Combination Hazard of Extreme rainfall, storm Surge & high Tide on estuarine infrastructure (CHEST)

Lead Research Organisation: University of Hull
Department Name: Geography

Abstract

Background and challenges: UK estuaries are at risk from combination flooding. Sea-level rise and predicted changes to UK storm patterns (affecting both surge and river flows) will alter the joint probability of multiple hazard events, making previous understanding of risk, and mitigation measures, potentially obsolete. Existing probabilistic methods for assessing combination hazard (e.g. FD2308) provide only water level hazards, with limited detail on where and when issues may occur, and vitally cannot readily accommodate revised estimates of event distributions due to sea-level rise and climate change.

Combination hazards, therefore, present a clear risk to project partner's infrastructure in estuaries (e.g. Flood defences, railways, roads, water treatment works, nuclear power stations), both under present conditions - and more with greater uncertainty in the future. The ability to better forecast the specific locations and timings of such combination hazards will enable effective planning and timely warnings to industry operators, partners and the public.

All of our project partners have considerable investments in areas prone to estuarine flooding including flood defences (EA), rail networks (Network Rail), water supplies (Welsh Water) and nuclear power stations (EDF Energy). The scale of national exposure to combination hazard is phenomenal. Taking the Humber alone, the projected economic costs of the 2015 surge tide occurring with 0.3m of sea level rise would be £12.5bn in direct flood damage and £10.8bn in consequential losses. Quite simply, our project partners cannot ignore how combination hazard may increase with climate change.

Aims & Objectives: CHEST will assess if the management of combined river-surge-wave-tide flooding in UK estuaries (and surrounding low-lying areas) needs to be amended, especially in the light of sea-level rise and changing climates. This leads to three objectives:
1. Establish if the interaction between the combination hazards of rivers, surge, waves and tides is important to resolve for estuarine flood risk.
2. Determine which of these factors and in what combination pose the greatest risk. i.e. What is the sensitivity to the different hazards and does it differ in different locations (a) within an estuary (b) between different estuaries?
3. With sea-level rise and climate change - ascertain whether the relative importance of the combination hazards change or shift. i.e. with SLR do surges become more dominant?

To achieve these objectives, CHEST will use novel, fast numerical models (developed in previous NERC and EA funded projects) to simulate the combination hazard for two contrasting pilot studies (Humber and Dyfi estuaries) where the primary project partners have particular concerns. In a second phase the CHEST modelling framework will be documented, packaged and where necessary tailored to a wider group of additional project partners allowing the application to estuaries around the UK.

Deliverables will be predictions in pilot basins of 'worst case' scenarios, system sensitivity to different combination hazards and for the impacts of climate change (Milestones 1, 3 & 4). As requested by PP's these will be in the form of digital maps/data of flood inundation areas, max depth, max velocities, water surface (AOD), flood duration, and a matrix/look up table for combination hazard sensitivity. For the model roll out phase, deliverables will be the software (source and exe) along with documentation and tutorials (Milestone 5). This may also involve tutorials/webinars - as requested by PP's at meetings C and D.

Project duration is 10 months involving researchers from the University of Hull and Bangor University. Cost is £104 700 (80%) including 10 months PDRA (Hull); CoI and PI costs and travel for meetings. Lewis' contribution is in kind (20% for 10 months), as a Ser Cymru NRN-LCEE Research Fellow on the QUOTIENT project.

Planned Impact

The main expected benefits to all project partners will be an increased understanding of the magnitude and likelihood of flooding from a combination hazard impacting on their infrastructure and its operation. This includes developing a method to determine how this hazard may shift due to climate change with the availability of the next generation of climate projections (e.g. UKCP18); as well as the how the sensitivity of estuaries (of different sizes and shapes) to hazards may alter (e.g. which combination of hazards is most potent in what context). For ALL project partners this will influence day to day management, incident response as well as planning for future investment decisions.

For our primary project partners, the Environment Agency, the findings from CHEST will feed directly into the Humber Strategic Review presently being developed, answering questions for future flood defence investment. Furthermore, this proposal directly addresses some of the questions raised by the National Flood Resilience Review, including using integrated modelling for flood forecasting and the potential to adjust fluvial flows to represent changing land use policies (e.g. NFM). Due to the ability of combinations of hazards to shift loci of flooding from typical vulnerable spots - this will provide important insight upon which to base investment decisions (e.g. defence weaknesses or where to locate flood storage/managed retreat sites). A key feature of CHEST for the EA is the sensitivity analysis enabling the EA to understand the role the different factors play, and which are most important in determining the picture of risk. This will help the EA more effectively manage risk during an actual incident. It will help inform the future EA Long- term Investment Strategy (LTIS) program as CHEST may highlight the potential "currently "unquantified" risk in these areas, especially if it is readily transferable to other key locations, nationally.

