There are many uncertainties in tsunami modelling; one of the key parameters for flow overland is friction. As a wave propagates over the land surface it interacts with surface features such as vegetation, buildings and water bodies, which can dissipate the energy and momentum of the flow. The friction of the land surface varies according to the surface features and Manning¿s friction coefficient n is often used to account for friction due to sub-grid features in inundation modelling. However, often a uniform value of n is applied even though it is known to vary spatially. Several studies have highlighted that the friction parameter has significant impacts on the resultant hazard maps from inundation modelling and thus implications for disaster management. This raises two questions: what is the best way to account for variability in land cover and at what resolution is this important for the production of inundation hazard maps?
To answer these questions a review of previous studies was conducted to determine n values used for different land classifications and identify the various techniques used to assign n values. Over 100 land classifications have been used in previous studies and there are four common techniques in accounting for land surface variability. These techniques are: using uniform friction (not accounting for variability); using bimodal friction (one value for the land and one for the sea); using land use/land class (LULC) data to assign friction values; and, including buildings in the topography model. It was found that in studies using LULC data there are many methods; with field data, government data or satellite data used for classification, 3 to 40 LULC classes determined and this being applied at various resolutions.
A study was undertaken to apply the four techniques to Patong Beach, Thailand and validate the results using field data from the 2004 Indian Ocean Tsunami. It was determined that the most effective method to account for land surface features is to incorporate buildings into the topography model. However, it is acknowledged that this data is very often unavailable, particularly in developing nations. There may also be a computational cost as a fine mesh is needed to capture building geometries. Thus there was a focus on the LULC approach. A method was developed to assign LULC classes by visual interpretation of satellite imagery at approximately 30 m resolution. This was verified with the inundation extent data from the actual event. This method was then applied to the production of a hazard map for the city of Pariaman, West Sumatra, Indonesia.
Presented at the 2017 International Tsunami Symposium
