Last year, Louise Slater published a paper on recent trends in U.S. flood risk in Geophysical Research Letters. The paper uses an interesting new approach to quantify changes in flood risk. We decided to ask Louise a few questions about the paper.
Q: Where are you from, where are you based, and what are your current research interests?
A: I grew up in the UK but spent a number of years in France, before returning to do a PhD at the University of St Andrews. After a brief first lectureship and a Postdoc in the USA, I have finally settled down at Loughborough University. My research focuses on disentangling the different drivers of flooding to improve flood predictions over a range of spatial and temporal scales, and across disciplines (hydrology, climatology, and geomorphology). This includes, for example, developing methods to quantify the influence of changing river channel capacity on flooding, as well as statistical models that forecast streamflows ‘dynamically’ using both global climate model outputs and projected changes in land use. In future work, I aim to better integrate these different geomorphic, climatic and land use drivers to enhance projections of flooding and streamflow.
Q: What is the take home message of your paper?
A: Our paper reveals strong regional patterns of changing flood risk across the United States using a new metric, changes in the number of flood days above flood thresholds (action to major flooding). Broadly speaking we found that, in recent decades, flood risks have been increasing across the central/northeastern U.S. and decreasing across the southern U.S. These patterns of change appear to be related to progressive spatial redistributions in long-term river basin wetness (measured using total freshwater data from NASA’s GRACE satellite).
Q: In this paper, you take a different approach to assessing trends in floods by using inundation frequency above government flood level warning categories, rather than more commonly used discharge magnitudes and frequencies – What are the advantages?
I believe there are several advantages of using a stage-based approach rather than traditional discharge-based metrics for communicating and understanding changing flood patterns. The first is that stage (water levels – which are measured directly at the gauge) has far fewer uncertainties than discharge (which is only estimated), especially during extreme flood events. Second, the ‘traditional’ approach of computing changes in discharge does not necessarily tell you how the flood risk is changing, and can even be counterintuitive. For example, if streamflow magnitudes increase over time, but the capacity of the river channel also increases (through widening or degradation), then there may not actually be any change in flooding at all. It is only by using the stage data, in association with local flood risk categories, that we may determine how the flood risk is actually changing, i.e. how people and assets may be affected. Last, and I think this point is most important from the perspective of flood risk communication and management, water levels are far more intuitive than discharge to most people. If you tell someone that the annual maximum flow in their local river has increased by, say, 50 m3/s/year, then that won’t really mean much to them. If, however, you say that the highest flood levels are now being exceeded 10 days per year, as opposed to just 1 day per year in the past, the changes in flood risk become much clearer.
Q: Your paper mostly describes interesting patterns of past flood changes. Can we also extrapolate your work to future conditions?
A: I would warn against any direct extrapolation of results based on past trend analyses, because there are too many different drivers at play: changes in the climate system, anthropogenic influences, changes in sediment delivery, and any subsequent channel adjustment. However, I do think there is a strong potential for forecasting changes in the frequency of exceedance of flood categories using different time-varying predictors (changes in land use, antecedent basin wetness, precipitation and temperature), and that is something that we are working on now.
Q: Your flood-stage based approach also seems relevant for assessing changes in flood risk in other regions. Are you planning to work on that in the future?
A: It would certainly be interesting to assess how flood risks are changing in other regions, and I would be keen to collaborate on the topic. However, to use a similar approach one would need stage time series and flood risk thresholds for each stream gauge (e.g. minor flooding at 5m, major flooding at 7m.). In the USA, the National Weather Service has already determined the flood risk thresholds locally at each gauge, so it is relatively easy to tie changes in water levels to flood risk indicators using a peak-over-threshold approach. In places where flood risk categories are not available, one could potentially set a statistical threshold instead (e.g. the water level that is exceeded 99% of the time), but that would provide information on the changes in flood hazard, as opposed to changes in flood risk for local populations and assets.
[END] – Interviewed by Wouter Berghuijs & Shaun Harrigan
This interview is part of the new YHS Research “Hylights” series to showcase interesting and outstanding work by early career scientists. Selection criteria are not set in stone, but reasons to select work can include e.g. novelty and relevance of findings, fun of reading, unique collaborations, media coverage and generated controversy. Selected work will be provided with a short layman summary, and a short written or video interview with the (first) author(s). Tips can be sent to young firstname.lastname@example.org or email@example.com.
Reference paper: Slater, L. J., and G. Villarini (2016), Recent trends in U.S. ﬂood risk, Geophys. Res.Lett., 43 , 12,428–12,436, doi:10.1002/2016GL071199.