For primary project partners, Network Rail, we will overlay their infrastructure maps with model results of flood inundation locations and depths - enabling them - for the first time to establish how vulnerable their assets are to existing flood hazard as well as combination hazards. We will also integrate the geotechnical properties of their assets (e.g. railway embankments) enabling them to make rapid decisions on asset viability based on inundation durations as well as depths. For both Dyfi and Humber estuaries this represents a significant improvement allowing very high resolution identification of likely sites of vulnerability to flooding and combination hazards.

Following successful piloting of this project on the Humber and Dyfi estuaries, we will involve our secondary project partners to help guide a roll out to other locations around the UK. The aim is to provide secondary project partners with the modelling tools to readily assess other estuaries for combination hazard risk. These include several nuclear power stations (EDF) and assets held by clients of partners ARUP, Atkins and Welsh Water. EDF are interested in the vulnerability of their nuclear installations especially to 1 in 10000 year extreme events as well as to climate change. Welsh Water in particular are concerned about asset damage (waste water treatment works) from saline intrusions and flooding extent from potentially untreated sewerage - two processes that they know little about. Their asset concerns extend throughout Wales and they will roll out the tools we develop for water quality studies.
 
Title Tracing and visualisation of contributing water sources in a model of flood inundation: video supplement 
Description These are video supplement files to Wilson & Coulthard (2021), produced using version 1.8f-WS of CAESAR-Lisflood software, available on Zenodo here. For a full description of the methodology and case studies, please refer to the paper which is available here: https://doi.org/10.5194/gmd-2021-340. Video animations (no audio) for the following case studies are included: 1. Carlisle, United Kingdom (carlisleanimation-sourcetracing.avi and carlisleanimation-depthonly.avi): Simulation of the January 2005 flood event at the confluence of the Rivers Caldew, Petteril and Eden, using a 5 m grid. Both water source tracing and depth only versions are provided. In the water tracing version, blue colours represent flows from the River Eden, reds are from the River Petteril and greens are from the River Caldew; darker shades represent deeper water. Available on YouTube here: https://youtu.be/xOtOi06cXvA In the depth only version, darker shades of blue represent deeper water, with no information about the water source in a grid cell. Available on YouTube here: https://youtu.be/aFz-sPRGHVE 2. Avon-Heathcote estuary in Christchurch, New Zealand (avonheathcoteanimation.avi): Simulation for July 2017, which included a high flow event on 22 July, using a model grid of 10 m. Blue colours represent flows from tide, reds are from the River Avon and greens are from the Heathcote River; darker shades represent deeper water. Available on YouTube here: https://youtu.be/Fczr5tczzXU 3. Amazon (amazonanimation.avi): Simulation at the confluence of the Solimões (mainstem Amazon) and Purus rivers in the central Amazon, Brazil, for the period of 1 October 2013 through December 2014, using a ~270 m model grid. Red colours are from the Solimões, green colours are from the Purus; darker shades represent deeper water. Available on YouTube here: https://youtu.be/PknAL_8fd1I 4. Planar slope (planaranimation.avi): A simple test case consisting of a 2000 x 1000 m planar slope (0.001 m/m), with walls added at 250 m intervals across the slope, each of which has several gaps through which water can flow. Model grid was 5 m. Eight water sources were traced in total, with three visualised in the animation: red = source 2, green = source 4, blue = source 6. Depths are shown in the middle plot. Available on YouTube here: https://youtu.be/DTw8ysJtx8o Please feel free to use these animations, under the terms of the CC-BY-4.0 license. Please provide a link back to this site and a citation to Wilson & Coulthard (2021). Reference: Wilson, M. D. and Coulthard, T. J.: Tracing and visualisation of contributing water sources in the LISFLOOD-FP model of flood inundation, Geosci. Model Dev. Discuss. [preprint], https://doi.org/10.5194/gmd-2021-340, in review, 2021 
Type Of Art Film/Video/Animation 
Year Produced 2021 
URL https://zenodo.org/record/5548534
 
