (m3/s) at a
specified time and constitute historical data

• Continuous measurement of discharge is very difficult.

• Direct measurement of discharge is very time consuming and

costly.

• Two step procedure is followed

– Discharge in a given stream is related to elevation of the
water surface (stage) through a series of careful
measurements.
– Stage of the stream is observed routinely and discharge is
estimated by stage discharge relationship.

• Have you seen a scale in river bridge piers???

Stage discharge relationship

stage discharge relationship curve

 Measurement of stage
• It is defined as the water surface elevation
measured above a datum (can be MSL or any
arbitrary datum).

• Manual stage measurement

– Staff gauge
– Wire gauge

• Automatic stage measurement

– Float gauge recorder
– Bubble gauge
 Staff gauges

Vertical Staff Gauge Sectional Staff Gauge

Float gauge recorder

Float Gauge recorders

 Stage data presentation
• Presented in the form of plot
of stage against
chronological time, known
as stage hydrograph

• Uses
– Determination of stream
discharge
– Flood warning and flood
protection works
– Reliable long term data to
peak floods can be analyzed
statistically to estimate design
peak river stages for use in the
design of the hydraulic
structures, such as bridges,
weirs etc.
 Stage Discharge relation ship (rating
curve)
• First a stage discharge relationship has to be
developed to use the stage data

• Thus the main aim of measuring stream-flow is to

establish this rating curve

• Let G be the stage and Q be the discharge. Once the

relationship is established, later just measure only the
stage and read the discharge from rating curve.

• First plot the discharges against the stages measured.

– If the relationship is permanent and does not vary with respect
to time it is called permanent control else it is called shifting
control
Stage discharge curve
 Relationship (for Permanent Control)

Q=Cr(G-a)
Where Q –discharge, G - is the stage, a – is a constant which represent
the gauge reading corresponding to zero discharge (normally
estimated by extrapolating in the plot, since the measurement is
almost very difficult)
Cr and  are rating curve constants

The best value are estimated either by method of least square or by

graphical method

logQ= log(G-a)+logCr
Plot the measured values and find the constants
The above equation is called as rating equation.
Step 1
Solve? Develop a stage discharge relationship for the
measured data. Find stage for a discharge of 2600
m3/s.
Stage Discharge Step 1
(m) (m3/s) ln(G-a) ln Q 26.00

21.95 100 -0.654 4.606

22.45 220 0.020 5.394 25.00

22.8 295 0.315 5.688

23 400 0.451 5.992 24.00

Stage (m)
23.4 490 0.678 6.195
23.57 500 0.761 6.215
23.00
23.65 640 0.798 6.462
24.05 780 0.963 6.660
22.00
24.55 1010 1.138 6.918
24.85 1220 1.230 7.107
21.00
25.4 1300 1.379 7.171
0 200 400 600 800 1000 1200 1400 1600 1800
25.15 1420 1.314 7.259 Discharge (cu.m/s)

25.55 1550 1.416 7.347 But actual measured is 20.5 m

25.9 1760 1.498 7.474 only????????
From the graph the value of a (discharge for zero stag is 21.50 m)
 Contd…
1.50

(G-a) =
Fit Results

Fit 1: Linear, Y=B*X+A

Equation:
1.00 Y = 0.726826 * X + -3.89026

0.0205Q0.7268
Number of data points used = 14
Average X = 6.46345
Average Y = 0.807547
Regression sum of squares = 4.83243
Residual sum of squares = 0.0538442
Coef of determination, R-squared = 0.988981
0.50 Residual mean square, sigma-hat-sq'd = 0.00448702

Stage (m)
G=27.63 m for a discharge
of 2600 m3/s
0.00

(G-a) =
0.163Q0.466 -0.50

G=26.85 m for measured a

of 20.5 m -1.00

Else??
0 1 2 3 4 5 6 7 8 9 10
Discharge (cu.m/s)
 Contd…
• This method holds good if the flow is virgin flow

• Stream-flow should be purely from runoff only

• If there is reservoir release or

• There is a back water effect

• Then we need site specific relationship and the

available rating curve need to be either
extrapolated or modified accordingly
First step in measuring the discharge is
the Measurement of velocity
• Commonly measured by a mechanical device called current meter.

• It measures the velocity at a point in the flow.

• It consists of a rotating element which rotates due to reaction of the

stream current with an angular velocity proportional to the stream
velocity.

• Two main types of current meter

– Vertical axis meters
– Horizontal axis meters

• Approximate stream velocity can be determined by floats.

 Current meter
• It is so designed that its rotation speed varies linearly
with the stream velocity at the location of the
instrument.
v = a Ns + b
Where
v = stream velocity at the instrument location in m/s
Ns = revolutions per second of the meter
a, b = constants of meter

• Each instrument has a threshold velocity below which

above equation is not valid.

• No of revolutions are counted for a known interval of

time.
 Horizontal axis meters

Vertical axis meters

 Calibration of Current meter
• The relation between stream velocity and revolutions per seconds
of the current meter is called calibration equation.

• Calibration equation is unique to each instrument.

• It is determined by towing the instrument in a special tank.

• The towing tank is a long channel containing still water with

arrangements for moving a carriage longitudinally over its surface
at constant speed.

• The instrument to be calibrated is immersed to a specified depth.

• Average value of revolutions per seconds (Ns) are determined for

a predetermined constant speed (v). This is repeated over a
complete range of velocities and a best fit linear relationship is
formed.
Velocity measurement by Floats
• Floating objects on the surface of a stream when timed can yield the surface velocity.
(first used by ????????? Remember it)

S
vS 
where t
S = distance traveled in time t

• Useful for
– A small stream in floods
– Small stream with rapidly changing water surface
– Preliminary surveys

• Types
– Surface floats
– Canisters floats
– Rod floats
 Area velocity method
• It consist of the measurement of the area of cross
section of the river at a selected section and measuring
the velocity through it.
Q = AV

• Criteria for gauging site

– The stream should have a well-defined cross-section which does
not change in various seasons.
– It should be easily accessible all through the year.
– The site should be in a straight, stable reach.
– The gauging site should be free from backwater effects in the
channel.
 Contd…
• Cross section area is divided into a large no of cross subsections by
verticals.

• The velocity in these subsections are measured by current meter.

• Accuracy of discharge measurement increase with no of

subsections which also increase time, effort and expenditure .

• Guidelines for selection of subsections

– The segment width should not be greater than 1/15 to 1/20 of the width
of the river.
– The discharge in each segment should be less than 10% of the total
discharge.
– The difference of velocities in adjacent segments should not be more
than 20%.
Cross-section of a stream to measure the
discharge
Variation of stream velocity with
depth

Variation of stream velocity with depth

 Steps in finding the discharge after
measuring the area and velocity
• If qi is discharge in elemental area
• Width is given by (W i+W i+1)/2
• If di is depth at velocity measurement
• Area of element is ai= di* (W i+W i+1)/2
» for I = 2, 3, 4, …..n-2
• For segment 1 and N it is triangle
2   2 

  W  W2    W  W N 1  
 1 2    N 2  
a1  d1   aN  dN   
2W  2W N 
 1 
   
   
• Discharge in elemental area qi=ai*vi
• Total discharge is N
Q   a i vi
i 1
 Solve?
The data for a stream gauging station is as follows. The rating equation
of current meter is given as v  0.51N s  0.03 m / s
Ns is number of revolutions per second
Distance from left Depth (m) Number of revolution of a Duration of
edge (m) current meter kept at observation
Average Velocity Segmental
0.6 m depth
width (m) (m/s) discharge cu.m/s

0 0 0 0 0 - -
1 1.1 39 100 2 0.229 0.504
3 2 58 100 2 0.326 1.303

5 2.5 112 150 2 0.411 2.054

7 2 90 150 2 0.336 1.344

9 1.7 45 100 2 0.260 0.882

11 1 30 100 2 0.183 0.366

12 0 0 0 0 - -

6.453
 Moving-Boat Method
• In large rivers standard
current meter is very time
consuming and it is difficult to
keep the boat stationary.

