(Note: attachments are not allowed in group posts)

Dear Babu

please note you may use 'OMRAMKEWALOM' or
'omramkewalom' password to open the *.wpd and *.wp6
files.
Congratulations.

You are doing good works. May you please refer to some
of publications from National Institute of Hydrology,
Roorkee. Dr. S. K. Mishra, has done some what similar
works. I am attaching some of the text from review of
literature from my ph.d. dissertation ‘ Harvesting
Rainwater for Optimal Agricultural Production from
PAU, Ludhiana, 1999.
You may choose to define tabulated value of r, the
regression coefficient ( underroot of coefficient of
determination) then compare with computed values. The
tabulated values should be more than computed values
at five and one percent level of significance for the
corresponding degree of freedom (N-2).
You can use either statgraphics for windows,
statistics, minitab, and so on to go for regression
and correlation . better if you could lay hands on
lotus smartsuite or regional softwares.
It is not necessary to study validity of all eight
models for all the watersheds. Each watershed over the
year, may yield different relations for given rest of
conditions. Say, Allred and Haan (1996) presented
partition of rainfall. Each rainfall event is
partitioned in 6 parts, each one over the year or
monsoon, gives varying values.
Using acceptable/non acceptable criteria is always
acceptable Haan C T (1977)-statistics in Hydrology.
You may read Hand book of applied hydrology 1988
edition, an exhaustive chapter on RR
If possible, try to develop frequency distribution
curves for each of the watershed separately, then you
can pick up selected frequency distribution models.
Choose upper and lower confidence limit for each of
the watershed. Your computed values should bye and
large contain within lower and upper bounds of
confidence say, 90 pc, 95 pc and 99 pc. You may use
statgraphics or minitab
CHAPTER II
REVIEW OF LITERATURE
Water harvesting, although an age-old practice, has
been defined as ' the practice of collecting water
from an area treated to increase runoff from rainfall.
It involves collection and storage of runoff water in
suitable locations by reducing the losses due to
seepage and due to evaporation (Singh, 1994 and Samra
et al., 1996) in scarcity area. Hydrologic modelling
involves the systematic evaluation and synthesis of
rainfall and different watershed parameters through
mathematical equations that simulate the physical
processes in the watershed to get a reliable estimate
of rain induced runoff. The first attempt at
modelling the rainfall-runoff process was developed
over 140 years ago in the rational formula for the
prediction of flood peak. Now, there are several
models available for calculating the runoff rates and
runoff volumes but mostly for localized regional
conditions. Although sizeable quantum of literature
on different aspects of the subject are collected but
in view of paucity, review of only the relevant and
latest literatures are reported in the following sub
head.
(a) Agroclimatic Data Processing
(b) Runoff Estimation
(c) Water Harvesting Structure
(d) Optimal Irrigation Scheduling

2.1 AGRO-CLIMATIC DATA PROCESSING
2.1.1 Rainfall Analysis
Frequent usages of Normal, Log normal, Pearson type
-III and Extreme Value Type I Distribution (Gumbel's
and Chow's) methods of frequency distribution have
been commonly used by many researchers for analysis
of rainfall and runoff time series (Bradley et al.,
1991; Mukherjee et al., 1991; Eltahir, 1992;
Farquharson et al., 1992; Rakhecha et al., 1992;
Singh et al., 1992; Win and Vandewiele, 1992;
Agnihotri et al., 1993; Rajput et al., 1993; Abtew
et al., 1995; Deora and Tiwari, 1995; Farquharson et
al., 1996; Hussein 1996; and Onof et al., 1996).

2.1.2 Effective Monsoon
A computer programme developed by Ashokraj (1979) for
study of the onset of effective monsoon, start and
duration of critical dry spells, start and duration of
wet spell with amount of rainfall received during the
wet spells and withdrawal of effective monsoon has
been used by Tiwari and Saxena (1987); Deora and
Tiwari (1995a) for prediction of onset of effective
monsoon and other features. Mavi (1996) presented
monsoonic characteristics of different districts in
Punjab. Parmar and Gandhi (1981) reported that the
probability of onset of south west monsoon for Rajkot
during the period June 11 to June 30 and cessation of
monsoon during Sept 24 to Oct 13 was 86% and 93%
respectively.

2.1.3 Water Balance Studies:
Adhikari et al., (1993) using 10 year data on
rainfall and evaporation for the low rainfall tract of
Bellary and using auto-regressive models, presented
monthly water deficit values from climatic water
balance studies. Karnieli and Ben (1993) measuring
runoff events from four watersheds in Southern Arizona
described two distinct cases of the water balance
equations (i) a storm predicting runoff (ii) a storm
without runoff. Boughton (1995) presented a water
balance model for semi arid region. Dasgupta (1995)
using the Thornthwaite and Mather procedure presented
computation of climatic water balance for Purulia
(West Bengal), Samastipur (Bihar), Puri( Orissa) and
Nagaon (Assam) districts as an alternative to
rainwater harvesting.