Description This project developed an integrated hydrodynamic model (CAESAR-Lisflood) to establish a method to characterise how multiple hazards are propagated through estuaries. This estuarine flood risk tool was piloted on two contrasting estuaries (the Dyfi and Humber) to determine combination flood risk. Sensitivity analysis showed that: (1) the region between the fluvial and marine flooding processes increased water-levels upstream and a reduced water-depth downstream; likely the result of a back-water effect and reduced pressure gradient driven flow of the storm tide into the estuary. Therefore, simplified inundation models are an appropriate tool for combination flood risk assessment in estuaries, (2) the magnitude of hazard drivers, not the timing, had significant effect on inundation extent; therefore, future changes to the magnitude of extreme river flows (and not changes in the shape of hydrographs) appear important to resolve. Fluvial timing was only found to have an effect in smaller estuaries, suggesting future river flow projections at a frequency less than daily-average values are need (especially for water quality).
Our process-based method addressed two weaknesses within the present joint probability approach (FD2308) for assessing combination hazards. Firstly, statistical methods may stack forecast surge/flow/tide levels together, but by using a 2D hydrodynamic model, CHEST can directly simulate the effect of surge/flow/tide/waves acting together on the volume of water within the estuary. This allows us to simulate the important nuances of flooding as well as non-linear interactions (volume and momentum effects; backing up of flow, water storage when overbank, return flows etc..) to generate detailed spatial and temporal forecasts of inundation.
Secondly, and crucially, our approach can account for the non-stationarity of climate change and sea-level rise - that is not accounted for in existing methods. This can be added directly in CHEST by driving the modelling with sea-level rise, storm, surge and wave predictions that will be based on the latest UKCP18 projections.
Exploitation Route The CAESAR-Lisflood model, used for CHEST, is freely available for all project partners to use. It is computationally efficient and optimised to operate over multiple processors/cores enabling its use on non-specialist hardware. The model is easy to use (graphical interface) facilitating its roll out for partners and other agencies to use in other estuary locations throughout the UK. Developed by the lead applicant, the IP and model code and methods have all been made freely available under a GNU licence so therefore free for all (and additional) project partners to use.
For the Dyfi and the Humber - maps of likely inundation areas for different combination hazards in present day and under climate change scenarios were created. These are of great use to the project partners for their strategic and day to day management of these sites.
For areas where the model was used (Humber, Dyfi and in the latter stages the Severn) corrected and merged DEM and bathymetry data is available for any partners to use.
For agencies/companies outside of the project partners, since CHEST JBA consultants are piloting CAESAR-Lisflood for additional applications, and the Met Office are interested in using this method to establish flood risk for estuaries under the new UKCP18 data.
Sectors Construction

Environment

Transport

 
Description Used to inform project partners (EA, EDF, ATKINS, Network Rail) about flood risk in the humber and dyfi catchments.
First Year Of Impact 2018
Sector Energy,Environment,Transport
 
Description UoH Present & Future Climate Hazard/Embedded Researcher Scheme
Amount £99,353 (GBP)
Funding ID NE/V004247/1 
Organisation Natural Environment Research Council 
Sector Public
Country United Kingdom
Start 07/2020 
End 08/2022
 
