Application of GIS in flood hazard mapping: A case study of Gangetic West Bengal, India
3.2. A closer look of the high hazard zone: A sub-regional study
The development block level regional hazard map is not capable of revealing the hazard scenario in adequate detail. This paper argues that a detailed large scale hazard mapping in suitable and cost-effective when it deals with maximum risk zone. The regional study leads the way to identify the high risk zone. For detailed sub-regional analysis very highly hazardous blocks in Jalangi River Basin has been considered. The study area constitutes of entire Domkal, Hariharpara, Noada and Tehatta-I, Tehatta-II blocks and part of Raninagar-I, Berhampur, Beldanga-I, Beldanga-II and Kaligunj blocks. Although Tehatta-II block was not classified in very high hazard category in Fig 2 it has been considered within the study area to maintain geographic contiguity between Kaligunj and Tehatta-I blocks. Revenue village is the smallest administrative unit of West Bengal. It is also the highest spatial resolution of census data collection in India. Therefore, revenue villages of the above mentioned development blocks have been chosen as the unit of inquiry for the sub-regional study. 3.2.1. Indicators of flood hazard at sub-regional scale: Flood hazard index for this village level study has been devised on the same line as the regional study but the sources of data vary to some extent. Three indicators of flood hazard have been chosen. Number of flood occurrence in each village for last 10 years has been considered as the measure of flood proneness. Irrigation and Waterways Department of West Bengal Government prepares maps annually to show the flood affected areas each year. These maps are not uniform in scale. The maps used for this study vary in scale from 1:250,000 to 1:2,000,000. Heterogeneity in the source map scale is one of the limitations of this study but considering the maximum spatial resolution of this investigation small scale maps have also been used. We named the indicator as ‘flood-fqr’. Population density per hectare has been taken into consideration as the indicator showing intensity of land use. This indicator is named as ‘pop’. The third indicator takes into account the aspect of flood emergency management. During the time of inundation affected population are required to be evacuated to a safe place for temporary shelter. Relatively higher ground, not likely to be submerged by flood water, can serve as the temporary flood shelter. Availability of such ‘higher’ land in each revenue village has been calculated from Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) digital elevation model (DEM). If the maximum elevation of a village is below a critical threshold the population is assumed to have no access to a flood shelter during contingency. These villages have been considered as under potential threat of flood and associated hazard. This indicator is given the mane ‘shelter’. 3.2.2. Preparation of data: Boundary of the revenue villages have been derived from individual Development Block maps. Scale of the map was 1:63360. Each block map has been registered to geographical coordinate system using a digitizing tablet. Then the village boundaries of each block have been digitized as polygon shape file. These shape files has been projected into UTM Projection, zone 45N, using Project Wizard of ArcToolbox and then integrated into a single shape file using the merge function in Geoprocessing Wizard of ArcMap. A total of 384 villages have been considered. Like the previous exercise a unique ID has been assigned to each of the polygon that represents a revenue village. All the Irrigation and Waterways Department’s maps have been registered to geographical coordinate using ArcInfo software. Flooded areas for year 1991, 1993, 1995, 1996, 1998, 1999 and 2000 have been vectorized in individual shape files as polygon using ArcMap and ArcCatalog. Later these layers have been projected into UTM Projection (Zone 45N) to conform the village level map of the sub-regional study area. Selection function of ArcMap was used to select polygons from the village layer that intersect with the above mentioned flooded area polygons. This process calculates the number of flood occurrence for each village and the result is tabulated under the indicator ‘flood-fqr’. Population density figure for each village has been derived from dividing the total population by the area. Source of the data was Village Directory of Census India, 1991.Same unique ID, as assigned to the polygons in the village shape file, have been given to the population data to make it compatible with the GIS layer. ASTER Relative DEM has been used to extract the highest point for each village. ASTER Relative DEM has a spatial resolution of 30m. This digital data is suitable to meet 1:50,000 to 1:250,000 map accuracy standard (United States Geological Survey, 2003).The village boundary layer has been overlaid to the DEM and highest elevation for each polygon