PubAg

Main content area

Modelling coastal flood vulnerability: Does spatially-distributed friction improve the prediction of flood extent?

Author:
Seenath, Avidesh
Source:
Applied geography 2015 v.64 pp. 97-107
ISSN:
0143-6228
Subject:
Landsat, friction, geographic information systems, geography, hydrodynamics, land cover, model validation, models, prediction, storms, tides, topography, Trinidad and Tobago
Abstract:
This paper examines whether the application of spatially-distributed versus static friction in hydrodynamic modelling increases the accuracy of predicted coastal flood extent using Pigeon Point, southwest Tobago, as a case in point. A two-dimensional hydrodynamic flood model is created from acquired and surveyed bathymetric, topographic and tidal data via the LISFLOOD-FP model code. Using a Landsat 8 image of the study area, a Maximum Likelihood (ML) supervised classification was performed to distinguish different land cover classes within the study site. The classified Landsat 8 image was further processed by assigning friction values to each land cover class to create a spatially-distributed friction file in ASCII format for use in LISFLOOD-FP. Using the flood model developed, simultaneous simulations were performed to assess the impact of storm surges (varying levels) on coastal flood extent at Pigeon Point utilising a static friction value, which broadly defined the area (i.e., 0.02), and the spatially-distributed friction file generated. Model outputs were compared to determine the extent of difference in flood prediction obtained from the application of static versus spatially-distributed friction through a Geographic Information System (GIS) based analysis. The flood model developed was subsequently applied to simulate an observed spring tide event using both static and spatially-distributed friction value(s) defined and model performance in each case was evaluated using the Root Mean Squared Error (RMSE) approach. Collated results indicated that using spatially-distributed over static friction do not increase accuracy of predicted coastal inundation extent, nor improve model performance. However, it appears to provide more insight on flood timings, which can be useful for coastal management.
Agid:
5407129