• Propeller type current meter is

used which is free to move
about its vertical axis is used.

• If Boat velocity is Vb, Flow

velocity is Vf and Resultant
velocity is VR (meter align
itself in the direction of
resultant velocity)
vb = vR cos θ and vf =vR sin θ
Moving boat method
 Contd….
• If the time of transit between two verticals is Δ t, then the width
between the two verticals
W = vb Δ t

• The flow in the sub-area between two verticals i and i+1 where the
depths are yi and yi+1 respectively, by assuming the current meter
to measure the average velocity in the vertical, is

 Stream discharge

Q = Σ ΔQi
 Solve
? The following data has been obtained in a moving boat.
Determine the discharge in the stream. The sections are
spaced at a constant distance of 75 m apart.
Angle made by current meter
Resultant velocity with the direction of boat
Section (m/s) movement (degrees) Depth (m)
0 Right bank - 0
1 1.75 55 1.8
2 1.84 57 2.5
3 2.00 60 3.5
4 2.28 64 3.8
5 2.30 65 4.0
6 2.20 63 3.8
7 2.00 60 3.0
8 1.84 57 2.5
9 1.70 54 2.0
10 Left bank - 0.0

Discharge = 3458 cu.m/s

Dilution technique for stream flow
measurement
• It depends upon the continuity principle applied to a
tracer which is allowed to mix completely with the flow.
– For a constant injection method if
Co = small initial concentration of the tracer in the stream-flow
C1 = high tracer concentration at section 1 where tracer is added
C2 = tracer concentration at section 2 where tracer mixes
thoroughly with the fluid due to the turbulent mixing process
Qt=discharge of tracer and Q-is discharge in the stream

i.e.
 Solve
?A 500 g/l tracer was injected at a constant rate of
4l/s. At downstream length the concentration of
tracer was measured as 4 ppm. If initial tracer
concentration in the stream was zero. Estimate
the stream discharge?
1 gram/liter [g/L] = 1001.142303 part/million [ppm]
The parts per million (ppm, 10⁻⁶) is a pseudo unit that describes small
values of miscellaneous quantities, for example, mole fraction or
mass fraction. These fractions are dimensionless because they are
quantity-per-quantity measures. Therefore, they are simply
numbers. One part per million denotes one part per 1,000,000 parts
and a value of 1 × 10⁻⁶. This can be compared to one drop of a
liquid substance diluted into 50 litres of water. Other part-per units
are commonly used: ppb (parts-per-billion, 10⁻⁹), ppt (parts-per-
trillion, 10⁻¹²) and ppq (parts-per-quadrillion, 10⁻¹⁵).
 Discharge Q = 500 m3/s
 Tracer
• Ideal properties of tracer:
– It should not be absorbed by the sediment, channel boundary
and vegetation. It should not chemically react with any of the
above surfaces and also should not be lost by evaporation.
– It should be non-toxic.
– It should be capable of being detected in a distinctive manner
in small concentrations.
– It should not be very expensive.

• Types of tracer:
– Chemicals (common salt and sodium dichromate )
– Fluorescent dyes (Rhodamine-WT and Sulpho-Rhodamine B
Extra )
– Radioactive materials (such as Bromine-82, Sodium-24 and
Iodine-132) (radioactive tracers).
 Hydrograph
A plot of the discharge in a stream plotted against
time chronologically

10
12
13
14
15
16
17
18
19
20
21
22
23
24
25
27
28
29
30
31
32
33
34
35
36
37
38
39

42
43
44
45
46
47
48
49
50
51
52
53
54
56
57
58
59
60
61
62
63
64
65
66
67
68
69
71
72
73
74
75
76
77
78
79
80
11

26

40
41

55

70
1
2
3
4
5
6
7
8
9
0 12

Rainfall
5 Crust 10

Time Lag 10
Recession
8
Limb
Runoff (V/t)

15

20

Peak Discharge 4

25

Base flow line

Rising 2
30
Limb
35 0

Time
• Depending upon the unit of time involved

– Annual hydrographs: daily or weekly or 10 daily

mean flows over a year

– Monthly hydrographs: daily mean flows over a

month

– Seasonal hydrographs: discharge in a particular

season such, as the monsoon season or dry season;
and

– Flood hydrographs or hydrographs due to a storm

representing stream flow due to a storm over a
catchment
• Each of these types have particular
applications

– Annual and seasonal hydrographs

• Calculating the surface water potential of stream
• Reservoir studies
• Drought studies

– Flood hydrographs
• Analyzing stream characteristics associated with
floods
 Water Year
• In annual runoff studies it is advantageous to
consider a water year

– Time when the precipitation exceeds the average

evapotranspiration losses

– In India, June Ist - May 31st of the following calendar year

– Complete cycle of climatic changes is expected and

hence the water budget will have the least amount of
carry over
 Runoff Characteristics of
Streams
• Classification of stream based on
annual hydrographs of streams

– Perennial

– Intermittent

– Ephemeral
 Perennial Stream
which always carries some flow
• There is considerable amount of groundwater flow throughout the year
• Even during dry seasons the water table will be above the bed of the stream
 Intermittent Stream
• During the wet season the water table is above the stream bed and
there is a contribution of the base flow to the stream flow

• During dry seasons the water table drops to a level lower than that of the
stream bed and the stream dries up
 Ephemeral Stream
• Does not have any base-flow contribution

• The annual hydrograph, shows series of short-duration spikes marking flash

flows in response to storms

• Becomes dry soon after the end of the storm flow

Characteristics of Streams based on
ground water contribution
– Effluent streams

– Influent streams
Flow characteristics of a stream
depend upon
• Rainfall characteristics
– Magnitude intensity
– Distribution time
– Space and its variability

• Catchment characteristics
– Soil
– Vegetation
– slope, geology, shape
– Drainage density

• Climatic factors which influence

evapotranspiration.
 YIELD
(ANNUAL RUNOFF VOLUME)
The total quantity of water that can be expected from a
stream in a given period

• The calculation of yield is of fundamental importance in all

water-resources development studies

• Methods to estimate yield

– Empirical equations
– Correlation of stream flow and rainfall,
• Time series modeling, ANN, GP and MT models
– Watershed simulations
Block Box Models
Conceptual models
Physical models
 Types of Rainfall-Runoff models Models

• Empirical models: based on observational data and

calibrated input-output relationship without description of
individual processes

• Conceptual models: basic processes (evaporation, etc.) are

separated to some extent, but their algorithms are
essentially calibrated input-output relationships

• Physically based models: based on reliable relationships

between watershed characteristics and the parameters
 Empirical Equations

• Binnie's Percentages
• Barlow's Tables
• Strange's Tables
• Inglis and DeSouza Formula
• Khosla's Formula