2.2 RUNOFF ESTIMATION
2.2.1 Rainfall - Runoff Models
Methods of estimating runoff using computer models
necessarily neglect some factors and make simplifying
assumptions regarding the influence of others.
Sometimes a black box approach is adopted with
rainfall as an input and runoff as an output. The
"black - box " contains the interaction of all
possible parameters that are either not well
understood or expressed quantitatively through
deterministic or stochastic relationships. The USDA,
SCS, CN (Soil Conservation Service, Curve Number)
method also known as Hydrologic Cover Complex method
(Hudson, 1987; Chow et al., 1988; Singh et al.,
1990; Varshaney, 1990; Dhruvanarayan, 1993; Hudson
1993; Schwab et al., 1993; Mutreja, 1995 and Murty,
1998) based on initial abstraction and recharge
capacity of a watershed has been suitably modified
largely for Indian conditions (Anonymous ,1972).
Hjelmfeelt (1991,1996) discussed the heritage of the
method and provided an interpretation of antecedent
moisture conditions. Ponce and Hawkinds (1996)
critically reviewed and examined this method,
clarified its conceptual and empirical basis;
delineated its capabilities, limitations, uses and
described future scope of research in SCS CN
methodology. Hjelmfeelt (1991), Hawkins (1993), Bonta
(1997), Mishra and Dwivedi (1998) and Mishra et
al., (1998) suggested procedures for deciding curve
numbers for a watershed using field data. The methods,
however, use discrete annual storm events for
computing curve numbers and, thus, the resulting curve
numbers are applicable to only high flow studies. Most
of the researchers used the SCS CN method with
modified values of initial abstraction and antecedent
moisture index for a variety of soil physical
conditions (Selvarajan 1990; Muzik and Gladwell, 1993;
Singh, 1995; Steenhuis et al., 1995 Hussein 1996;
Ponce and Hawkins 1996; Ramana Rao et al., 1996;
Mishra,1997; Mishra and Singh 1998; Mishra et al.,
1998 and Gaur 1999).

2.2.2 Models for Estimating Runoff Rates
Katyal et al., (1990) and Singh et al., (1990)
showed that maximum peak discharge could be obtained
as a function of rainfall depth, rainfall intensity,
rainfall duration(equal to time of concentration) and
the previous discharge. Cook's method to estimate peak
runoff rate considers four watershed characteristics
viz., relief, infiltration, vegetal cover and surface
storage. The accuracy of the method depends upon the
location specific appropriate numerical values of the
parameters/factors. Mullem (1991a, 1991b) using green
ampt infiltration model, studied the runoff and peak
discharge. Bonta and Rao (1992) presented a method
estimating peak runoff rates in the CREAMS model by
analysing precipitation and runoff data from
Agricultural North Appalachian Experimental watersheds
(USA). The SCS developed computer programme TR-55 to
predict runoff rates from small urban watershed with
uniform rainfall, and this method was also found
satisfactory for estimating runoff from rural area
(Schwab et al.,1993). Onyando and Sharma (1995) using
22 year data (1958-80) from headwaters of the Lake
Victoria drainage basin in Kenya, presented simulation
of direct runoff volume and peak rates for rural
catchments in Kenya. Michael and Sorooshian (1994)
compared simple versus complex runoff models in a
midsted semi arid watershed. Singh (1995 vp)
described various computer models on watershed
hydrology, being used in different regions and for
different set of conditions. Buchtele et al., (1996)
used the conceptual soil moisture accounting model
’SACRAMENTO’, and the physically based forest
hydrological model ’BROOK’ for runoff simulations for
small and medium sized basins to investigate the
different component, contributing to the total runoff.
Chiew et al., (1993); Chiew and McMohan (1994); and
Chiew et al., (1996) compared six rainfall - runoff
modelling approaches viz., simple polynomial equation,
simple process equation (tanh equation), simple time
series equation (Tsykin equation), complex time series
model(IHACRES), a simple conceptual model(SFB) and
complex conceptual model (MODHYDROLOG) to simulate
daily, monthly and annual flows in the eight
unregulated catchments. The complex conceptual model
gave, by far, the best simulation of daily high and
low flows and could estimate adequately daily flows
for the wetter catchments. Chiew et al., (1993 and
1996) explored similarities and differences between
conceptual rainfall runoff models developed by
hydrologists and land surface parameterisation schemes
for incorporation into atmospheric general circulation
models (AGCMs). A few conceptual models have also
reported estimation of runoff rates (Franchini and
Pacciani, 1991; Jain, 1993; Khan, 1995 and
Franchini and Galeati, 1997).

