Towards forecast-based climate resilience and adaptation in the water sector

Usual applications of forecasts to resilience assessments in water systems ask the question "What are the benefits of forecasting product X for water system Y?". This project proposes to start asking instead: "What are the forecast characteristics that would increase the resilience of a water system to climate-related risks? what variables? what lead times? and with what accuracy?" Such an approach puts the focus on the needs of forecast users. This will enable water managers, government agencies, and communities, to identify more easily which forecasts would be useful to them. It will also help forecast providers such as the Met Office to focus forecast improvement efforts to areas where they would be most beneficial.

The work as part of this embedded researcher scheme aims to:

A) Start tackling the question of mapping the potential benefits of forecasts depending on their performance, by building a freely available, open-source Python toolbox that does that for a single planned water infrastructure asset (e.g. a storage reservoir with pumps and treatment plant). The toolbox will implement a stress testing procedure to determine which events or combination of events present a risk to the performance of the asset (supply disruption, financial risk, etc.). It will then incorporate a simple synthetic forecast generator to evaluate the ability of forecasts to accurately forewarn of climate-related hazards that can affect system performance. In a final step, the toolbox will be linked with simple multi-objective optimisation algorithms to trade off the benefits of investing in mitigation / adaptation actions to avoid bad performance, vs. the cost of implementing these actions as a result of a false alarm given by the forecast. This will help to understand which forecasts should be used to trigger appropriate mitigation and / or adaptation actions at the asset level, and what forecast precision is required for this.

B) Develop a long-term collaboration between the host organisation Anglian Water (AW) and Dr Charles Rougé (CR). The successful implementation of the open-source Python toolbox, and its application to a key asset in AW's long-term adaptation plans, will only be a first step in that direction. Planned activities during with CR embedded at AW will lead to the submission of grant proposals to extend that work, with AW as key partner and beneficiary.

1) A first proposal will (i) design the next generation of synthetic forecast generators to simulate forecasts for several climate variables at once, with different forecast lead times, while reproducing desired statistical properties (precision, correlation between the different forecasts, etc); and (ii) apply this new synthetic forecast generator to the development of flexible forecast-based adaptation plans where new water infrastructure investments decisions would be triggered not only by climate events but also by the availability of new forecast products with the potential to improve how water systems can be managed. This proposal will be submitted during the project and will have the Met Office as its other key partner.

2) Further work will scope out how forecast-based resilience tools can help to support the further development of strategic water planning models used by water utilities to make long-term adaptation plans. This work will focus on assessing a new functionality of one such model, which enables return flows (effluents from water treatment plants) to vary dynamically as a function of water demand. Representing them would enable to detect unintended consequences of demand management as it may reduce effluent discharges sustaining environmental flows at key locations. Consequences on adaptation depend on forecast supply and demand during drought conditions, as they are projected to evolve in coming years and decades.