SCS-CN technique
 Binnie's Percentages
• Sir Alexander Binnie measured the runoff from a
small catchment near Nagpur (Area of 16 km2 )
during 1869 and 1872

• Developed curves of cumulative runoff against

cumulative rainfall

• Established percentages of runoff from rainfall

• These percentages are beening used in Madhya

Pradesh and Vidarbha region of Maharashtra for
the estimation of yield
 Barlow's Tables
• Barlow, the first Chief Engineer of the Hydro-
Electric Survey of India (1915)

• Conducted study on small catchments (area~

130Km²)in Uttar Pradesh expressed runoff R as
R = Kb P

where Kb = runoff coefficient

– which depends upon
• type of catchment
• nature of monsoon rainfall.
 Barlow’s Runoff Coefficient Kb in Percentage

Class Description of catchment Values of Kb (percentage)

Season Season Season
1 2 3
A Flat, cultivated and absorbent soils 7 10 15
B Flat, partly cultivated and stiff soils 12 15 18
C Average catchment 16 20 32
D Hills and plains with little cultivation 28 35 60
E Very hilly, steep and hardly any 36 45 81
cultivation
Season 1: light rain, no heavy downpour
Season 2: Average or varying rainfall, no continuous downpour
Season 3: Continuous downpour
 Strange's (1928) Tables
• Data on rainfall and runoff in the border areas, of Maharashtra and
Karnataka and obtained the values of the runoff coefficient

Ks = R/P
as a function of the catchment character

• Catchments were characterized as "good", "average" and "bad'.

• Strange also gave a table for calculating the daily runoff from daily
rainfall.

• In this the runoff coefficient depends not only on the amount of

rainfall but also on the state of the ground.

• Three categories of the original ground state as 'dry', 'damp' and

'wet' are used by him
Extract of Strange’s Table of Runoff Coefficient Ks in
Percent

Total monsoon Runoff coefficient Ks percent

rainfall (cm)
Good Catchment Average Bad Catchment
Catchment
25 4.3 3.2 2.1

50 15 11.3 7.5

75 26.3 19.7 13.1

100 37.5 28 18.7

125 47.6 35.7 23.8

150 58.9 44.1 29.4

 Inglis and DeSouza (1929)
Formula
• Stream gauging in 53 sites in Western
India resulted, two regional formulae
between annual runoff R in cm and annual
rainfall p in cm as follows:
• For Ghat regions of western India
R = 0.85 P - 30.5
• For Deccan plateau
1
R P P  17.8
254
 Khosla's Formula
• Monthly data on rainfall, runoff and temperature data for
various catchments in India and USA considered

Rm = Pm - Lm
Lm = 0.48 Tm for Tm > 4.5°C

where Rm = Monthly runoff in cm and Rm ≥ 0

Pm = monthly rainfall in cm
Lm = monthly losses in cm
Tm = mean monthly temperature of the catchment in °C

• For Tm ≤ 4.5°C, the loss Lm may provisionally be assumed as

ToC 4.5 -1 -6.5
Lm (cm) 2.17 1.78 1.52

Annual runoff = Rm

 Watershed Simulation
• Water-budget equation for the determination of
runoff
Rs = P - Eet – f - Δ S
R = Rs + Go
Rs = surface runoff,
P = precipitation,
Eet = actual evapotranspiration,
G0 = net groundwater outflow and
ΔS = change in the soil moisture storage
f = infiltration loss

The sum of Rs and Go is considered to be given by

the total runoff R, i.e. stream-flow
 Disadvantages of Conceptual and
Physically based Models
• Data Intensive
• Complex and large number of parameters are
involved
• Implementation and calibration of such a model
can typically present various difficulties
• Requiring sophisticated mathematical tools,
significant amounts of calibration data and
some degree of expertise and experience with
the model.
• Data on the physical characteristics of the
catchment are available at the grid scale is
required
 Data required for RR Modeling
• The data required to develop a model for particular
watershed includes
– Soil characteristics,
– initial soil moisture,
– infiltration, and
– groundwater flow.
• But these data are not easily available as compared
to the rainfall (P) and runoff (Q) data.
• In most of the studies, variable ‘P’ is selected in order
to forecast Q.
• All the rainfall data are lumped with respect to time
and a single time series rainfall data will be used to
predict the runoff
 Two simple RR models, but widely
used RR models
1. Rainfall-Runoff Correlation n  PR   P  R
a
R= aP+b
 
n  P   P 
2 2

R = runoff
P = precipitation
 R  a P
a = is the slope b
b = is the intercept n
r = coefficient of correlation

n PR   P  R
r
n P   P X n R   R 
2 2 2 2
 Develop a linear RR model
Month P R Month P (cm) R
(cm) (m3/s) (m3/s)
1 5 0.5 10 30 8.0
2 35 10.0 11 10 2.3
3 40 13.8 12 8 1.6
4 30 8.2 13 2 0.0
5 15 3.1 14 22 6.5
6 10 3.2 15 30 9.4
7 5 0.1 16 25 7.6
8 31 12.0 17 8 1.5
9 36 16.0 18 6 0.5
 20
Fit Results

18 Fit 2: Linear, Y=B*X+A

Equation:
Y = 0.380197 * X + -1.55603
Number of data points used = 18
16 Average X = 19.3333
Average Y = 5.79444
Regression sum of squares = 405.607
14 Residual sum of squares = 30.5424
Coef of determination, R-squared = 0.929973
Residual mean square, sigma-hat-sq'd = 1.9089
Runoff (cu.m/s)

Coef. of correlation, R = 0.964

12

10

0 6 12 18 24 30 36 42
Rainfall in cm
Developing Multi Linear Regression Models
 In the linear regression model, the dependent variable is assumed to
be a linear function of one or more independent variables plus an
error introduced to account for all other factors.
 The multiple linear regression models are represented in equation
as:
 y = m1x1 + m2x2 + m3x3 +………+ mnxn + c
 The general multiple linear regression model considered is given in
equation
 Various models have been generated, by varying the value of m and
n in the equation and 26 different models have been developed.70
% of data used for training and 30 % used for testing
 Q (t+1) = f {P (t), P (t-1),…… P (t -m), Q(t)Q (t-1),……, Q (t - n)}
 Study Area
 Study area : Koyna watershed, Krishna Basin, District-Satara,
Maharashtra, India
 Latitude 730 34' 43" and 730 44’28’’ N ’ Longitude -170 23’ 33’’
and 170 56’ E
 Watershed has a drainage area of 891.78 km2
 Daily rainfall (P) data for nine rain gauge stations (1961-2007)
 Daily runoff (Q) data at Koyna dam available for (1961-2007).
 Hourly rainfall -runoff data available for 2005-2007 (3 years)
 Raingauge Latitude Longitude % Contri Area
Station (N) (E) bution in Sq.Km
Pratapgad 730 34' 43" 170 56' 02" 6.2 55.31
Mah. 730 40' 21" 170 55' 23" 5.72 50.99
Sonat 730 42' 30" 170 50' 14" 16.24 144.83
Location of Koyna Walawan 730 35' 43" 170 44' 17" 13.34 119
watershed Bamnoli 730 45' 43" 170 43' 46" 14.84 132.32
Karagaon 730 76' 47" 170 39' 17" 17.49 155.96
Kathi 730 49' 36" 170 29' 18" 14.61 130.29
Nawja 730 43' 24" 170 25' 37" 7.73 68.92
Koyna Dam 730 44' 28" 170 23' 33" 3.83 34.16
 Correlation Matrix
Station Mahabl Walvan Pratap Navja Sonat Kati Kargaon Bamnoli Koyna Koyna
eshwar gad Inflow