2.2.3 Models for Estimating Runoff Volumes
Total volume is of primary interest in the design of
water harvesting structures for irrigation and
drainage measures. It is thus essential to predict
the total volume of runoff that may result from a
watershed during the design flood. Analysis of
watershed characteristics (rainfall amount, rainfall
duration, rainfall intensity, and antecedent
precipitation) have been used in linear and multiple
linear regression models for predicting runoff volumes
from small agricultural and non agricultural (treated
and untreated) watersheds by (Singh et al., 1990;
Agnihotri et al., 1993; Chen and Singh, 1993; Rajput
et al., 1993; Onyando and Sharma, 1995; Pandit and
Gopalkrishnan, 1996; and Singh et al.,1996). The
regression relationships of runoff on rainfall on
daily/monthly and annual bases from a variety of
gauged and ungauged micro/mini watersheds have also
been reported by a section of researches. Uncertainty
associated with prediction of runoff rates, and runoff
volumes on different sizes of catchments have also
been reported (Goldman et al., 1990; Melching et
al., 1990 and Melching 1992).

2.2.4 Monthly Water Yield Models
Some regional monthly rainfall-runoff models have
been developed for estimating runoff from rainfall
for different regions under the assumed
hydro-meteorological characteristics. The NPB model
(Bhattarai, 1982; Rai, 1993; Sur et al.,1998 and
Narda et al.,1999) needs the following input:

a) Daily rainfall, average pan evaporation, lower and
upper steady infiltration rate, lower and upper
exponent in the Horton's infiltration equation and
surface runoff and
b) Capacity of interception storage, upper soil
storage, sub soil storage and drainage component of
upper soil storage

The NPB programme simulates changes in soil moisture
in different storage which are influenced by
evapotranspiration and rainfall. First, daily rainfall
is added to the soil moisture storage and runoff (if
any) is predicted. Soil moisture storage is adjusted
by calculating evapotranspiration losses and water
movement by drainage. The next daily rainfall data is
examined and above operation is repeated. Daily
rainfall is added to the interception storage. If the
interception storage overflows then excess is added
to the upper soil storage. If the excess in upper
soil storage overflows, then excess is added to the
drainage storage. If the drainage storage also
overflows then runoff occurs on that day and then on
similar lines, daily and monthly values of runoff are
printed. Evapotranspiration demand is met from the
interception storage when the storage contains
moisture. When this interception storage is empty,
actual evapotranspiration coming from upper soil
storage is computed according to the moisture
available in the storage. Movement of moisture from
the drainage storage to the sub soil storage occurs at
the daily infiltration rate. Movement of moisture from
the sub soil storage to ground water is allowed by
multiplying the moisture level in the sub soil storage
by daily depletion factor.


> In this regard, Hughes (1995) applied monthly
rainfall runoff models to arid and semiarid catchments
for estimating water resources. Ibrahim and Cordery
(1995) based their studies on partitioning of monthly
rainfall into soil water storage, ground water
recharge, surface runoff, evaporation and other
losses for estimating runoff volume. Allred and Haan
(1996) using daily precipitation, monthly
evapotranspiration and six watershed parameters,
developed a conceptual computer modelling programme
'SWMHMS' to simulate monthly runoff from a small
non-urban watersheds. They used Blaney?Criddle
equation to calculate evapotranspiration and SCS curve
number method to estimate surface runoff volume. The
'SWMHMS' provides options for both optimization and
sensitivity analysis. The programme is tested with
data from six watersheds diversely located in Georgia,
Texas, Oklahoma, North Carolina, Ohio and Idaho. The
optimal curve number for majority of watersheds was
closest to an SCS type II value. The model predicted
the best on Georgia, Texas, Oklahoma and North
Carolina (with low accumulation of snowfall).

22.4. Critics on rainfall runoff modelling:
Although USDA SCS CN method with modifications on
either initial abstraction or antecedent precipitation
index is used in regression forms, conceptual model,
mathematical forms and some of the computer
programmes. The computer programme with limited and
dynamic natured soil physical properties and watershed
characteristics are still not widely usable in other
regions. Variation in use of the existing computer
programmes, inspite of using the same USDA SCS CN
method, are attributed to assumption with respect to
soil properties, watershed characteristics,
periodicity of study (event based daily, monthly,
annual). Such limitation on nature and requirement of
data, necessitated to make use of the existing
computer programmes with necessary improvement on data
input, computation procedure and input-output pattern.
In the present study, a computer programme based on
the USDA, SCS, CN method is developed in dBASE
(III+) and presented in Appendix 'A'. The SWMHMS
programme has been modified with respect to the data
requirement, computation procedure and input-output
pattern, required for the specific region.