Floods in Meso-Scale Catchments - On Processes and Projections of Floods and Flood Loss

Keller, Luise (2019). Floods in Meso-Scale Catchments - On Processes and Projections of Floods and Flood Loss. (Dissertation, Gegraphisches Institut der Universität Bern, Philosophisch-naturwissenschaftlichen Fakultät)

Text 761_Keller.pdf - Published Version Restricted to registered users only Available under License BORIS Standard License. Download (16MB)

Understanding the hydro-meteorological processes that lead to fluvial floods is important for various applications. Several approaches exist for classifying flood generating processes based on hydro-meteorological and catchment state conditions. However, these approaches rely on data of subdaily temporal resolution that may not be readily available in certain situations. This is frequently the case for example in the context of climate change impact assessments or for the analysis of historical floods. In the first part of this thesis we thus present an approach to delineate flood generating processes in meso-to macroscale catchments based on daily meteorological and catchment state conditions. The approach is developed and tested for a meso-scale, pre-alpine catchment for the period 1961 to 2014. Seven parameters were derived to describe the flood events with regard to spatio-temporal precipitation patterns, snow cover extent, snowmelt, and antecedent catchment wetness due to precipitation and snow melt. Based on these parameters, 5 storylines of flood generation were identified with a cluster analysis. These are (a) long duration, low intensity precipitation events with high precipitation depths, (b) long duration precipitation events with high precipitation depths and episodes of high intensities, (c) shorter duration events with high or (d) low precipitation intensity, respectively, and (e) rain-on-snow events. Distinct differences in the hydrological characteristics of the storylines were identified and could be linked to the storylines’ hydro-meteorological properties. This indicates that flood generating processes in meso-scale catchments can be distinguished on the basis of daily meteorological and catchment state parameters. The approach may thus facilitate an analysis of flood generating processes in meso-to macroscale catchments using e.g. climate and discharge projections, or observational data of daily temporal resolution.
The second part of this thesis makes a contribution to the assessment of uncertainties in projections of floods for future climate conditions. Here, we focus on the scenario-neutral method of climate change impact assessment that allows to analyse the sensitivity of a catchment to a range of changes in selected meteorological variables. The main challenges of the scenario-neutral approach are the choice of meteorological variables and statistics thereof, and how to generate time series representing altered climatologies of the selected variables. Different methods have been proposed for the scenario-neutral assessment of future flood occurrence, yet a comparison of conceptually different methods is still missing. In the presented study we thus apply three different scenario-neutral methods to derive discharge projections for a meso-scale, pre-alpine catchment and compare resulting climatologies of annual maximum floods (AMF). The methods differ with regard to the statistical attributes of precipitation and temperature change they consider, in the tools to generate altered meteorological time series, as well as in the type of input data they rely on. Resulting average change in mean AMF peak magnitudes and volumes was in the order of -6% to +7%, and of -11% to +14%, respectively. In addition, differences in the variability of mean AMF peak magnitude and volume change were observed (interdecile range of 20% to 46%for peak magnitudes and of 29% to 44%for flood volumes, respectively). Moreover, different relationships between flood volume and peak magnitude change resulted for the different methods. These differences were linked to disparate flood regime shifts caused by the three approaches, i.e. different flood seasonality and hydro-meteorological flood event properties. The results highlight that focussing only on selected aspects of climate change when performing scenario-neutral assessments of flood occurrence can lead to differences in the prevailing flood generating processes and thus different hydrological flood event characteristics.
In the third part of this thesis we present an approach for assessing uncertainty in flood loss estimates by large ensemble modelling. Various sources of uncertainty affect the estimation of flood loss under climate change and ensemble modelling is commonly applied to represent uncertainty in few selected aspects. However, the consideration of several factors can be hampered by the high computational demand of performing resource-intensive computations a large number of times. Surrogate models can be used to reduce the computational burden of numerical models and can thus facilitate comprehensive uncertainty assessments. In the presented study we demonstrate the use of a surrogate inundation model for large ensemble flood loss modelling. The approach is exemplified for the floodplains of a mesoscale, pre-alpine study area. The surrogate model describes the relationship between flood peak magnitude and inundation depth at the flood exposed buildings for the study area and thereby replaces a fully integrated two-dimensional hydrodynamic simulation of projected peak discharge. Four generic sources of uncertainty were considered: The choice of a scenario-neutral method, climate projection uncertainty, hydrological model parameter sets, and vulnerability function choice. In total, 81000 estimates of flood loss were derived. From these, 46500 estimates were selected according to the change of annual mean precipitation and temperature of an ensemble of regional climate models, and considered for an attribution of uncertainty. Estimated loss due to a100-year flood event showed large uncertainty ranging from a decrease of loss compared to present day climate to more than a 7-fold increase. Vulnerability function choice was identified as the most important source of uncertainty explaining almost half of the variation in the loss estimates. Choice of a scenario-neutral method and climate projection uncertainty jointly contributed nearly as much. These results point to a large potential of improved vulnerability modelling for reducing uncertainty in flood loss estimates for the study area. More generally, the study demonstrated how a surrogate inundation model can facilitate large ensemble flood loss modelling and thereby comprehensive uncertainty assessments.

Interest & Impact

Downloads

3 since deposited on 08 Apr 2020
0 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item