Estimation of Design Flood for Rivers of Saurashtra Region contributing into the Gulf of Khambhat

Design flood has been estimated for rivers of Saurashtra region contributing into the Gulf of Khambhat using deterministic as well as statistical approach for planning, design and management of hydraulic structures. By comparing the results obtained by these approaches, one can easily estimate the flow rate or peak discharge to a given design return period and can establish the suitability of approach for this study area. Nine river basins with 20 dams of Saurashtra region were analyzed in this study. Though Saurashtra is one of the most water scarce regions of India yet it suffers from the flooding problem, as the numbers of rainy days are very less and the rainfall intensity is very high. Due to being a regulated basin, dam wise study was preferred. Deterministic approach was carried out using synthetic unit hydrograph (SUH) and regional flood formulae (RFF) methods for subzone-3a provided in Central Water Commission (CWC) report, 2001. Statistical approach was carried out using Rainfall frequency analysis employing Gumbel’s EV1distribution. As there is no spill by these hydraulic structures and the annual flood data for the nine river sites are heavily affected by the storage dams in the upstream. Hence these data violate the basic principle of virgin flow. Hence the analysis of these data was not attempted further. The main objective of study was to carry out the rainfall frequency analysis for these river basins to get 24 hour rainfall for a return period of 25, 50 and 100 years for an individual basin instead of using the value obtained by iso-pluvial map to estimate the design flood. The overall results reveals that due to construction of number of dams in 9 river basins, design flood estimation on each dam by using deterministic approach is more feasible. Revised design floods using SUH and RFF method on the basis of estimated rainfall indicates over-estimated and under-estimated design floods. Since the percentage difference is very less between revised SUH and revised RFF method. So, for safety purpose one with higher value should be used.


intRODuctiOn
Flood, a natural disaster is responsible for loss of life and property world over.Floods damage property and endanger the lives of humans and animals and also affect the environment and aquatic life negatively.Floods have been occurring repeatedly in India.Approximately 40 million ha area (12%) in India has been identified as flood prone 18 .For mitigating the flood disasters, various structural and non-structural measures are adopted.Structural measures include protection works and flood embankments while non-structural measures include flood forecasting, flood warning and flood plain zoning.Design flood estimates are required for the design of various hydraulic structures such as weirs, barrages, dams, embankment etc. and flood protection / relief schemes 5,14 .Flood forecasts are required for operation of various flood control structures, for taking emergency measures such as maintenance of flood levees, evacuating the people to safe localities etc. Whenever rainfall or river flow records are not available at or near the site of interest, it is difficult for hydrologists or engineers to derive reliable flood estimates directly.In such situation, flood formulae developed for the region are one of the alternative methods for estimation of design floods, particularly for small-to-medium catchments.The conventional flood formulae developed for different regions of India are empirical in nature and do not provide estimates for a desired return period.A number of studies have been carried out for estimation of design floods for various structures by different Indian organizations.Among these the prominent studies are carried out jointly by the Central Water Commission (CWC), Research Designs and Standards Organization (RDSO), and India Meteorological Department (IMD) using the method based on synthetic unit hydrograph and design rainfall, considering physiographic and meteorological characteristics for estimation of design floods 3 and regional flood frequency studies carried out by RDSO using the USGS and pooled curve methods 12 for various hydrometeorological subzones of India.The concept of the geomorphologic instantaneous unit hydrograph (GIUH) was introduced by Rodriguez-Iturbe and Valdes 17 .The topographic and geometric properties of the watershed and its drainage channel network are reflected by geomorphology 6 .Snyder (1938)  proposed synthetic unit hydrograph approach (SUH) for ungauged basin 21 .A desirable method should satisfy the requirements of universal acceptability; ease in use with a minimum of data; robustness in nature; and reliability 14 .Now a days GIS and remote sensing techniques are being used extensively to monitor the disasters like droughts and floods 7 .
Practically in the design of all hydrologic structures the peak flow that can be expected with an assigned frequency (say 1 in 100 years) is of primary importance to adequately design the structure to accommodate its effect.The design of bridges, culvert waterways and spillways for dams and estimation of scour at a hydraulic structure are some examples wherein flood-peak values are required.To estimate the magnitude of a flood peak Fig. 2: location of G&D sites and rain-gauge in river basin map the following methods are available: (1) Rational method; (2) Empirical method; (3) Unit-hydrograph technique and (4) Flood-frequency studies 10 .The use of a particular method depends upon (i) the desired objective, (ii) the available data and (iii) the importance of the project.Further, the rational method is applicable only to small-size (<50 km 2 ) catchments and the unit-hydrograph method is normally restricted to moderate-size catchments with areas less than 5000 (km 2 ) 13,15 .
In present study, design floods for various structures in the 9 river basins namely Wadhavan-Bhogavo, Limbdi-bhogavo, Sukhbhadar, Utavali, Padalio, Khalkhalia, Ghelo, Keri and Kalubhar have been estimated.Deterministic approach based on unit hydrograph theory developed by CWC 4 and statistical approaches based on frequency analysis has been used for the design flood estimation.