Title CAESAR-Lisflood v1.8f-WS (water source tracing and visualisation) 
Description Update 9 May 2023: The Wilson & Coulthard paper in GMD is now published and available here: https://doi.org/10.5194/gmd-16-2415-2023. Updated code is also available in version 1.9j-WS, available on Zenodo here: https://doi.org/10.5281/zenodo.7589023 This is the code for the CAESAR-Lisflood model with a proposed method for water source tracing added (software version 1.8f-WS). The code is based on version 1.8f of the CAESAR-Lisflood model which is available here. This additional methodology will be included in a future version of the main software branch. The methodology, example applications, and computational benchmarking are described in the accompanying paper by Wilson & Coulthard [https://doi.org/10.5194/gmd-2021-340]. The methodology developed allows additional insights into flood dynamics to be gained by accounting for flow pathways and is independent of the hydraulic formulation used. The software provided here includes the easy to use GUI of the main CAESAR-Lisflood software, with additional functionality added to enable water source tracing and for visualisation. Along with the code, two example applications are provided: the planar test case and a 15 m version of the Carlisle model. Additional examples are available in these videos (applications as described by Wilson & Coulthard), each produced with output from this software: Carlisle flood model, UK (5 m version). Avon-Heathcote shallow estuary model, New Zealand. Amazon-Purus confluence, Brazil. Planar test case. For details of each of these, please see the Wilson & Coulthard paper [https://doi.org/10.5194/gmd-2021-340]. Video files are also available. For details of in general use of the CAESAR-Lisflood software, please see the main software website here. The additional GUI components for water source tracing are as follows: Enable water source tracing in the Hydrology tab, with the checkbox under the reach input variables. Note that each source file specified will be counted as one source, even if it is used in multiple grid cells. If you are using rainfall as an input, you can also provide a rainfall zonation map to split the tracing of rainfall depending on where it falls (e.g. in different catchments). Under Save Options, optionally place a check next to the water tracers and rain zone tracers outputs. Set up the rest of the model in the same way as any other simulation, then load data and press Start. To visualise the water sources during a simulation: an extra water tracer visualisation control box is added to the GUI. To view the fraction of depth from each source, assign numbers to each of the red (R), green (G) and blue (B) colour options. Note that sources 1 and 2 are reserved for downstream stage and rainfall, so the fluvial sources begins at 3 and increases in the order specified in the reach input variables. The enhance slider control changes the value of beta used for emphasising lower water source fractions (see equations 11 and 12 in Wilson & Coulthard). The point info window checkbox opens an extra dialog which allows a user to click on the map and obtain information about depth, flow velocity etc. The fraction of each water source in the cell is included in this dialog - not that the total water source tracer should sum to 1.0: very small deviations from this are possible due to numerical errors. Getting started: run the software and open (using Config File > Open) either the Carlisle 15 m case study (carlisle15m load file.xml) or one of the planar test files (e.g. model_trace_8.xml for 8 tracers). These are all set up and you should be able to click load data then Start. Reference: Wilson, M. D. and Coulthard, T. J.: Tracing and visualisation of contributing water sources in the LISFLOOD-FP model of flood inundation, Geosci. Model Dev. Discuss. [preprint], https://doi.org/10.5194/gmd-2021-340, in review, 2021 
Type Of Technology Software 
Year Produced 2021 
Open Source License? Yes  
URL https://zenodo.org/record/5541122
 
Title CAESAR-Lisflood v1.9j-WS (water source tracing and visualisation) 
Description This is the code for the CAESAR-Lisflood model with a proposed method for water source tracing added (software version 1.9j-WS). The code is based on version 1.9j of the CAESAR-Lisflood model which is available here. This additional methodology will be included in a future version of the main software branch. The methodology, example applications, and computational benchmarking are described in the accompanying paper by Wilson & Coulthard [https://doi.org/10.5194/gmd-2021-340]. The methodology developed allows additional insights into flood dynamics to be gained by accounting for flow pathways and is independent of the hydraulic formulation used. The software provided here includes the easy to use GUI of the main CAESAR-Lisflood software, with additional functionality added to enable water source tracing and for visualisation. Along with the code, an example application is provided of 15 m version of the Carlisle model. This is in the folder Carlisle Example Getting started: run the software and open (using Config File > Open) either the Carlisle 15 m case study (carlisle15m load file.xml). These are all set up and you should be able to click load data then Start. The additional GUI components for water source tracing are as follows: Enable water source tracing in the Hydrology tab, with the checkbox under the reach input variables. Note that each source file specified will be counted as one source, even if it is used in multiple grid cells. If you are using rainfall as an input, you can also provide a rainfall zonation map to split the tracing of rainfall depending on where it falls (e.g. in different catchments). Under Save Options, optionally place a check next to the water tracers and rain zone tracers outputs. Set up the rest of the model in the same way as any other simulation, then load data and press Start. To visualise the water sources during a simulation: an extra water tracer visualisation control box is added to the GUI. To view the fraction of depth from each source, assign numbers to each of the red (R), green (G) and blue (B) colour options. Note that sources 1 and 2 are reserved for downstream stage and rainfall, so the fluvial sources begins at 3 and increases in the order specified in the reach input variables. The enhance slider control changes the value of beta used for emphasising lower water source fractions (see equations 11 and 12 in Wilson & Coulthard). The point info window checkbox opens an extra dialog which allows a user to click on the map and obtain information about depth, flow velocity etc. The fraction of each water source in the cell is included in this dialog - not that the total water source tracer should sum to 1.0: very small deviations from this are possible due to numerical errors. Additional examples are available in these videos (applications as described by Wilson & Coulthard), each produced with output from this software: Carlisle flood model, UK (5 m version). Avon-Heathcote shallow estuary model, New Zealand. Amazon-Purus confluence, Brazil. Planar test case. For details of each of these, please see the Wilson & Coulthard paper [https://doi.org/10.5194/gmd-2021-340]. Video files are also available. For details of in general use of the CAESAR-Lisflood software, please see the main software website here. 
Type Of Technology Software 
Year Produced 2023 
Open Source License? Yes  
URL https://zenodo.org/record/7589022