Epistemic uncertainties and natural hazard risk assessment – Part 2: Different natural hazard areas

Abstract

This paper discusses how epistemic uncertainties are considered in a number of different natural hazard areas including floods, landslides and debris flows, dam safety, droughts, earthquakes, tsunamis, volcanic ash clouds and pyroclastic flows, and wind storms. In each case it is common practice to treat most uncertainties in the form of 5 aleatory probability distributions but this may lead to an underestimation of the resulting uncertainties in assessing the hazard, consequences and risk. It is suggested that such analyses might be usefully extended by looking at different scenarios of assumptions about sources of epistemic uncertainty, with a view to reducing the element of surprise in future hazard occurrences. Since every analysis is necessarily conditional 10 on the assumptions made about the nature of sources of epistemic uncertainty it is also important to follow the guidelines for good practice suggested in the companion Part 1 by setting out those assumptions in a condition tree.

Assessing future vent opening locations at the Somma‐Vesuvio volcanic complex: 2. Probability maps of the caldera for a future Plinian/sub‐Plinian event with uncertainty quantification

“…While a strong dichotomy between uncertainty on the knowledge of the system (epistemic) and uncertainty on its future behavior (physical/ aleatoric) is not universally accepted, it is a simplifying, pragmatic paradigm adopted in many natural hazard assessments [e.g., Marzocchi and Bebbington, 2012;Selva et al, 2012]-most notably in probabilistic seismic hazard assessment, where there is an extensive and exhaustive literature. For recent reviews in relation to several natural hazards, see, e.g., Beven et al [2015Beven et al [ , 2016. We adopt these two main classes because they are consistent with the "doubly stochastic modeling" concept which comes from statistics literature [e.g., Cox and Isham, 1980;Daley and Vere-Jones, 2008;Harte and Vere-Jones, 2005] and has been recently applied to volcanology [e.g., Jaquet et al, 2008Jaquet et al, , 2012.…”

A Vulnerability‐Based, Bottom‐up Assessment of Future Riverine Flood Risk Using a Modified Peaks‐Over‐Threshold Approach and a Physically Based Hydrologic Model

“…Accurately simulating the physical processes governing intense small-scale convective precipitation in climate models is a challenging task that exceeds the prediction skill of many existing models (Dai, 2006;Kendon et al, 2012;Kysely et al, 2015;Pritchard et al, 2011). As discussed previously, some of the summer risk is also related to tropical moisture exports, sometimes in the form of tropical cyclones (Lu & Lall, 2017;, 2016. Warmer tropical Atlantic sea-surface temperatures should facilitate increases in the frequency of land-falling tropical cyclones in the Northeast US (Emanuel, 2005), though this result is not consistently simulated by CMIP5 GCMs (Wuebbles et al, 2014).…”

“…The method is applicable where the occurrence of hazard exceedances over a decision‐relevant threshold is sufficient to characterize vulnerability. Our approach relates causal meteorological processes to local catchment flooding (e.g., Merz et al, ), is based on public interest in flood risk (Botzen et al, ), is applicable where any risk exceedance is deemed not acceptable (e.g., Srinivasan & Rethinaraj, ), is simplistic in its presentation of risk as recommended by Pappenberger et al (), and is in line with the recommendations of Beven et al () to acknowledge assumptions and uncertainties.…”

“…A second problem stems from the difficultly of incorporating emerging hydroclimate information into conventional risk analysis for long-term planning. It is well established that simple extrapolations of flood frequency curves beyond gage records can underestimate flood hazard if instrumental records are too short and exclude past extreme events (Greenbaum et al, 2014;Halbert et al, 2016;Shin et al, 2015;Yan & Moradkhani, 2015, 2016. There is a well-developed literature on methods to reduce estimation uncertainty by pooling gages of variable record length (Haddad et al, 2012;Renard, 2011;Stedinger & Tasker, 1985) and leveraging paleoflood evidence or historical written records of past floods (Kjeldsen et al, 2014;Mei et al, 2016;Stedinger & Cohn, 1986).…”