Study area and data availability
Saurashtra basin is a region of western India, located on the Arabian Sea coast of state of Gujarat.Saurashtra is bounded on three sides by waters of sea, namely in the north by the Gulf of Kutch with some part by the little Rann, in the west and south by the Arabian Sea and in the South-East by the Gulf of Khambhat; while in the east is the Mainland of Gujarat and are shown in Figure 1,8,9,19 .The area covered by Saurashtra region is 59,360 sq.km. of which 9000 sq.km.area is under study 20 .Suarashtra basin lies between latitude 20°N to 24°N and longitude 69°E to 73°E.The rivers of Saurashtra region under study are: Wadhavan-Bhogavo, Limbdi-Bhogavo, Sukhbhadar, Utavali, Khalkhalia, Padalio, Keri, Ghelo and Kalubhar.There are 20 dams situated in these river basins.Details of river basins and dam situated in these river basins are shown in Table 1 and 4. Basin maps with dam site are shown in Figure 3 to 10.
There are 13 rain gauge stations and 9 G&D stations in these river basins which are shown in Figure 2. The rainfall data are collected from IMD as well as Kalpasar Department and G&D data are collected from Kalpasar Department of Gujarat.Details of G&D stations and raingauge stations are shown in Table 2 and 3.For Synthetic Unit Hydrograph analysis, data related to catchment like river length, catchment area and equivalent slope are required and the same are computed using SWAT model and Arc-GIS.SRTM data of 90 m resolution are used for this purpose.

mEthODOlOGy
In this study, deterministic approach based on unit hydrograph theory and statistical approaches based on frequency analysis are used for design flood estimation.

Deterministic approach
Due to paucity of data, regional approach based on synthetic unit hydrograph developed by Central Water Commission (CWC), 1987 has been used 2 .The study area falls under the subzone 3(a).

Synthetic unit hydrograph (Suh) method
The following relationship for SUH method has been developed by CWC (1987): ...(3) W 50 = 2.284/(q p ) 1.00 ...(4) Q p =q p * A ...( 5) W 75 = 1.331/(q p ) 0.991 ...( 6) WR 50 = 0.827/(q p ) 1.023 ...( 7) WR 75 = 0.561/(q p ) 1.037 ...(8) T m =t p + 0.5 ...( 9) Where, A = Total catchment area in km 2 L = Length of longest main stream along the river course in km S c = Equivalent stream slope in m/km t p = Time from the centre of effective rainfall duration to the peak in hr.q p = Peak rate of discharge in cumec per sq.km.Q p = Peak discharge of U.G. in m 3 /s T B = Base width of U.G. in hr.T m = time from the start of rise to the peak of U.G. in hr.W 50 = Width of U.G. measured at 50% of peak discharge ordinate in hr.W 75 = Width of U.G. measured at 75% of peak discharge ordinate in hr.W R50 = Width of rising limb of U.G. measured at 50% of peak discharge ordinate in hr.W R75 = Width of rising limb of U.G. measured at 75% of peak discharge ordinate in hr.

Regional flood formulae method
The regional flood formulae have been developed by CWC to estimate 25, 50 and 100 year return period flood values.The meteorological variability has been accounted from region to region in these formulae.The others factors such as shape of the catchment, slope of the stream etc, which have influence on the peak, have also been included in these formulae thereby improving over most of the limitations of the empirical / rational formula.Thus to estimate design flood for sub-zone 3(a), Regional flood formula is given as 2 :

Statistical approach
The statistical approach, otherwise also called frequency analysis, may be performed on the past recorded data of annual peak data series.Frequency analysis is carried out on the available record of annual flood peak discharge or annual rainfall events of the region.

