Flood Inundation Mapping along the Lower Reach of Kelani River Basin under the Impact of Climatic Change

The downstream low lying region of the Kelani River including the Colombo suburbs, experience severe inundation due to heavy rainfalls in the upper basin of the Kelani River. Occurrence of heavy rainfalls is expected to be more frequent in the tropics with the impact of climatic change (IPCC, 2007). Therefore, understanding future rainfall intensity in the river basin and inundation in the low lying region along the lower reach of the Kelani River is extremely important as this is a region with a high population density and economic activities in the suburbs of the capital. The present study analyses the potential extreme rainfalls and resulting flood inundation along the lower Kelani River. Coarse grid atmospheric parameters provided by Global Climate Model (GCM) models for A2 (high emission scenario) and B2 (low emission scenario) scenarios of Intergovernmental Panel on Climate Change (IPCC, 2007) were downscaled to local scale by applying Statistical Downscaling Model (SDSM). Flood discharge and inundation along the Kelani River reach below Hanwella were analyzed by applying two-dimensional flood simulation model (FLO-2D). Inflow to the model at Hanwella, is estimated by the Hydrologic Engineering Center – Hydrologic Modeling System (HEC-HMS) model under future extreme rainfall events. Areas vulnerable to inundation under the above climatic change scenarios are presented.


Introduction
Climate change is defined as statistically significant variation in either mean state of the climate or in its variability, persisting for an extended period (typically decades or longer).It may be due to natural internal processes or external forcing or to persistent anthropogenic changes in the composition of the atmosphere or in land use [15].Climate change directly affects precipitation and temperature.High intense rainfalls, frequent and prolonged droughts are some examples.It has been predicted that the implications of climatic change on Sri Lanka are variations in rainfall patterns, sea level rise, and temperature [3].
Two monsoon systems, the Southwest (May-Sep) and the Northeast (Nov-Feb) and development of extreme low pressure conditions in the Bay of Bengal also have direct impacts on the rainfall patterns in Sri Lanka.Anomalously high seasonal precipitation typically associated with La Nina phenomenon and cyclonic storms which originate from the Bay of Bengal are usually the main reasons for devastating floods in the island.
This study discussed the frequencies of high intensity precipitation with their inundation extents in the Kelani River basin, produced under IPCC Special Report on Emissions Scenarios (SRES) A2 and B2 scenarios [16].Both A2 and B2 emphasize on rapidly growing selfreliant nations while B2 accounts for more ecologically friendly growth.If extreme precipitation events are becoming more frequent in the 21 st century as manifested by ever changing global climate system, the Kelani River basin is one of the most vulnerable river basins for floods and costly flood damages since the river flows through the commercial capital [12].

Study Area
The Kelani River basin is located in between Northern latitudes 6° 47' to 7° 05' and Eastern longitudes 79° 52' to 80° 13'.The basin area is about 2230 km 2 with two distinct types, as the upper basin is mountainous and the lower basin which is below Hanwella, has plain features.
The basin receives about 2400 mm of average annual rainfall and carries a peak flow of about 800-1500 m  [18], while more than 78,000 families were affected during the 2010 flood [17].Moreover, Irrigation Department records show that there were two consecutive severe floods which occurred during the year 2008.

Preparation of data
Topographic data of the basin were obtained from Shuttle Radar Topographic Mission (SRTM) data as Digital Elevation Model (DEM).
The SRTM-DEM data, with a horizontal resolution of approximately 90 m near the equator and a vertical resolution of 1 m, constitutes the finest resolution and most accurate topographic data available for most of the globe [11].According to a study carried out in the Ruhuna basin, the most useful data were the SRTM DEM at 90m resolution.The slope map generated using the SRTM DEM was very useful to identify low lying areas [9].SRTM DEM has also been used in the hydrological modeling of upper Kotmale basin in Sri Lanka [6].
Inflow hydrograph at Hanwella was generated by HEC-HMS and the rainfall of the lower basin was generated by Statistical DownScaling Model (SDSM).Land use data was from the Department of Survey, Sri Lanka at 1:50,000 scale.