Mahble 1.00 - - - - - - - - -
shwar
Walvan 0.96 1.00 - - - - - - - -
Prtapga 0.97 0.96 1.00 - - - - - -
d
Navja 0.97 0.96 0.97 1.00 - - - - - -
Sonat 0.97 0.97 0.97 0.96 1.00 - - - -
Kati 0.97 0.97 0.97 0.96 0.96 1.00 - - - -
Kargaon 0.89 0.89 0.89 0.90 0.90 0.88 1.00 - - -
Bamnoli 0.97 0.96 0.95 0.96 0.98 0.95 0.90 1.00 - -
Koyna 0.98 0.97 0.95 0.97 0.97 0.96 0.91 0.97 1.00 -
Koyna 0.95 0.90 0.94 0.93 0.91 0.93 0.80 0.90 0.91 1.00
Inflow
 1.000

Coefficient of correlation
0.900
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
Lag in days

Cross correlation plot average Rainfall/ daily Inflow

1.0
1.0
0.8

Partial Autocorrelation
0.8
0.6
0.6 0.4
(with 5% significance limits for the autocorrelations)
0.4
Autocorrelation

0.2 (with 5% significance limits for the partial autocorrelations)

0.2 0.0
0.0 -0.2
-0.2 -0.4
-0.4 -0.6
-0.6 -0.8
-0.8 -1.0
-1.0
1 5 10 15 20 25 30 35 40
1 5 10 15 20 25 30 35 40 Lag
Lag

Autocorrelation function plot Partial autocorrelation plot

 750.00 0.00
700.00 50.00
650.00
100.00
600.00
150.00
550.00
200.00
Daily Inflow in ( Mm3)

Average Rainfall(mm)
500.00
450.00 250.00

400.00 300.00

350.00 350.00
300.00 400.00
250.00
450.00
200.00
500.00
150.00
550.00
100.00
50.00 600.00

0.00 650.00
1/1/1961
1/1/1963
1/1/1965
1/1/1967
1/1/1969
1/1/1971
1/1/1973
1/1/1975
1/1/1977
1/1/1979
1/1/1981
1/1/1983
1/1/1985
1/1/1987
1/1/1989
1/1/1991
1/1/1993
1/1/1995
1/1/1997
1/1/1999
1/1/2001
1/1/2003
1/1/2005
1/1/2007
Day start from 1/1/1961 to 31/12/2007 (47 Years)

Time series plot of daily average rainfall and daily inflow

 Time series plot of annual rainfall
8000

7000

Annual Rainfall
6000 Mean Rainfall
Total annual rainfall in mm

5000

4000

3000

2000

1000

9 61 963 965 967 969 971 973 975 977 979 981 983 985 987 989 991 993 995 997 999 001 003 005 007
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2
Years starting from 1961
 99
N 47
AD 0.897
P -V a lu e 0.020
90

Percent
50

10

1
0 2500 5000 7500

Probability plot of annual rainfall

99
N 47
AD 0.317
90 P-Value 0.528
Percent

50

10

1
-2 0 2

Normal probability plot of annual rainfall Histogram with normal distribution plot for
(Transformed data) annual rainfall series
 Statistical properties of different series of annual rainfall
Trend Analysis plot for annual rainfall for stationarity

Coeff of
Series Mean Std.Dev Skewness Kurtosis
var.
Entire Series 3930.88 1466.43 0.37 -0.25 0.15
First half 3269.01 1489.29 0.46 -0.22 -1.11
Second half 4621.53 1095.88 0.24 1.03 0.20
First one
3195.24 1562.50 0.49 0.04 -1.06
third
Second one
3815.05 1004.49 0.32 -1.27 1.44
third
last one third 4775.11 1203.43 0.25 0.89 -0.53
 8000
An nual Rainfall
+ ve Outliers 7000 Me an Rainfall

Annual Rainfall in mm
Me an +S.D
Me an-S.D
Upper 6000 S eries5
whisker 5000
4000

3000

2000
Median
1000
Lower
0
whisker

1961
1964
1967
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
2006
-ve Outliers

Time in Years

Box plot of annual rainfall series Time series plot showing annual rainfall, mean and
standard deviation.

1.0 1.0
(with 5% significance limits for the autocorrelations)
0.8 0.8
(with 5% significance limits for the partial autocorrelations)
0.6 0.6

Partial Autocorre lation

0.4 0.4
Autocorrelation

0.2 0.2
0.0 0.0
-0.2 -0.2
-0.4 -0.4
-0.6 -0.6

-0.8 -0.8

-1.0 -1.0

1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12
Lag Lag
Autocorrelation function plot annual rainfall Partial autocorrelation function plot for annual
rainfall
 3000.00
99 .9 N 153
99 AD 6 .1 2 9
2500.00
P-Va l u e <0 .0 0 5
90
Daily average rainfall in mm

Percent
2000.00
50
1500.00
10
1000.00
1
0 .1
500.00
0 2 50 0 50 0 0

0.00
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148
Probability plot of daily average rainfall
Time in days starting from 1 June to 31 Oct data
P r obability P lot for T r ansfor med Data
Daily average rainfall for a period of 47 years (1961-2007) 99.9 N 153
35 Mean -0.05237
99 AD 0.717
StDev 0.9683 P-Value 0.060
30 N 153 90

Percent
25 50
Frequency

20
10
15
1
10 0.1
-4 -2 0 2
5
Normal probability plot of daily
0
-3 -2 -1 0 1 2 3 average rainfall (transformed data)
Transformed rainfall values
 350
O b served Rain f all
T ren d lin e
300
Daily Rainfall in mm

Yt = 8 .0 3 0 7 6 + 0 .0 0 0 3 1 8 3 5 3 * t
250

200

150

100

50

0
1 1717 3434 5151 6868 8585 10302 12019 13736 15453
T im e in D ays s tarting fro m 1 J an 1 9 6 1 to 3 1 D e c 2 0 0 7
Trend analysis of daily rainfall values
 3000

Upper
2500 whisker
Avearge daily Rainfall in mm

2000

1500

Median
1000

1.0
500
0.8 (with 5% significance limits for the partial autocorrelations)
Lower whisker
0.6
0

Partial Autocorrelation
0.4
0.2
Box plot of daily average rainfall 0.0

-0.2
1.0 -0.4
(with 5% significance limits for the autocorrelations)
0.8 -0.6

0.6 -0.8

0.4 -1.0
Autocorrelation

0.2 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Lag
0.0
-0.2

-0.4
Partial autocorrelation function plot of daily average
-0.6 rainfall
-0.8
-1.0

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Lag

Autocorrelation function plot of daily average rainfall

 7000.00

6000.00
Annual Inflow
Mean Inflow
5000.00
3
Annual Inflowin Mm

4000.00

3000.00

2000.00

1000.00

0.00
1961

1963

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007
Time in Years

Time series plot of annual Inflow

 99

N 47
95 AD 0.665
P-Value 0.078
90

80
Probability %

70
60
50
40
30
20

10
5

1 9
0 1000 2000 3000 4000 5000 6000 7000 Mean -0.04936
StDev 1.079
Probability plot of annual inflow (Original data) 8
N 47
7
99 6

Frequency %
N 47
AD 0.335 5
90
Probability %

P-Va lu e 0.496
4

50 3
2

10 1
0
1 -2.4 -1.6 -0.8 0.0 0.8 1.6 2.4
-4 -2 0 2 Inflow

Normal probability plot of annual inflow Histogram with normal distribution plot for annual inflow
(transformed data)
 7000
A nnual Inflow 7000
Trend line

6000 +ve Outliers

Annual Inflow in Mm3

6000

Yt = 3775.03 + 1.38094*t

Annual Inflow in Mm3

5000 Upper
5000 whisker

4000
4000
Median
3000
3000
Lower whisker
2000
1 5 10 15 20 25 30 35 40 45 2000
Times in Years

Trend Annual Inflows Box plot of Annual inflow series.

