FLOOD ASSESSMENT STUDIES FOR KELANI RIVER BASIN, SRI LANKA

Abstract

Kelani River Basin in Sri Lanka experiences frequent flooding which resulted in loss of lives and properties of the people in this basin. Keeping this in view, a study was taken up for Flood Assessment in Kelani River Basin, Sri Lanka up to Hanwella gauging site using Hydrologic Modelling System tool developed by HEC, USA (HEC-HMS). In the HEC-HMS version 4.2 software, various options and methods are available under each sub models. In the study, most of the available options were considered for analysis and the values of various goodness of fit criteria, available in HEC-HMS, such as Nash-Sutcliffe Model Efficiency (NSE), Percent Error in Peak, and Percent Error in Time to Peak and Percent error in Discharge Volume (Volume Deviation (Dv)) were computed. Based on these, the performance of various methods for simulating the observed hydrographs was judged. The Basin Model was selected considering the Kelani river basin up to Hanwella gauging site as a single basin for the simulation of flood hydrographs. In the Meteorological Model, Gaged weight option, available in HEC-HMS, was considered for rainfall analysis. For the Transform Model or Direct Runoff Model in HEC-HMS, Clark UH, SCS UH & Snyder UH models were considered. The Calibration (manual and automatic) and Validation of model parameters were carried out analysing the available hourly rainfall-runoff data of the four storm events observed during the monsoon seasons of the years 2017, 2014 and 2012. The Arc Map-ArcGIS version 10.3 and HEC-GeoHMS version 10.3 were also used to process the different types of spatial data required as input for the HEC-HMS model applications. From the results of the analysis, it was found that the Snyder UH model is best suited model and the Clark UH model is the second best suited for the flood assessment of Kelani river Basin whereas SCS UH model is found to be least performing model. The calibrated & validated Snyder Model yielded 0.82 of NSE value, 2.4 % of percent error in peak, 25 % of percent error in time to peak and 15 % of percent error in discharge volume. The representative unit hydrograph for the basin, derived from the Snyder’s Model, was applied to formulate the real time flood forecast for the flood event of June 2014, considering the blocks of v hourly excess rainfall available up to the period of forecast. The forecasted and observed flood hydrographs were compared considering all the blocks of excess rainfall. It was found that both the hydrographs are in close agreement. 100-year flood was estimated convoluting the 100-year design sequence of excess hourly rainfall values for the design duration with the representative unit hydrograph. The recommended value of storm duration was considered and the 100-year rainfall value was obtained from DDF curve. The hourly rainfall values were obtained using the relationship developed between 100-year rainfall for the design duration and the storm durations. The design loss rate and base flow were considered as per the recommendations. The design loss rate was applied to compute the excess rainfall whereas the design base flow was added to the design direct surface runoff hydrograph ordinates in order to compute the flood of 100-year return period. Standard Project Flood (SPF) was estimated using the representative unit hydrograph and hourly values of excess rainfall critically sequenced as per the recommended practices. The design duration was obtained as per recommended practices. The DAD analysis was carried out to develop the DAD curve which was used to compute the depth of desired duration for the basin. The relationship developed between the depth and duration was used to get the incremental values of hourly rainfall. Finally, the recommended values of loss rate and base flow were considered to get the estimate of SPF. Flood frequency analysis was carried out using Gumbel and Log-Pearson type III distributions based frequency factor approach. Those flood estimates for different return periods were compared for both the distributions. The outcome of the study would be very much beneficial for the Flood Disaster Management Authorities of Govt of Sri Lanka. It would help them for implementing various structural and non- vi structural measures in Kelani river basin to minimise losses of lives and properties due to frequent occurrence of severe flood events.