Rainfall Modeling
Statistical Downscaling Model (SDSM) which is designed to downscale GCM data into regional level, was applied to downscale precipitation for the basin up to 2099 under A2 and B2 scenarios published by IPCC.The model was calibrated for the period from 1961 to 1975 and validated from 1976 to 1990.SDSM uses linear regression techniques between predictor (observed large scale climate fields) and predictand (local observed meteorological variables) to produce multiple realizations of synthetic daily weather sequences.The predictor variables provide daily information about large scale atmosphere condition, while the predictand describes the condition at the site level [2].
Accordingly in order to ensure that the downscaling method in reproducing the mean and variability of observed variable in the basin, an uncertainty analysis was carried out.Table 1 shows the statistical parameters of annual maximum daily rainfall for simulated and observed data during the periods of calibration (1961)(1962)(1963)(1964)(1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975) and validation (1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990).The upper basin is about 1740 km 2 and the time of concentration from upper basin which is from the "Samanala" mountain to Hanwella is about 2 days and at the edge of the lower basin it is about 2.5 days.This could be due to substantial losses because of infiltration and depression storage, within the upper basin which has a vegetation cover of about 70%.Frequency analysis of return period was carried out for 3 day total rainfall of the upper basin by using forecasted data from 2020 to 2099.
The lower basin has an area of about 500 km 2 with plain features and ample amount of builtup areas.The time of concentration of lower basin at the edge of the basin is around 12 hours.Therefore, one day rainfall in the lower basin can cause severe floods.Therefore, the daily rainfall forecasted from 2020 to 2099 was used for the frequency analysis within the lower basin.
Accordingly the flood and inundation analysis in the lower basin was carried out using 3 day rainfall events of 50 year and 100 year return periods of upper basin and daily rainfall events of 50 year and 100 year return periods of lower basin.Table 2 and 3 illustrate the summary of results of frequency analysis for upper and lower basins respectively by applying the Gumbel distribution [4].

2D flow modeling
Two-dimensional flood simulation model, FLO-2D was utilized to analyse and map inundation areas.FLO-2D is two-dimensional flood routing model that distributes a flood hydrograph over a system of square grid elements.FLO-2D numerically routes a flood hydrograph while predicting the area of inundation and simulating flood wave attenuation [13].
The FLO-2D system consists of processor programs to facilitate graphical editing and mapping that simulate channel and floodplain details.The Grid Developer System (GDS) generates a grid system that represents the topography as a series of small tiles.The FLO-2D model has components for rainfall, channel FLO-2D moves the flood volume around on a series of tiles for overland flow or through stream segments for channel routing.Flood wave progression over the flow domain is controlled by topography and resistance to flow.Flood routing in two dimensions is accomplished through a numerical integration of the continuity equation and the equations of motion [14].
-Flow depth -Depth-averaged velocity -Excess rainfall intensity -Friction slope -Bed slope The differential form of the continuity and momentum equations in the FLO-2D model is solved with a central, finite difference numerical scheme as one dimensional flow in the channel and two dimensional flow in the flood plain.

Model calibration
The GDS was used to generate 250 m x 250 m grids covering the river basin.The topography of the basin area was determined by using the downloaded DEM from SRTM.Manning's coefficients over the basin were allocated according to the land use patterns Normalized Objective Function , Nash-Sutcliffe efficiency and percentage bias given in equations ( 3) to ( 5) respectively, were used to evaluate the Goodness of Fit between the simulated and observed flows.4 shows the Goodness of Fit of the simulation results for selected flood events.The Good discharge a and the inu show good Accordingly future forec  Further, inundation maps could be published in public places along with awareness programs which would guide people to find safe places during severe floods.

Conclusions
2-D inundation model has been developed to lower Kelani basin to estimate inundation under extreme rainfall to the basin.
Rainfall under climatic change scenarios, A2 and B2 were downscaled from GCM shows that the 50 year return period daily rainfall was 244 mm and 227 mm under A2 and B2 scenarios respectively and 100 year return period daily rainfall was 268 mm and 248 mm under A2 and B2 scenarios respectively at lower basin.The 50 year return period 3 day rainfall was 415 mm and 393 mm under A2 and B2 scenarios respectively and 100 year return period 3 day rainfall was 456 mm and 432 mm under A2 and B2 scenarios at upper basin.Some areas in Hanwella, Kaduwela, Kolonnawa, Biyagama, Kelaniya and Colombo DS divisions are identified as vulnerable areas for inundation under rainfalls of 50 year return period.The inundation extents under 100 year return period rainfall expands to Kelaniya, Thimbirigasyaya and Sri Jayawardanapura Kotte GN divisions too.
The findings are useful to plan implementation strategies to minimize flood damages in the lower Kelani river basin under expected climatic change impacts.

Figures 1 (Figure 1 -
Figures 1(a) and 1(b) show the uncertainty of forecasted annual maximum daily rainfall with observed values for calibration and validation periods respectively, for lower basin.The simulation results for validation period under A2 and B2 scenarios show good correlation with observed data flow, infiltration, levees and other physical features[14].
. The model was calibrated for the extreme event which occurred in November 2005, by comparing the flow at Nagalagam Street gauging station.The validation was carried out for the discharge at Nagalagam Street by using the events which occurred in April 2008, May 2008 and May 2010. Figure 2 and 3 show the observed and simulated hydrographs for 2005 flood event and 2010 flood event respectively.The observed data during 2010 flood event, the rising and falling limbs show some oscillations and which could be due to external factors.

Figure 2 -Figure 3 -
Figure 2 -Time series of observed and simulated flow at Nagalagam Street during 2005 flood

Figure 4
Figure 4 -I in May 201 Figure 5rainfall of basin and period in th