Series Mean Std. Dev coeff of var. skewness Kurtossis

Entire Sereis 3847.80 726.69 0.19 -0.12 0.25

First half 3847.80 726.69 0.19 -0.12 0.25

Second half 3766.82 1280.96 0.34 0.97 -0.08

First one third 3862.34 860.16 0.22 -0.19 -0.22

Second one
third 3646.26 705.32 0.19 0.11 -0.48

last one third 3919.30 1410.44 0.36 0.81 -0.64

Stationarity for annual inflow series.

 1.0
0.8

0.6 (with 5% significance limits for the autocorrelations)

0.4

Autocorrelation
0.2

0.0

-0.2
-0.4

-0.6

-0.8
-1.0

1 2 3 4 5 6 7 8 9 10 11 12
Lag

Autocorrelation function plot of annual inflow

1.0
0.8

0.6
(with 5% significance limits for the partial autocorrelations)
Partial Autocorrelation

0.4
0.2

0.0
-0.2
-0.4

-0.6

-0.8
-1.0

1 2 3 4 5 6 7 8 9 10 11 12
Lag

Partial autocorrelation function plot of annual inflow

 99.9 35
N 153 Mean 0.01476
99 AD 6.750 StDev 0.9265
P-Value <0.005 30
N 153
Probability %
90
25

Frequency %
50
20
10 15
1 10
0.1
-50 0 50 100 5

0
Probability plot of daily average Inflow -3 -2 -1 0 1 2
Daily Inflow
original data
-50 0 50 100
99.9 Normal Distribution plot with histogram
N 153
99
AD 0.670
(Transformed Data)
Probability %

P-Value 0.079
90

50

10

1
0.1
-4 -2 0 2
Probability plot with daily average
inflow (Transformed data )
 350
O b served d aily in f low
T ren d lin e
300
Daily Inflow in Mm3

250 Yt = 1 0 .2 2 0 4 + 0 .0 0 0 0 2 4 0 9 9 4 * t

200

150

100

50

0
1 17 17 34 34 51 51 6 86 8 8 58 5 1 030 2 1 20 19 137 36 15 453
T im e in D ays s tarting fro m 1 J an 1 9 6 1 to 3 1 D e c 2 0 0 7

Trend Plot of Daily Inflow

 Statistical properties of different series of daily average inflow for stationarity
1.0
(with 5% significance limits for the autocorrelations)
Coeff 0.8

Series Mean Std.Dev of var Skew Kurt 0.6

0.4

A utocorrelation
Entire Series 24.89 20.11 0.81 0.40 -1.39
0.2
First half 35.36 21.25 0.60 -0.49 -1.45 0.0
-0.2
Second half 14.55 12.06 0.83 0.88 -0.25
-0.4
First one third 26.00 19.88 0.76 0.13 -1.70 -0.6

-0.8
Second one
-1.0
third 42.13 13.90 0.33 -0.21 -1.43
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
last one third 7.22 4.62 0.64 0.61 -0.79 Lag
Autocorrelation function plot of daily
inflow series
60
Upper
1.0
Whisker
50 0.8
0.6

Partial Autocorrelation
(with 5% significance limits for the partial autocorrelations)
40 0.4
Daily Inflow

0.2
30
0.0
-0.2
20
Median -0.4
-0.6
10
Lower -0.8
0 Whisker -1.0
1 5 10 15 20 25 30 35
Lag
Box plot of daily average inflow
Partial Autocorrelation function plot of daily inflow
 Performance criteria's
N
Pearson’s Correlation  Q
t 1
obs 
(t )  Q obs Qest (t )  Q est 
Coefficient (R) N 2 2

 Q
t 1
obs (t )  Q obs  Q est (t )  Q est 

E1  E2
Nash– Sutcliff efficiency E
E1
(E) N 2
E 2   Qest (t )  Qobs (t )
t 1

N
 
2

E1   Qobs (t )  Q obs
t 1

N 2

Mean Square Error MSE 

i 1 Qest (t )  Qobs (t )
(MSE) N
 Selection of input variables
1.000

Coefficient of correlation
0.900
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
Lag in days

Cross correlation Average Rainfall/ Daily Inflow

1.0
1.0 0.8

Partial Autocorrelation
0.8 0.6
0.6 0.4
(with 5% significance limits for the autocorrelations)
0.4 0.2
Autocorrelation

(with 5% significance limits for the partial autocorrelations)

0.2 0.0
0.0 -0.2
-0.2 -0.4
-0.4 -0.6
-0.6 -0.8
-0.8 -1.0
-1.0 1 5 10 15 20 25 30 35 40
1 5 10 15 20 25 30 35 40 Lag
Lag

Partial autocorrelation plot

Autocorrelation plot
 Model Input-Output Structures
Models Input Structure No. of Output
variables
Time series Models
Model 1 Q (t) 1 Q (t+1)
Model 2 Q (t-1), Q (t) 2 Q (t+1)

Model 3 Q (t-2), Q (t-1), Q (t) 3 Q (t+1)

Model 4 Q(t-3), Q(t-2), Q(t-1), Q(t)} 4 Q (t+1)
Model 5 Q(t-4),Q(t-3), Q(t-2), Q(t-1), Q(t)} 5 Q (t+1)
Model 6 Q(t-5),Q(t-4),Q(t-3), Q(t-2), Q(t-1), Q(t) 6 Q (t+1)
Model 7 Q(t-6), Q(t-5),Q(t-4),Q(t-3), Q(t-2), Q(t-1), Q(t)} 7 Q (t+1)
Model 8 Q(t-7),Q(t-6), Q(t-5),Q(t-4),Q(t-3), Q(t-2), Q(t-1), Q(t) 8 Q (t+1)

Casue Effect Models

Model 9 P (t) 1 Q (t+1)
Model 10 P (t-1), P (t)} 2 Q (t+1)
Model 11 P (t-2), P (t-1), P (t)} 3 Q (t+1)
Model 12 P(t-3),P(t-2),P(t-1),P(t)} 4 Q (t+1)
Model 13 P(t-4),P(t-3),P(t-2),P(t-1),P(t)} 5 Q (t+1)
Model 14 P (t-5), P(t-4),P(t-3),P(t-2),P (t-1), P (t)} 6 Q (t+1)
Model 15 P(t-6),P (t-5), P(t-4),P(t-3),P(t-2),P (t-1), P (t)} 7 Q (t+1)
Model 16 P(t-7),(t-6),P (t-5), P(t-4),P(t-3),P(t-2),P (t-1), P (t)} 8 Q (t+1)
Model 17 P(t-8),P(t-7),(t-6),P (t-5), P(t-4),P(t-3), P(t-2),P (t-1), P (t)} 9 Q (t+1)
Model 18 P(t-9),P(t-8),P(t-7),(t-6),P (t-5), P(t-4),P(t-3),P(t-2),P (t-1), P (t) 10 Q (t+1)

