Water demand forecasting sits at a critical juncture. As climate patterns intensify, populations shift, and the ripple effects of events like COVID-19 permanently reshape how and when people use water, the traditional models that have guided UK water resource planning for decades are no longer fit for purpose on their own. This UKWIR guidance confronts that reality head-on, offering the sector a coherent framework for evolving its forecasting methods, not by prescribing a single “best” approach, but by equipping planners with the tools to make smarter, more defensible choices.
The report, Improving Approaches to Demand Forecasting, provides updated guidance for Water Resource Management Plans (WRMPs) from 2029 onwards. It frames water demand as a complex system shaped by the interplay of demographic, economic, behavioural, environmental and policy drivers, many of which interact in non-linear and unpredictable ways. A structured evaluation of forecasting methods across five modelling families (time-series, econometric, end-use, simulation and machine learning) confirms that no single technique adequately serves all planning purposes, and that method selection should be driven by planning risk and decision context rather than model sophistication alone. The guidance gives particular attention to three areas of growing importance: embedding climate change directly within weather-driven demand analysis rather than treating it as a bolt-on uplift; strengthening peak demand assessment through stochastic methods and peak demand libraries that move beyond limited historic records; and integrating smart meter data to improve behavioural insight and model calibration, while cautioning against over-reliance on datasets that remain unevenly deployed and incompletely linked to explanatory drivers. Throughout, the report emphasises proportionality, transparency and governance, ensuring that forecasting advances are matched by clear documentation, auditability and organisational capability to sustain them across planning cycles.
