Open AccessA Bayesian network model for assessments of coastal inundation pathways and probabilities
Author(s) -
Narayan S.,
Simmonds D.,
Nicholls R.J.,
Clarke D.
Publication year - 2018
Publication title -
journal of flood risk management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.049
H-Index - 36
ISSN - 1753-318X
DOI - 10.1111/jfr3.12200
Subject(s) - flood myth , coastal flood , computer science , identification (biology) , bayesian network , environmental science , floodplain , process (computing) , hydrology (agriculture) , environmental resource management , geology , geography , oceanography , ecology , cartography , artificial intelligence , climate change , sea level rise , geotechnical engineering , archaeology , biology , operating system
Coastal flood assessments are often required to describe networks of flood sources, pathways and receptors. This can be challenging within traditional numerical modelling approaches. In this paper, we assess coastal flood plains as networks of interlinked elements using a Bayesian network (Bn) model. The Bn model describes flood pathways and estimate flood extents for different extreme events and is constructed from a quasi‐two‐dimensional Source – Pathway – Receptor (2 D SPR ) systems diagram. The Bn model is applied in Teignmouth in the UK , a coastal flood plain of typical complexity. It identifies two key flood pathways and assesses their sensitivity to changes in sea levels, beach widths and coastal defences. The process of 2 D SPR and Bn model construction helps identify gaps in flood plain understanding and description. The Bn model quantifies inundation probabilities and facilitates the rapid identification of critical pathways and elements before committing resources to further detailed analysis. The advantages, utility and limitations of the Teignmouth Bn model are discussed. The approach is transferable and can be readily applied in localscale coastal flood plains to obtain a systems‐level understanding and inform numerical modelling assumptions.