Model 19 P(t-10),P(t-9),P(t-8),P(t-7),(t-6),P (t-5), P(t-),P(t-3),P(t-2),P (t-1), P (t) 11 Q (t+1)

Models Input Structure No. of Output
variables

Combined Models
Model 20 Q (t+1) = P (t), Q (t) 2 Q (t+1)

Model 21 P (t-1), P (t), Q (t) 3 Q (t+1)

Model 22 P(t-2),P (t-1), P (t), Q (t) 4 Q (t+1)

Model 23 P (t-3), P (t-2), P (t-1), P(t),Q(t) 5 Q (t+1)

Model 24 P (t-1), P (t),Q(t-1), Q (t} 4 Q (t+1)

Model 25 P(t-10), P(t-9), P(t-8), P(t-7), P(t-6) P(t-5),P(t-4),P(t-3), P(t-2), P(t-1), P(t), Q(t-6),Q(t-5), Q(t- 18 Q (t+1)
4),Qt-3), Q(t-2), Q(t-1), Q(t)

Model 26 P(t-10), P(t-9), P(t-8), P(t-7), P(t-6), P(t-5) 17 Q (t+1)

P(t-4),P(t-3), P(t-2), P(t-1), P(t), Q(t-5), Q(t-4), Q(t-3), Q(t-2), Q(t-1), Q(t)
 Models equations developed by regression Method
Time series models

Model 1: Q (t+1) = 0.889Q (t) +1.1480

Model 2: Q (t+1) = -0.037Q (t-1) +0.921Q (t) +1.190

Model 3: Q (t+1) = 0.108Q (t-2)-0.137Q (t-1) +0.925Q (t) +1.06

Model 4: Q (t+1) = 0.022Q (t-3) +0.088Q (t-2)-0.134Q (t-1) +0.923Q (t) +1.038

Model 5: Q (t+1) = 0.041Q (t-4)-0.0162Q (t-3) +0.0093Q (t-2)-0.137Q (t-1) +0.922Q (t) +0.996

Model 6 : Q(t+1) = 0.019Q(t-5)-0.024Q(t-4)-0.013Q(t-3)+-0.092Q(t-1)+0.922Q(t)-0.137Q(t-

1)+0.921Q(t)+0.977

Model 7: Q (t+1) = -0.0632Q(t-6)-0.039Q(t-5)+0.032Q(t-4)+-0.0192Q(t-3)-0.922Q(t-2)-

0.1382Q(t-1)+0.921Q(t)+0.915

Model 8: Q (t+1) = 0.001 Q (t-7) + 0.0621 P (t-6) –0.0392 Q (t-5) + 0.032 Q (t-4)- 0.0192Q(t-
3)+0.092 Q(t-2)-0.138Q(t-1)+0.920Q(t)+0.915
 Cause Effect Models
Model 9: Q (t+1) = 0.832P (t) +2.137

Model 10: Q (t+1) = 0.675P (t-1) +0.178 P (t) +1.930

Model 11: Q (t+1) = 0.669 P (t-2) +0.258P (t-1)-0.085P (t)+2.066

Model 12 : Q (t+1) = 0.672 P (t-3) +0.253P (t-2)-0.059P (t-1)-0.027P (t) +2.066

Model13 :Q(t+1) = 0.673P(t-4)+0.254P(t-3)-0.062P(t-2)-0.010P(t-1)-.018P(t)+2.085(45)

Model 14 : Q(t+1) = 0.674 P (t-5) +0.254P(t-4)-0.061P(t-3)-0.012P(t-2)-0.008P(t-1)0.010P(t) +2.075

Model 15 : Q(t+1) = 0.160P(t-6)+ 0.207P (t-5) +0.352P(t-4)-0.074P(t-3)-0.008P(t-2)-0.021P(t-1)-0.007P(t)

+0.692Q(t)+0.341

Model 16 : Q(t+1) = 0.354P(t-7)- 0.159P(t-6)+ 0.207P (t-5) +0.352P(t-4)-0.074P(t-3)-0.024P(t-2)-0.006P(t-

1)-0.014P(t) +0.692Q(t)+0.354

Model 17 : Q(t+1) = -0.159 P(t-8)+0.208P(t-7)+ 0.352P(t-6)-0.074 (t-5) -0.007P(t-4)-0.023P(t-3)-0.003P(t-

2)-0.001P(t-1)-0.014P(t) +0.692Q(t)+0.366

Model 18 : Q(t+1) = -0.158 P(t-9)-0.208 P(t-8)+0.352P(t-7)- 0.074P(t-6)-0.007 (t-5) -0.022P(t-4)+0.005P(t-

3)-0.005P(t-2)+0.007P(t-1)-0.022P(t) +0.692Q(t)+0.386

Model 19 : Q(t+1) = -0.158 P(t-10)-0.208 P(t-9)+0.352P(t-8)- 0.074P(t-7)-0.007 (t-6) -0.023P(t-

5)+0.005P(t-4)-0.005P(t-3)+0.007P(t-2)-0.024P(t-1) +0.003P(t)+0.692Q(t)+0.383
 Combined Models

Model 20 : Q(t+1) = 0.214P(t)+0.766Q(t)+0.306 111.386

Model 21 : Q (t+1) = -0.186P (t-1) +0.471P (t) +0.679Q (t) +0.510 60.016

Model 22: Q (t+1) = -0.177P (t-2) +0.223P (t-1) +0.259P (t) +0.659Q (t) +0.159

Model 23 : Q(t+1) = -0.164P(t-3)+0.205P(t-2)+0.356P(t-1)-0.102P(t)+0.693Q(t)+0.270

Model 24 : Q (t+1) = -0.081P (t-1) +0.239P (t) -0.026Q (t-1) +0.777Q (t) +1.015

Model 25 : Q(t+1) = -0.025P(t-10)+0.015P(t-9)-0.041P(t-8)-0.042P(t-7)-0.020P(t-6)-0.113p(t-5)-0.073P(t-

4)+0.224P(t-3)=0.350P(t-2)-0.073P(t-1)-0.029P(t)+0.054Q(t-6)-0.009Q(t-5)+0.039Q(t-4)+0.009Q(t-
3)+0.134Q(t-2)+0.005Q(t-1)+0.601Q(t)+0.106

Model 26 : Q(t+1) = -0.015P(t-10)-0.056(t-9)-0.043P(t-8)-0.022P(t-7)-0.112P(t-6)-0.049P(t-5)+0.225P(t-

4)+0.351P(t-3)-0.076P(t-2)-0.006P(t-1)-0.024P(t)+0.006Q(t-5)-+0.041Q(t-4)+0.015Q(t-3)+0.135Q(t-
2)+0.007Q(t-1)+0.600Q(t)+0.163
 Statistical performance from various regression Models

Model No. of input parameters R MSE E

Time series Models

Model 1 1 0.912 132.46 0.832
Model 2 2 0.911 135.94 0.827
Model 3 3 0.914 129.20 0.7918
Model 4 4 0.914 128.80 0.836
Model 5 5 0.915 128.63 0.836
Model 6 6 0.925 128.49 0.837
Model 7 7 0.914 129.07 0.837
Model 8 8 0.914 129.06 0.836
Casue Effect Models
Model 9 1 0.911 141.42 0.820
Model 10 2 0.906 149.46 0.810
Model 11 3 0.906 149.44 0.810
Model 12 4 0.906 149.60 0.810
Model 13 5 0.906 149.58 0.810
Model 14 6 0.906 149.32 0.809
Model 15 7 0.956 68.21 0.914
Model 16 8 0.956 67.92 0.923
Model 17 9 0.954 67.77 0.913
Model 18 10 0.956 67.37 0.914
Model 19 11 0.955 69.07 0.912
 Statistical performance from various regression Models
Ctd.

