Development and application of efficient methods for the forward propagation of epistemic uncertainty and sensitivity analysis within complex broad-scale flood risk system modelsThis article is one of a selection of papers published in this Special Issue on Hydrotechnical Engineering.

Abstract

Increasingly, an understanding of flood risk across regions and nations, and an ability to explore how these might change in time, is seen as a prerequisite to effective and efficient flood risk management. In response, specific flood risk analysis methods have been developed that are both accurate and fast to run. Although widely acknowledged as desirable, it has not previously been possible to quantify the uncertainty associated with the assessed flood probability, consequence, or risk. To help overcome this deficiency, an efficient method for the propagation of epistemic uncertainties through large-scale flood risk system models has been developed and trialed for three pilot catchments. The approach is allied to an efficient sensitivity analysis that enables the influence of individual uncertainties on the output quantity of risk to be isolated, enabling future research, development, and data-gathering efforts to be focused.

Ranking sources of uncertainty in flood damage modelling: a case study on the cost‐benefit analysis of a flood mitigation project in the Orb Delta, France

“…Nevertheless, many authors first focused on the uncertainty in a single component of the flood damage assessment chain: hydraulic modelling (e.g. Bernardara et al ., ; Gouldby et al ., ; de Rocquigny, ), inundation mapping (Bales and Wagner, ; Stephens et al ., ), damage functions (Merz et al ., ; Kutschera, ; Merz and Thieken, ; Merz et al ., ) or land use (Te Linde et al ., ). To go further, a number of recent studies investigated how combinations of these uncertainty sources interact and propagate through flood damage assessments.…”

Believe it or not? The challenge of validating large scale probabilistic risk models

“…Inaccuracies in data (introduced, for example, through inaccuracies in measured defence crest levels or the land use in at-risk areas) and the performance of individual model components (introduced, for example, through simplifications of the breach processes) lend themselves to direct inclusion within a local or global sensitivity analysis to determine their influence on the assessment of 13 . These errors are referred to here as 'quantifiable knowledge uncertainty'.…”

“…A number of studies have therefore been undertaken to explore the potential sources of error within NaFRA 12, 13,14,15,16 . These studies attempt to quantify uncertainties that may have a significant influence on the credibility of the NaFRA results including model structure errors (such as the lack of interaction between the floodplain and river channel), model component uncertainties (reflecting the simplification of the floodplain process used) and data uncertainties (particularly around basic data such as crest level and the location and type of property that maybe exposed to flood waters).…”

A flood risk analysis model with topographical inundation and life-loss

“…Approaches have been developed that seek to quantify some of these other sources (e.g. Apel et al, 2004;De Moel and Aerts, 2011;Gouldby et al, 2010;Merz and Thieken, 2009). These approaches are not commonly applied in practice, primarily due to the computational burden associated with the implementation of the methods.…”