University of Twente Student Theses


Evaluating methods to assess the coastal flood hazard on a global scale : a comparative analysis between the Bathtub approach and the LISFLOOD-AC model

Bootsma, J. (2022) Evaluating methods to assess the coastal flood hazard on a global scale : a comparative analysis between the Bathtub approach and the LISFLOOD-AC model.

[img] PDF
Abstract:The already present global coastal flood risk is a threat to population, infrastructure and coastal ecosystems (Vafeidis et al., 2008). Due to changes in storm patterns and sea level rise, a significant increase in coastal flood hazard is expected (Church & Gregory, 2019; Forzieri et al., 2016). Since hazard estimation is a necessary precursor to risk assessment, coastal flood hazard assessment at large spatial scales is needed. This study investigates global inundation estimations from the Bathtub approach, and critically compares the approach and results to the outcomes of the more detailed, process-based, reduced complexity model LISFLOOD-AC. The LISFLOOD-AC model used for coastal flood inundation estimates on a global scale is a relatively computationally expensive model. The commonly used Bathtub approach does not have this limitation, although this model can overestimate the flood magnitude (Vousdoukas et al., 2016). This study investigates whether the computational time of models can be reduced without compromising on quality of results. The spatial domain of this study involves the entire stretch of coastline around the World. Both models were compared for the baseline period 1980-2014. The model scenarios included span return periods of extreme sea levels from 5 to 200 years. Identifying differences in estimated flood extent was the main task in this project. The flooded area estimates of both modelling approaches were subdivided based on administrative boundaries, coastal typologies and terrain classes. The outcomes were compared as flooded area in (km2), flooded area normalized by shoreline length in (km2/km) and as factor difference (Bathtub / LISFLOOD-AC). Trapezoidal numerical integration was applied to the flooded area estimates spanning all baseline return periods to compare the model outcomes as Expected Annual Flooded Areas (EAFA) in (km2/km/yr). Furthermore, the fit metrics: hit rate, false alarm rate, and critical success index wer determined for all baseline return periods. In this, the hit rate represents the flooded area correctly predicted by the Bathtub approach, the false alarm rate is a metric of overestimation and the critical success index is a metric of agreement between both modelling approaches. Results from this study show that the Bathtub modelling approach overestimates the flooded area significantly compared to the process-based reduced-complexity model LISFLOOD-AC. This holds for most countries on Earth (subquestion 1), every coastal typology (subquestion 2) and every terrain class (subquestion 3). The Bathtub approach estimates a global EAFA that is 5.9 times larger than the estimate by the LISFLOOD-AC model. Furthermore, this overestimation was observed for all baseline return periods included, although the factor difference reduced when moving to higher return periods of extreme sea levels. Smaller differences were observed for steeper coastlines, for which the influence of the topography is larger. This results in less overestimation by the Bathtub approach. The propagation of the coastal flood is only limited by the topography in the Bathtub approach, while the LISFLOOD-AC model also incorporates the effect of landscape roughness. Therefore, in flat terrains like plains, the flood propagation in the Bathtub approach experiences no deceleration which causes significant extents of the flood. This especially holds in combination with the fact that the Bathtub approach does not include conservation of mass and therefore assumes an unlimited amount of water that can propagate inland. The hit rates reveal that the Bathtub approach generally estimates flooding in areas where the LISFLOOD-AC model does as well. However, from the false alarm rates it is clearly observed that the Bathtub approach significantly overestimates the flooded area. Lastly, the critical success index values are generally low, representing an insufficient agreement between the flooded area estimates by both modelling approaches. From the results of this study, it can be stated that the Bathtub approach should never be used in any serious assessment of the coastal flood hazard on large spatial scales.
Item Type:Essay (Master)
Faculty:ET: Engineering Technology
Subject:56 civil engineering
Programme:Civil Engineering and Management MSc (60026)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page