No. of input
Model R MSE E
parameters
Combined Models
Model 20 2 0.926 111.387 0.858
Model 21 3 0.961 60.01 0.923
Model 22 4 0.955 69.37 0.911
Model 23 5 0.955 68.07 0.913
Model 24 4 0.924 114.2 0.859
Model 25 18 0.958 64.45 0.918
Model 26 17 0.957 66.91 0.915
 350.0

300.0
Observed Runoff Plot of observed and predicted runoff for
Predicted Runoff
250.0 Model 21 (During validation period)
Runoff Mm3

200.0

150.0

100.0

50.0

0.0
1 312 623 934 1245 1556 1867 2178 2489 2800 3111 3422 3733 4044 4355 4666 4977
Time in Days

350
3
Predicted runoff in Mm

300 Scatter Plot for Model 21 (During

250
validation Period for best Regression
Model)
200

150

100

50

0
0 50 100 150 200 250 300 350
3
Observed Runoff in Mm
 DISTRIBUTED DATA MODELS
Models Input Structure No. of Output
variables
Distributed data Casue Effect Models
Model 1 P1 (t),P2(t), P3 (t), P4(t), P5 (t), P6 (t), P7 (t), P8 (t), P9 (t) 9 Q (t+1)

Model 2 P1 (t-1), P1(t), P2(t-1), P2(t), P3 (t-1), P3(t), P4(t-1), P4(t), P5 18 Q (t+1)
(t-1),P5(t), P6(t-1), P6(t), P7 (t-1), P7(t), P8(t-1), P8(t), P9(t-
1), P9(t)
Distributed data Combined Models
Model 3 10 Q (t+1)
P1 (t),P2(t), P3 (t),P4(t), P5 (t),P6(t), P7 (t),P8(t),P9(t),Q(t)

Model 4 P1 (t-1), P1(t), P2(t-1), P2(t), P3 (t-1), P3(t), P4(t-1), P4(t), P5 19 Q (t+1)
(t-1), P5(t), P6(t-1), P6(t), P7 (t-1), P7(t), P8(t-1), P8(t), P9(t-
1), P9(t), Q(t)
DISTRIBUTED DATA CAUSE EFFECT MODEL 1 (Developed Equation)

Model 1 : Q (t+1) = 0.296 P1 (t) +0.0185 P2(t) + 0.038 P3 (t) -0.0580 P4(t) +
0.0261P5 (t) + 0.0972P6 (t) + 0.0366 P7 (t) + 0.0217 P8 (t) + 0.2308 P9(t) +
1.269

DISTRIBUTED DATA CAUSE EFFECT MODEL 2 (Developed Equation)

Model 2 : Q(t+1) = 0.2386 P1 (t-1) + 0.1572P1 (t) -0.028 P2 (t-1)- 0.0165P2 (t) - 0.055
P3 (t-1)+ 0.001 P3 (t)+ 0.0042 P4 (t-1)+ 0.0043 P4 (t) - 0.0017 P5 (t-1) + 0.0232 P5 (t)
+0.078 P6 (t-1)+ 0.054 P6 (t)+ 0.0283 P7 (t-1)+ 0.013 P7 (t)+ 0.032P8(t-1)+ 0.031P8 (t)+
0.1403 P9 (t-1)+ 0.0709P9 (t)+ 0.856
 DISTRIBUTED DATA COMBINED MODEL 3 (Developed Equation)

Model 3 : Q (t+1) = -0.022 P1 (t) +0.063 P2(t) +0.140 P3 (t) - 0.077 P4 (t) + 0.0033
P5 (t) +0.021 P6(t) +0.0991 P7 (t) -0.048 P8 (t)+ 0.0435 P9 (t)+ 0.779 Q(t)+ 0.7355

DISTRIBUTED DATA COMBINED MODEL 4 (Developed Equation)

Model 4 : Q (t+1) = 0.0061 P1 (t-1) + 0.049 P1(t)+ 0.0558P 2(t-1) + 0.0443 P2(t) -
0.03268 P3 (t-1) -0.031 P3(t) - 0.0332 P4 (t-1) +0.0453 P4 (t) + 0.037P5 (t-1)
+0.0419 P5(t) +0.0088 P 6 (t-1) +0.0378 P 6(t) + 0.0566 P7 (t-1) +0.0195 P7 (t) + P8
(t-1) -0.0343 P8 (t) + 0.1857P 9 (t-1)+ 0.6521 P9(t)+ 0.3818 Q(t)+ 0.381852
 MODEL RESULTS BOTH DIST. DATA AND LUMPED DATA

Model No. of input parameters R MSE E

Casue Effect Models

Model 1
1 0.911 141.422 0.820
(Lumped)
Model 1 (Dist) 9 0.917 137.57 0.825
Model 2
2 0.906 149.464 0.810
(Lumped)
Model 2 (Dist) 18 0.912 145.135 0.816
Combined Models
Model 3
2 0.926 111.387 0.858
(Lumped)
Model 3 (Dist) 10 0.936 101.96 0.88
Model 4
3 0.961 60.01 0.923
(Lumped)
Model 4 (Dist) 19 0.967 65.27 0.923
 Conclusions from Data analysis
Conclusion from Annual rainfall series
 Original data of annual rainfall series do not follow a normal distribution
hence log transformation was carried out and found that it follows a
normal distribution

 Annual rainfall series data is found to be non stationary

 From trend analysis it was found that annual rainfall data follow a linear
upward trend

 The box plot analysis showed positive and negative outliers in rainfall
series
Conclusion from Daily average rainfall series

 Daily average rainfall series data required transformation to follow normal

distribution hence log transformation is carried out and found to follow normal
distribution

 A trend analysis showed a linear falling trend in daily average rainfall data
 Conclusion from Daily average rainfall series (Ctd.)

The daily average data found to be non stationary

Box plot analysis showed no positive and negative outliers

Trend analysis of daily rainfall data shows almost horizontal

trend.
Annual Inflow series

Annual inflow data required transformation to follow the

normal distribution.

Trend analysis shows almost horizontal trend

The data is found to be non stationary

 Box plot analysis showed positive outliers in the data set

 Conclusion from Daily average Inflow

Daily average inflow data required transformation to follow the normal

distribution

Trend analysis showed falling trend, similar to that of daily rainfall

series

Daily average Inflow data found to be non stationary

Box plot showed no positive and negative outliers

 Conclusions from Multiple linear Regression Models

MLR models resulted in better scenario in predicting average

and normal peak flow. Shows significant improvement from time
series to cause effect and cause effect to combined models.

 MLR model with combined input of rainfall and runoff predicted

better than considering rainfall runoff separately.

