Flood hazard and vulnerability assessment in upper Gin river basin in Sri Lanka under climate change

Abstract

Floods are a frequent major disaster throughout the world, usually resulting in fatalities andmassive economic and environmental damage. Seasonal and localized flooding is one of theextremely common natural disasters in Sri Lanka. There are two monsoon seasons (Southwestand Northwest monsoon) and two inter- monsoons (First Inter and Second Inter monsoon) inSri Lanka, each of these monsoon seasons are followed by floods induced by heavy rainfall.The Southwest monsoon, which comes between May and September, has the greatest impacton the southern region of Sri Lanka. This research is developed to assess the flood hazard,vulnerability, and risk of the Thawalama watershed for climate change in future representativeconcentration pathways (RCP) 8.5. The Research methodology begins with selecting Events,which was determined by different statistical approaches. The Gumbel method was the bestfit for determining the event's return periods. A 12-year return period (2003) was selected forcalibration, and a 5-year return period (1999) was selected for validation. Further, the future climate rainfall data was bias-corrected using the linear scaling method. The future climaterainfall data was divided into two centuries: mid-century (2040-2070) and end-century (20702099). Thereafter, the 5-year Return period and 12-year Return period were estimated throughthe

Gumbel method for both mid and end centuries. The Hydrologic Engineering Centre'sHydrologic Modeling System (HEC-HMS) was calibrated for 2003 and validated for 1999 atthe gauging station of the Thawalama catchment to obtain lateral flows and inflow inside thecatchment. Thereafter, Hydrological Engineering Centre's River Analysis System (HEC-RAS)was calibrated and validated for the lateral flows and inflows obtained from HEC-HMS for2003 and 1999 respectively. Similarly, the future lateral flows and inflow were derived usingHEC-HMS by importing climatic rainfall data for selected events of 5-year and 12-year returnperiods in both mid and end centuries. Thereafter, HEC-RAS was used to get flood inundation,flood depth, and flood velocity maps. Finally, to achieve objectives, flood depth and floodvelocity were imported to the Arc-GIS interface to develop flood hazards, and populationdensity was used to develop flood vulnerability. Hence, a flood risk map was prepared bymultiplying flood hazards and flood vulnerability. The HEC-HMS was calibrated with NashSutcliffe (NSE)=0.80, Root mean square error standard deviation (RMSE st dev.) =0.40, andPercent Bias (P-bias) =17.65% and Validated with NSE=0.67, RMSE st dev.=0.60 and Pbias=15%. Thereafter, HEC-RAS was calibrated with NSE=0.66, Coefficient of determination(R²) =0.83 and P-bias=3.98% and Validated with NSE=0.62, Coefficient of determination (R²)

=0.79 and P-bias=3.28%.

The results show an increasing trend of flood inundation area forboth the 5-year return period (17.36 km²,17.40 km²,19.77 km² for years 1999,2052,2091,respectively) and 12-year return period (19.55 km², 20.06 km², 21.18 km² for years 2003, 2058,2098, respectively). Thereafter, sudden increment of flood hazard, flood vulnerability, andflood risk was obtained after mid-century in both 5-year and 12- year return periods. Almost22 Grama Niladhari Division (GND) were found to be a very high-risk category and 21 GNDwere found to be at a high-risk category at the end century of the 12-year Return period in the year 2098 whereas 19 GND were found to be a very high-risk category and 23 GND werefound to be at a high-risk category at the end century of 5-year Return period in the year 2091.Flood hazard, flood vulnerability, and flood risk is increasing suddenly after mid-century inboth 5-year and 12- year return periods. Hence, from the viewpoint of disaster reduction, theinformation derived from this study can help to estimate the probability of flood damage forthe local population.