2D Flood Simulation and Development of Flood Hazard Map by using Hydraulic Model

A flood is an overtopping of water from channel banks which submerges land which is usually dry. In India, floods in Godavari and Sabari River are observed several times but 2006 flood was one of the severe events which affected Khammam district extensively. The present paper demonstrates the analysis of severe flood of Khammam using hydraulic modeling. For this purpose HEC-RAS 2D a Hydraulic model was used to simulate the behavior of study area which covers part of Godavari River and Sabari River in Telangana and Andhra Pradesh states. Model provides detail temporal simulated parameters like depth of water, water elevation with respect to M.S.L and Velocity of flowing water. SRTM 30m posting, river discharge data, LULC for assigning Roughness coefficient was used as input to the model. HEC-RAS Simulated flood results were validated with remotely sensed Radarsat satellite observations. Performance of HEC-RAS model was evaluated according to the criteria of measure of fitting raster cells with satellite data. Validation results showed that performance of HEC-RAS model is very effective and can be used by hydrologist or water resources engineers for planning and development.


Introduction
Floods are one of the major natural disasters affecting the South-Asian region. India and Bangladesh are two from eight South-Asian countries which are reported to be worst affected countries in the world, accounting to be 1/5 th of global death count due to floods (Agarwal and Sunita, 1991). In India, nearly 40 million of land is prone to floods. Floods are frequent phenomena in the country occurring during monsoon season (June-October), which affects to crop lands, infrastructure as well as lives.
Repetitive flood inundation is threatening human life and property which indeed requires effective flood risk assessment (Matgen et al., 2007). Prediction of flood inundation is not straightforward since the flood inundation extent is highly dependent on topography and it changes with time. Flood prediction is a very complex process in both spatial and temporal contexts whereas the Conventional engineering methods is time consuming. Application of the hydraulic numerical modelling for flood analysis and flood plain management is a strategic and essential tool for an integrated flood plain management Research Article (Cheung, 2003;Chang, 2000). Flood risk assessment and management are fundamental steps for identifying prone risk areas, current hazards, and reducing them in future flood events (Ranzi, 2011). To propose measures on flood management it needs basic understanding of flood analysis. There are two approaches of understanding, first is, in situ flood observations (Hagen and Lu, 2011) but in situ observation are not available all the time, so another one is observing flood by Remote sensing satellite data (Haq et al., 2012;Chormanski et al., 2011) but it still limits us to know the temporal behavior in flood event. So, in order to develop flood hazard and risk zone maps and to know the temporal behaviour of flood event it is essential to simulate the flood inundation by numerical models. The studies of Flood inundation modelling using hydrodynamic models approach for designing river engineering and irrigation schemes and mapping flood risks has been carried out by various researchers worldwide (Werner, 2004;Bates et al., 2005;Patro et al., 2009)

Data Used
Spatial Hydraulic modeling requires data of two type's Geographical data and Hydraulic data.

HEC-RAS Model Setup
The projected DEM is basic input to hydraulic modeling. Model converts DEM into terrain in the form of Triangular Irregular Network (TIN) file. TIN file is a vector representation of DEM Figure 2 shows terrain created from DEM. Further, study area was defined by 2D closed polygon. The 2D polygon Perur Konta Kunavaram area covers probable areas that are liable to inundate. This area can be defined by two methods in order to minimize optimization of 2D Mesh cells. First method is to define 2D polygon area based on previous floods extend observed by satellite. Another method is to define polygon by digitizing the 2D polygon boundary connecting higher elevated areas like hills and mountains. For present study, second method was adopted as high elevated hills are surrounding to the study area. After Defining 2D Polygon area computational cells area generated. The cells are rectangular in shape and at boundary it may be rectangular or polygon up to 8 sides. Cell size was kept to be 30 m in order to enclose DEM cell size. Figure 3 shows 2D polygon with computational cells, these cells are also called as Mesh.
Further, three boundary condition assigned to study area. Two inflow hydrograph boundary condition and one normal depth boundary condition. Inflow hydrograph one at Perur gauge site on Godavari River and one at Konta gauge site on Sabari River. Figure 4 shows the assigned boundary condition to study area. Similarly Inflow hydrograph at Perur and Konta gauge site for the event obtained from CWCis shown in Figure 5. The normal depth 0.00032 slope was assigned according to the topographic conditions of the study area. Where, p and q are the specific flow in the x and y directions (m 2 /sec), n is the manning's resistance, s is the surface elevation in (meters), h is the water depth (meters), g is gravitational acceleration (m/sec 2 ), ρ is the density of water (kg/m 3 ), τ xx, τ yy, τ xy are the components of effective shear stress and f is the Corollis(/sec).