It is found that MLR is performing better due to longer length of

observed data.
 2. SCS-CN Technique
• Developed by Soil Conservation Service of USA for
small agricultural watersheds, latter extended to urban
watershed.

• Data needed is the rainfall and type of soil and land

use

• Based on the concept of reservoir operation

– Let the water storing capacity of the soil (S)
• S-maximum retention of water in the soil – capacity of the reservoir
– Runoff will start only when this reservoir is full
But actual soil moisture is Sa = P-Ia-R

Similarly let
R (actual runoff)
P-Ia = Potential runoff
R =0 if P<Ia

The hypothesis of SCS-CN technique is that the ratio of actual and potential
moisture retained is equal to the ratio of actual and potential runoff

P  Ia  R R

S P  Ia

R
P  I a 2
By rearranging this the SCS-CN equation is
P  Ia  S

SCS-CN found that Ia = 0.2 S  P  0.2S 2

R
P  0.8S
• A number of curves has been
plotted for the measured P and
R

• SCS fund a number of

dimensionless curves with
numbers called Curve
Numbers 0CN100

– For impervious and water

bodies CN=100
– It varies with respect to
antecedent soil moisture, soil
type, and land use.
– This CN is related to S
1000
S  10
CN
In fps system
P and R in inches
25400
S  254 For various land uses the CN is taken as
CN weighted curve Number
Where P and R in mm
 Runoff curve numbers for selected agricultural, suburban and urban land uses (antecedent moisture condition II, Ia = 0.2S)
Hydrologic Soil Group
Land Use Description
A B C D
without conservation treatment 72 81 82 91
Cultivated Land:
with conservation treatment 62 71 78 81
Poor condition 68 79 86 89
Pasture or range land:
Good condition 39 61 74 80
Meadow: Good condition
30 58 71 78
thin stand, poor cover, no mulch 45 66 77 83
Wood or Forest land:
Good condition 25 55 70 77
Good condition: grass cover on 75% or more of the
Open spaces, lawns, parks, golf area 39 61 74 80
course, cemeteries, etc.
fair condition: grass cover on 50% to 75% of the area
49 69 79 84
Commercial and business areas (85% impervious)
89 92 94 95
Industrial districts (72% impervious)
81 88 91 93
Average lot size Average % impervious
1/8 acre or less 65 77 85 90 92
1/4 acre 38 61 75 83 87
Residential
1/3 acre 30 57 72 81 86
1/2 acre 25 54 70 80 85
1 acre 20 51 68 79 84
Paved parking lots, roofs, driveways, etc. 98 98 98 98
Paved with curbs and storm sewers 98 98 98 98
Street and roads gravel 76 85 89 91
dirt 72 82 87 89
Soil Group Description of the Hydrologic Soil group by USSCS

A Lowest runoff potential. Includes deep sand with very little clay and silt.
B Moderately low runoff potential. Mostly sandy soil less deeper than A
C Moderately high runoff potential. Comprises shallow soil of clay, colloids.
D Highest runoff potential. Includes mostly clay of high swelling percent, nearly
impermeable soils.

Antecedent moisture 5-day antecedent rainfall (cm)

condition class Non-monsoon period Monsoon period
I. Optimum soil condition Less than 1.25 Less than 3.5
from plastic limit to wilting
point
II Average value for 1.25 to 2.75 3.5 to 5.25
annual floods
III. Heavy rainfall or low Over 2.75 Over 5.25
rainfall and low
temperature during five
days preceding the storm
 ? Determine the runoff from a watershed of 60 ha.
The following data is available
• Rainfall = 100 mm
• Row crop good condition 40 ha (soil type C)
• Wood land good condition 20 ha (soil type B)

From the CN table find the CN corresponding to these two soil type and
land use
– For row crop 40 ha = 82
– For wood land 20 ha = 55
– The weighted CN = CN  (82 * 40)  (55 * 20) = 73
60

25400
S= S  254 = 93.9 mm
73

 P  0.2 S 2
R = 37.7 mm
P  0.8S
 Fixing the Capacity of a reservoir
Required Storage = Dt- St (over a year)
 Solve and find the capacity of the reservoir
required??

Mean flow Cumulative Cumulative

Month cu.m/d Volume/month Demand flow demand

Jan 60 6696000 4464000 6696000 4464000

Feb 45 4536000 4032000 11232000 8496000

mar 35 3906000 4464000 15138000 12960000

Apr 25 2700000 4320000 17838000 17280000

May 15 1674000 4464000 19512000 21744000

Jun 22 2376000 4320000 21888000 26064000

Jul 50 5580000 4464000 27468000 30528000

Aug 80 8928000 4464000 36396000 34992000

Sep 105 11340000 4320000 47736000 39312000

Oct 90 10044000 4464000 57780000 43776000

Nov 80 8640000 4320000 66420000 48096000

Dec 70 7812000 4464000 74232000 52560000

 80000000.00

Cum. Inflow
60000000.00 Cum. Demand
Cum. Inflow/demand

40000000.00

20000000.00

0.00

0.00 4.00 8.00 12.00

Time
 Mean
flow
cu. Volume/ Cumulative Cumulative
Month m/d month Demand Surplus Deficit surplus deficit

Jan 60 6696000 4464000 2232000 2232000

Feb 45 4536000 4032000 504000 2736000

mar 35 3906000 4464000 -558000 -558000

Apr 25 2700000 4320000 -1620000 -2178000

May 15 1674000 4464000 -2790000 -4968000

Jun 22 2376000 4320000 -1944000 -6912000

Jul 50 5580000 4464000 1116000 1116000

Aug 80 8928000 4464000 4464000 5580000

Sep 105 11340000 4320000 7020000 12600000

Oct 90 10044000 4464000 5580000 18180000

Nov 80 8640000 4320000 4320000 22500000

Dec 70 7812000 4464000 3348000 25848000

 Sequent Peak Algorithm
• If N year flow data are available
• It is assumed the cyclic pattern repeats
700

600

500

400

300

200

100

0
0 100 200 T i me i n mont hs 300 400 500

T i me s e r i e s pl ot of T ot a l i nf l ow
 Sequent Peak Algorithm
• Estimate the Net-flow volume
= Inflow-outflow (demand)
• Plot the cumulative net-flow Vs time
• Find the sequent peak (high) and trough (low)
• The maximum difference gives the capacity of the reservoir.
 Drought
• Meteorological drought
– If rainfall is decreased more than 25% of normal rainfall
• Severe drought – if rainfall is 75% of normal rainfall
• Severe drought – if rainfall is 40-75% of normal rainfall
• Hydrological drought
– Depletion of surface and ground water
– Drying of lakes reservoirs aquifers
• Agricultural drought
– When soil moisture and rainfall are inadequate during growing period of a crop
• Economic drought
 Self Study
• Solve all the problems given in this section

• Others techniques of stream flow measurement

– Electromagnetic method
– Ultrasonic method
– Indirect methods
• Flow measuring structures (weirs and notches?)
• Slope area methods (based on resistance equation for
uniform flow and Manning’s formula)
 Current research in this area
1. Application of ANN and GP in rating
curve parameter estimation
2. Finding new sophisticated instruments
for stream-flow measurement
3. Etc…..
 Research area in this topic
• Establishment of rainfall-runoff relationship
– Black box models –ANN
– Conceptual models
– Physical models
• Inclusion of GIS and RS
• Drought prediction and mitigation