Guidelines for Streamflow Forecast Verification

WMO’s newly released Guidelines on the Verification of Hydrological Forecasts (WMO-No. 1364) provide a structured approach for assessing the accuracy and usefulness of hydrological forecasts. These Guidelines contain detailed methodologies and considerations for streamflow forecast verification that will assist practitioners in evaluating their operational systems. It focuses on several metrics to evaluate forecasts:

• Accuracy: Measures how close forecasts are to actual outcomes
• Bias: Identifies systematic deviations in forecasts
• Reliability: Assesses consistency between forecast probabilities and observed frequencies
• Resolution: Evaluates the ability to distinguish between different outcomes
• Sharpness: Reflects the confidence level of the forecasts.

By systematically assessing these factors, forecasters can identify areas for improvement and build more reliable and trustworthy forecasting systems. Some examples of the systems detailed in the Guidelines are provided below.

In Québec, Canada, the Ministry of the Environment uses the Système de Prévision Hydrologique (SPH) to predict river streamflow. SPH blends multiple weather forecasts to help experts issue flood warnings. It also reviews the accuracy of past forecasts, providing forecasters with rapid feedback. However, effective use of the system can be challenging due to constantly changing weather patterns.

In Australia, the Bureau of Meteorology uses the Hydrological Forecasting System (HyFS) to meet strict targets, including issuing timely flood warnings, predicting floods early and ensuring forecast accuracy. HyFS tracks performance at over 500 locations, generating monthly reports that highlight successes and areas for improvement.

Meanwhile, in the Netherlands, Rijkswaterstaat, the national water authority, operates forecasting systems for key water bodies like the North Sea and the Rhine and Meuse rivers. Their system, known as RWsOS, is currently being enhanced with a verification tool designed to detect issues such as model performance deterioration over time or poor data quality. The tool is further being expanded with the Ensemble Verification System (EVS), which analyses forecasts and shares the results through an online dashboard. In the coming years, Rijkswaterstaat aims to provide additional training to forecasters, empowering them to use this feedback to further improve forecast skills.

Collectively, all these systems demonstrate how technology, data, and expert judgment work together to protect communities from flooding. The process of forecast verification, comparing forecasts with actual observation, ensures these predictions are reliable and actionable. This is critical, as accurate hydrological forecasts are crucial for:

• Emergency response planning
• Water resource management
• Agricultural scheduling
• Infrastructure protection.

Thus, the new Guidelines are an important contribution to the "Early Warnings for All" initiative. By providing verification guidelines, the WMO helps nations enhance their trust in water-related hazard warning and strengthens global resilience against water-related disasters.