has been extracted using Zonal Statistics function of ArcInfo Spatial Analyst. Unlike the regional analysis a multiplicative model has been adopted for creating a composite flood hazard map in sub-regional scale. Villages have been assigned rank for each of the 3 hazard indicators. Nature of the indicators, especially ‘shelter’, does not allow application of any statistical treatment to the original data set for creating composite flood hazard index. Hazard ranks are commonly integrated into a multiplicative model to create a composite hazard index (Islam et al 2000b). The ranking scheme for first two flood hazard indicators is presented in Table 2, 3 and 4. Knowledge based hazard ranking of the selected hazard indicators for the sub-regional study
Villages that never experienced inundation in last 10 years have been assigned a hazard rank of 0 in Table. 2. This method has been adopted to ensure that these villages get a composite hazard index of o in the multiplicative model. In Table 3 the revenue villages exhibiting a population density of 0 are mostly marshy land and abandoned river channels. These villages have been assigned a hazard rank of .25 so that they acquire a smaller composite hazard index value. Kamalpur village, showing a population density of 3091person per hectare, has been given a hazard rank of 6. Excluding the value 0 and 3091 rest of the series has been subdivided into 4 quartiles and given hazard rank from 1 to 5. Assigning hazard rank to indicator ‘shelter’ requires a detailed knowledge of the local topography. Our main objective is to roughly identify the critical elevation above which flood water is not likely to submerge the ground. Villages having their highest point below this critical elevation threshold are subject to high hazard rank. To determine the break of slope or the critical elevation value a number of transverse profiles have been drawn across Jalangi River using the ASTER digital elevation mode. One such transect profile is shown in Figure 3 whose location is marked in Figure 4.  Figure: 3 A scheme of hazard ranking for indicator ‘shelter’ has been devised after analyzing a series of transverse profiles across Jalangi River. Details of the scheme are tabulated in table: 4. Final flood hazard index for sub-regional scale is created as follows: Flood Hazard = (R_fld-fqr × R_pop × R_shelter) Since the data ranges are not very familiar it has been classified into 4 hazard categories by natural break (Jenks) scheme. In this process ArcMap identifies break points by identifying inherent clustering pattern of the data. Class boundaries are set where there are relatively big jump in the data values (Minami, 2000). Details of the classification scheme are shown in Table 5.
 Figure: 4. Flood Hazard map prepared by village level sub-regional scale study. Inset showing location of the sub-regional study area within the entire study region Figure 4 exhibits that there is no defined pattern in the disposition of flood hazard zones. Contrary to the overall topographic configuration moderate and high hazard zones are not necessarily located very near to Jalangi River. In the NE side of the area a cluster of villages depict high to very high hazard situation. More or less eastern side of Jalangi River is more hazard prone than the western part. Most probably the higher western bank of the river and natural levees prevent the flood water to spill its right bank very frequently. In the central portion of the area a few villages show moderate to high hazard potential by virtue of their higher population density and low road density. It should be noted that the reliability of this hazard map is more where the revenue village sizes are small. It is obvious that the larger villages mask out the finer details and reduce the spatial resolution of the map. The village boundaries have been derived from 1: 63360 maps but inclusion of the flood related data from small scale maps makes it unsuitable for large scale representation. The village level hazard map is recommended to be presented at 1: 250,000 scale. 4. Conclusion: This study shows a simple and cost effective way to use geographical information system for creating flood hazard map from the available data base. It is acknowledged that accuracy of the key information, past records of flooding, depends upon the scale of the map that represents them. Due to lack of access to government resource we could not lay our hand on the large scale flood maps for the early 1990s. Incorporation of those resources would definitely enhance accuracy of the analysis. The weightage schemes are also open to suggestion and further improvement. Acknowledgement We are thankful to Mr. Kamal Pal of Riddhi Management Pvt. Ltd. (www.riddhi.org) for providing us some important data set and logistic support for the project. We also express our thanks to the Irrigation Department of West Bengal Government, India, for allowing access to consult annual flood reports of the state. Reference
Page 3 of 3
| Previous | |