Model Performance
Model Performance has been evaluated based on measures of fit F 1 and F 2 (MoyaQuiroga et al., 2015;Horrit et al., 2007;Di Baldasarre et al., 2009). It depends on the simulated raster cells that are fitted with the satellite image. It gives the degree of accuracy of the model, F 1 ranging from 0 to 1 and F 2 ranging from -1 to 1. The equation of F 1 and F 2 is given in the equation 3 and 4 respectively.
Where, A is correctly predicted cells, B is Over-predicted cells and C is under predicted cells. The value closer towards 1 indicates model performance is better.

Flood Hazard Map
The flood Hazard was developed for the study area according to the simulated flood depth in inundated areas. The hazard classification was assigned according to the Japan ministry of land infrastructure and transport (MLIT) shown in Table 1. hazard (depth less than 0.5 meter) is entitled to be very low as people can evacuate easily on their feet. H 2 hazard (depth 0.5-1 meter) is entitled to be very low, in this zone, evacuation becomes difficult for adults and infants, animals may get exposed to hazard. H 3 hazard (depth 1-2 meters) is a medium zone, where people may get drowned but they will be safe in their homes having plinth level to be 0.6 to 1 meter. H 4 hazard (depth 2-5 meters) is entitled to high hazard zone where all people in this zone are not safe in their homes but at the most they may be safe on their roofs. H 5 hazard (depth greater than 5 meters) is an extreme hazard zone where people are not safe even on their roofs.

Results and Discussion
The HEC-RAS model results are in two decimal floating point raster stored in tiff file format. The simulated results are in the form of flood inundation with depth, velocity, water surface elevation with respect to time. Results for desired date and time can be achieved within the simulated period such as 04 Aug 2006 12.00 hours to 10 Aug 2006 24.00 hours at one hour time interval. For present study, simulated results for 07 Aug 2006 at 08.00 hours was taken into consideration for validating with RADARSAT satellite image acquired on same date and time Figure 6 shows the RADARSAT image and its derived flood extent. The evaluated measures of fit F 1 and F 2 by using the GIS tools like Arc-GIS and ERDAS shown in Table 2. Since F 1 and F 2 are closer to 1 indicates the accuracy of model performance which tends to be better and the ratio of flood simulated to flood observed is 0.96 this means 96 percent of inundated area is matched. The Figure 8 shows the variation of inundation with respect to time. The flood hazard Map was developed by considering the extreme inundation (maximum depth) simulated by the model. The hazard classification was done based on Japan ministry of land infrastructure and transport (MLIT) shown in Table 1. Figure 9 shows the flood hazard map for study area.
The depth is classified into five classes from H 1 to H 5 shown in Figure 9. Many villages on east and west side of Sabari river and north and south side of Godavari river got affected.

Conclusion
The HEC-RAS model performance shows better performance when compared to RADARSAT satellite image. Inundation increases with increases in discharge values as shown in Figure 2. The villages falling in extreme category H 5 are Chidumurum, Chatti, Kummur, Bojaraigudem, Jallagudem, Markandeyulapeta, Tallagudem, Regulapadu, Abhicherla, Kuturgutta, Kuturu, Muluru, Bhagvanpuram, Repaka, Peddarukur, Ravigudem, Chinnaruku, Gundugudem, Chintharajupalle, Waddegudem, S. Kothagudem on the either side of Sabari River and Rekapalle, Pocharam, Pochavaram, Kukunoor, Tipurapendamedu, Gommuru, Morampalle, Nagineprolu, Dantenam, Tekulagudem, Peddagollagudem, Ramachandrapuram forest, Thathakur, Dacharam and Tirumalapuram are on Either side of Godavari River are most vulnerable villages as depth simulated in some part of this villages are greater than 5 meters so people from this villages an event occasion are suggest to move towards higher elevation such that away from river towards northeast if village are on northern side of bank of Godavari river and move towards southeast if village are on southern side of bank of Godavari river. For Sabari river, villages lying on East side of are suggest to move towards east direction and villages lying on west side of Sabari river are suggested to move towards to the west side away from river.

List of Abbreviations
DEM -Digital Elevation Model; CWC -Central Water Commission; MLIT -Ministry of Land Infrastructure and Transport (Japan); AWiFS -Advanced Wide Field Sensor; LULC -Land Use Land Cover