HYDROLOGICAL AND HYDRODYNAMIC MODELLING FOR FLOOD DAMAGE MITIGATION IN BUDHABALANG RIVER BASIN, INDIA

Abstract

Budhabalang River Basin situated in the north-east of the Peninsular India experiences frequent flooding causing a lot of damage to the property and lives of people. Odisha being a coastal state to the Bay of Bengal is always hit by cyclones causing major floods. Recently occurred floods during super cyclone October 1999, August and September 2007, June 2008, October 2009, September 2011 and Phailin in October 2013 and fresh Hudhud during October 2014 are a few examples. Keeping these in view, a study has been carried out on Budhabalang river basin for assessment of flood and its forecasting by hydrological modelling up to Govindpur outlet using HEC-HMS software and generating its flood inundation extent by hydrodynamic modelling using HEC-RAS. The study will be very much useful for planning suitable flood management measures. In this study, ArcGIS version 10.5 and HEC-GeoHMS software have been used for different geospatial processing. HEC-HMS version 4.2.1 is used for rainfall-runoff modelling with three transform methods namely Clark UH model, SCS UH model and Snyder UH model adopting exponential recession model for base flow and initial & constant loss rate model for rainfall abstractions. Out of seven major historical flood events, five events are considered for calibration whereas two extreme events are considered for validation of three transform models. The performance of the three transform models is compared during calibration as well as validation based on error functions such as the percentage errors in peak & time to peak, Nash-Sutcliffe Error (NSE) values, Mean Absolute Error and Root Mean Square Error. The validated model for the event of Oct-2013 (A) yielded NSE value of 0.923, 5.51% increase in peak, 2.5% increase in time to peak. The values of all the performance criteria computed based on observed and simulated flood hydrographs are within the desirable limits. Based on the comparisons, it is found that the Clark model has performed extremely well for simulating the observed flood events of the basin. The 1-hour 1-mm unit hydrograph based upon Clark model parameters is applied for computation of design flood hydrograph for the 100-year return period. The design flood hydrograph is estimated by convoluting the unit hydrograph with hourly rainfall excess for 100-year return period, based upon rainfall frequency analysis and critical sequencing with single bell and double bell pattern of rainfall and finally adding the baseflow. It is found that the double bell storm pattern underestimates the peak discharge by 0.97% and the peak water level by 0.38% in comparison to single bell. The time to the peak has been underestimated by11.36%. Thus, in case v of the double bell rainfall pattern, although there is a very little underestimation of peak discharge, the time to peak occurs early. Hence the design flood hydrograph obtained from Double Bell design storm pattern may be adopted as 100-year return period design flood. In the probabilistic approach, the 100-year flood is estimated based on frequency analysis of annual maximum peak flood series using Gumble distribution which is found to be the best fit distribution. The 100-year flood has been estimated to be 4236.18 and 4194.93 cumec corresponding to single bell and double bell storm patterns respectively by the deterministic approach using Unit Hydrograph techniques together with frequency analysis of rainfall whereas the probabilistic approach using flood frequency analysis of annual maximum peak flood series has given 100-year return period flood as 3522.41cumecs. The water levels of the river at Govindpur gauging site have been categorized to Warning Level and Danger Level as per Central Water Commission, New Delhi as 7.21M and 8.13M respectively. Real-time forecasting with warning schedule and lead time so established would be very much useful for the concerned Central & State Government departments for taking suitable flood management measures during the flood for minimizing the flood damages. The unsteady flow analysis is carried out using HEC-RAS version 5.06 and the model is calibrated with the observed water level at the railway bridge site downstream to Govindpur. The flood inundation map is prepared on RAS-Mapper for the flood event of Oct-2013 (A) and it is compared with the processed satellite photograph of the inundated area as on 14th Oct 2013 at 06:00 hrs obtained from the Disaster Management Support Service available on BHUVAN website. It is observed that the simulated flood inundation area closely resembles with that obtained from the BHUVAN web site. The flood inundation map for design flood hydrograph of the 100-year return period, generated with a flood water depth of 2m contour interval, would be very much useful for adopting suitable flood management measures.