Comparative assessment of the GIS based bathtub model and an enhanced bathtub model for coastal inundation

Abstract

Coastal zones are dynamic spaces where human activities and infrastructure are exposed to natural forces, climate change and extreme weather events such as storm surges. Coastal inundation is regarded as one of the most dangerous and destructive natural hazards, and while there are many studies to analyse these events, GIS based methods are limited. This research aimed at develo** a GIS based enhanced Bathtub Model (eBTM) that improves on the widely used simple Bathtub Model (sBTM) to make it more appropriate to a storm surge related coastal inundation context. The eBTM incorporates beach slope, surface roughness and instils hydrological connectivity relevant for event scale coastal flooding, unlike the sBTM which only uses topographic elevation above sea level as input. For a test site in Cape Town, South Africa, inundation levels for 3 independent scenarios were calculated using the average spring tides level, extreme sea level for a 1-in-100-year storm and two sea-level rise scenarios. Each scenario was run on both the sBTM and the eBTM developed through this study. Comparing the results, the eBTM method overall produced more conservative inundation results and also produced less disconnected areas of (unrealistic) inundation. The eBTM also produces inundation water levels relative to structures, thus showing the potential for quantifying the coastal inundation risk to infrastructure, which is of relevance in the disaster response context. Additionally, the impact of using Digital Terrain Models (DTMs) instead of Digital Surface Models (DSM) on the inundation results was tested. The use of a DSM, including buildings and other objects, showed more realistic trajectories of the inundation water moving through the model area.

Appendix: Detailed description of the eBTM model steps

Numbers refer to the processing step numbers in Fig. 4:

• 1. The input surface and user specified inundation water level were added to the Raster Calculator tool to identify areas lower than the specified inundation water level. For this study a DSM was used. The output layer was a binary layer ‘grid1’, where 0 are areas higher than the pre-set inundation water level and 1 are areas lower than the pre-set inundation water level;

• 2. Grid1 was reclassified into 2 classes where zero values were recoded to ‘NoData’ and 1 = 1. This step eliminated areas higher in elevation than the input inundation water level, as the NoData cells act as barriers, forcing inundation to go around them when using the Cost Distance tool (in step 5). Therefore this step excluded the buildings from the water pathways. The output is ‘grid2’;

• 3. Grid2 was then converted into a polygon .shp mask. The output is ‘Shp1’;

• 4. Shp1 was used to extract the DSM of the focus area, excluding buildings. The output is ‘grid3’

• 5. Grid3 was used as the input elevation model for the slope tool which calculates the steepness of each cell across the study site. The output comprises of the slope angles. The output is ‘grid4’;

• 6. Grid 4 was divided by the roughness coefficient (RC) (based on Sekovski et al. 2015), which in this case was 1, using the raster calculator. The output is ‘grid5’;

• 7. The Cost Distance tool used grid5 and the coastline (.shp line) to calculate the inundation areas and pathways which favoured water movement (Sekovski et al. 2015; Perini et al. 2016), based on the least cumulative cost route per cell. The coastline essentially instilled the hydrological connectivity in the model and served as the baseline from which water propagates inland. It therefore must intersect the input DSM. The output is ‘grid6’ (the cost raster) where the value of each cell represents the least horizontal cost distance from the coastline⁷ (Perini et al. 2016). This step included an optional output, the Backlink Raster, that can be used to show directional pathways;

• 8. The following steps required the input data to be in integer format, however, grid 6 is in floating point format. In order to preserve grid 6’s decimal values, the raster calculator was used to multiply grid 6 by 1000 (producing grid 7);

• 9. Grid 7 was converted into an integer raster, using the Integer tool (grid 8);

• 10. The Raster to Polygon tool is used to create an inundation mask ‘shp2’ in shapefile format;

• 11. The values reflected in shp2 were based on the values of each cell (from grid 8) which represented the least horizontal distance from the coastline as calculated by the Cost Distance tool i.e. it no longer reflected the elevation. Shp2 was used as a mask to extract the inundated area from the input DSM. The output is grid 9 which contains the original elevation values of the DSM;

• 12. The raster calculator is then used to subtract grid9 from the user defined initial inundation water level. The output is ‘grid10’ which gives the actual inundation depth;

• 13. The raster calculator is used to eliminate cells where the inundation water level is calculated to be negative i.e. below the DTM, so only water levels occurring on the ‘surface’ are reflected. The final output is the Inundation Depth Raster.