Apply for this PhD here https://applyto.newcastle.ac.uk/ using application studentship code FLOOD245. Please contact Caspar Hewett (caspar.hewett@newcastle.ac.uk) if you have any questions about the application process.
Natural flood management is now established as an important part of catchment management of flood risks. These interventions slow the flow of water through catchments, thereby reducing the investment needed on traditional defences. Measures include creating and restoring ponds and wetlands, enhancing soils and afforestation to slow overland flows and aid infiltration, and constructing features like leaky dams to hold back water in channels.
Design of NFM strategies, especially over urban or river catchment scales, is recognised to be a complex and combinatorial problem - because of the number of potential measures, the many potential combinations of measures and the many locations where they could potentially be installed, and the heterogeneity of hydrological processes in a catchment. This is typically more complex than traditional flood defences as the flood risk reduction benefits arise from an interacting combination of many small measures (Ramsbottom et al., 2019).
This project will combine innovative modelling approaches with hydrological modelling to identify optimal NFM strategies that balance multiple and often competing objectives, including capital and maintenance costs, location, size and choice of NFM measure, effectiveness at risk reduction. Catchment stakeholders will be able to explore the relative merits, and potential tradeoffs, of different spatial NFM strategies.
At least one urban (e.g. Newcastle) and one rural (e.g. Irthing) catchment will provide the initial and distinctive case studies for this research.
1. Flood risk analysis. Physically-based models such as SHETRAN or CityCAT can be used to assess flood risk and simulate NFM measures in catchments (Barnes et al., 2023) and cities respectively.
2. Spatial optimization and tradeoff analysis. A spatial optimization method (Caparros-Midwood et al., 2017) will be developed and adapted to assess NFM strategies. The location and design of a number of spatial NFMs strategies will be randomly generated. The impact of each strategy on objectives including cost and flood risk reduction will be evaluated. A spatial genetic algorithm (GA) will, over successive iterations, find and improve strategies against these objectives. Tradeoffs between different design choices will be visualized using Pareto plots so stakeholders can identify preferred NFM strategies.
3. Machine Learning (ML). Results from thousands of NFM spatial strategies simulated in the case studies will be used to train a ML algorithm. The performance of the subsequent ML in rapidly identifying optimal strategies using just remote sensing and hydrological model input data will be benchmarked against the GA and Arup’s NFM suitability tool.
Ramsbottom et al. (2019) Effectiveness of natural flood management measures. Proc. IAHR World Congress. https://static.iahr.org/34/368
Barnes et al. (2023) Leaky dams augment afforestation to mitigate catchment scale flooding. Hydrological Processes, 37(6), e14920. https://doi.org/10.1002/hyp.14920
Caparros-Midwood et al. (2017) Spatial Optimization of Future Urban Development with regards to Climate Risk and Sustainability Objectives, Risk Analysis, 37(11): 2164–2181. https://doi.org/10.1111/risa.12777
