Coupling water quality and quantity models to integrate climate risk to reservoir water quality into water planning

Ensuring public access to clean water faces unprecedented challenges. The rising frequency and intensity of hot-dry conditions, coupled with population growth, strain water stocks. Concurrently, the presence of excess phosphorus and nitrogen in freshwater lakes and reservoirs leads to harmful algal blooms (HAB) precisely when hot-dry conditions occur, impacting aquatic life and complicating water treatment for human consumption. Due to uncertainties in future climate, water demands, nutrient discharge, and ecological factors, determination of HAB risk is a complex task. This hard-to-quantify risk impacts water planning since it is unknown whether reservoir water will be usable in the hot, dry summers when it is most needed.

Water planning increasingly involves fast water resource simulators. These tools evaluate the performance of adaptive infrastructure investments within complex water resource systems under changes in supply-demand conditions. Relying on water balance calculations, these fast models prioritise water quantity targets and typically neglect water quality. This overlooks water quality impacts on resource availability. Conversely, advancements in aquatic ecosystem modelling have produced complex water quality simulators, incorporating numerous space and time variant equations for hydrodynamics, biogeochemical and ecological processes, and particularly addressing HABs. These processes are much more complex than water balance dynamics, leading to models with much slower run-times than water resource simulators. In addition, to account for water quality in the design of operating strategies, we need two-way coupling of these models to communicate the simulated water quality and quantity states with each other frequently throughout simulations.

To achieve this two-way coupling, we integrated a high-performing lake model, the General Lake Model (GLM), into a recently developed water resource simulator called Pywr. Thanks to its 1D resolution of physical processes, GLM operates at speeds comparable to Pywr, and this work is, to our knowledge, the first to apply it to water planning. Enhanced communication between the models is facilitated by Pywr's extended parameters, allowing the execution of customised tasks at each timestep. Furthermore, Pywr's advanced scenario handling capability renders the coupled framework ideal for risk assessment. Coupled models are pivotal for designing operating strategies aimed at minimising HAB risk under diverse future conditions (climate, water demands and nutrient transport). Successful implementation will shed light on the feasibility of constructing new reservoirs, evaluating their susceptibility to algal blooms, and informing billion-pound investment decisions.