Frequency analysis for individual gauged sites
Frequency analysis study interprets a past record of events to predict the future probabilities of occurrence and estimate the magnitude of an event corresponding to a specific return period 1 .If the event records are of sufficient length and reliability, they may yield satisfactory estimates.The method, however, does not provide a hydrograph shape but gives only a peak discharge of known frequency.The processed data series are to be analysed to ensure that the fundamental assumption of frequency analysis are satisfied.The data series is to be checked for randomness, presence of trend and outliers.The presence of trend is tested by using Kendall's rank correlation test and Turning point test.The presence of randomness and outliers is tested by Anderson's correlogram test and Chow test respectively.Detailed at site flood frequency analysis is carried out by using various distributions like Normal, Log-Normal, Pearson type III, Log-Pearson type III, Gumbel's Extreme value distribution 9 .Gumbel EV1 is the commonly used distributions and the details about these distributions are given below 1,15,16 .

Gumbel EV-1 type distribution
It is one of the most commonly used distributions in flood frequency analysis and was introduced by Gumbel in 1941.It is widely used for extreme values in hydrologic and meteorological stud ies for prediction of flood peaks, maximum rainfalls, maximum wind speed, etc.It is the double exponential distribution (known as Gumbel's distribution or extreme value type 1 or Gumbel's EV-1 distribution).The CDF of EV-1 distribution is defined as Regional flood frequency analysis Kumar (2009), developed the Regional flood frequency relationship using L-moment approach for ungauged catchments for 17 Subzones hydro-meteorologically homogeneous.Out of 17 subzones, Saurashtra region falls under Subzone 3(a) and the relationship for this subzone developed by Kumar (2009) is given as follows 11 : Where, Q T = Flood estimate for an ungauged catchment in m 3 /s for T year return period C T = a regional coefficient A = Catchment area in km 2 b = a regional coefficient, for subzone 3(a) this value is 0.383.Value of C T for Various return period for Subzone 3(a) are shown in Table 5.

RESultS anD DiScuSSiOn
In this study initially the above approach are used for 20 dams as well as for 9 river basins on the basis of 24 hour rainfall for T year return period given in the iso-pluvial map.After rainfall frequency analysis, it is revised only for dams because these basins are heavily affected by dams situated on upstream.The result obtained by the above approach by the use of 24 hour rainfall for a T year return period given in the iso-pluvial map (IMD, Pune) are shown  8. Thus this estimated value of 24 hour rainfall for return period of 25, 50 and 100 years is used to revise design floods for the dams present in these river basins.Revised design floods for dams in these river basins for return period of 25, 50 and 100 years are computed and tabulated in Table 9 and from Table 9 it is found that the % difference is very less between revised SUH and revised RFF method.
By using the relationship developed by Kumar (2009), the design flood estimates for return period of 25, 50 and 100 years for dams and rivers are computed below in Table 10 and 11.From Table 10 and 11 it is found that the % difference is very large between L-moment and revised SUH method.L-moment method underestimates the design floods for dams as well as river basins.
The annual flood data for the nine river sites are heavily affected by the storage dams in the upstream.Hence these data violate the basic principle of virgin flow.Hence the flood frequency analysis of these data was not attempted further.

cOncluSiOnS
After the analysis of these river basins and dams situated on it, the following conclusions are drawn: • For the study area, 24 hr rainfall for the return period of 25, 50 and 100 years are different   for 9 river basins which also differs from isopluvial map recommended by IMD, Pune for this region.• Revised design floods using SUH and RFF method on the basis of estimated rainfall indicates over-estimated and under-estimated design floods.

•
Due to construction of number of dams in 9 river basins, design flood estimation on each dam by using deterministic approach is more feasible.

•
The percentage difference is very less between revised SUH and revised RFF method.So, for safety purpose one with higher value will be used.

•
Regional flood frequency relationship based on L-moment under-estimates the design floods with average percentage difference of 32.023% for dams and 28.28% for river basins.

•
The reason for large average percentage difference was investigated and the data analysis reveals that there are large storages in these basins and hence application of either RFF or L-moment based methods may not be applicable.

table 9 :
Revised design flood for t year return period by Suh and RFF methods for 1119

Table 6
and 7 as well as developed by rainfall frequency analysis for basin wise are shown in Table8 and 9. From Table6, it can be seen that design flood estimates for return period of 25, 50 and 100 years for dams namely Wadhavan-Bhogavo, Limbdi-Bhogavo, Sukhbhadar, Utavali, Khalkhalia, Padalio, Keri and Kalubhar are underestimating except Ghelo which is overestimating when compares with the result obtained from Table8.The reason behind this variation in result is the use of value T year return period 24 hour rainfall.By rainfall frequency analysis it has been found that the river basins namely Wadhavan-Bhogavo, Limbdi-Bhogavo, Sukhbhadar and Kalubhar have higher value of rainfall from what