Finer Rainfall Maps Did Not Buy Britain Better River Forecasts Everywhere

Researchers at the UK Centre for Ecology and Hydrology and the Met Office Hadley Centre have published an evaluation of the river flow forecasting scheme that has driven Great Britain's monthly water outlook since December 2023. The scheme takes rainfall recorded in past years whose broad weather patterns resembled the one now forecast, and uses that rainfall to drive a national river model on a 1 km grid.

The forecasts cover the month and the season ahead and feed the UK Hydrological Outlook, the monthly national outlook for river flows and groundwater levels. The authors say sharper forecasts would support decisions that are costly or slow to reverse, such as changing how much water is held in reservoirs or switching on transfers between regions. The evaluation is open access in Hydrology and Earth System Sciences.

The authors report that the river flow forecasts performed better than the rainfall forecasts that drive them, tested against what actually happened between 1993 and 2017, and that much of that advantage came from knowing how wet the ground already was rather than from the weather forecast. Both were most skillful in winter and performed adequately in spring and autumn.

Finer input did not uniformly buy more accuracy. Moving to the 1 km grid degraded some of the skill earlier work had found using the same method at lower resolution, and the rainfall forecasts now match, rather than beat, what coarser models achieve. The version the outlook ran before applied one national rainfall figure to the whole country, scaled by local long-term averages, from a weather model with a 50 km grid.

The paper also marks out where the rainfall forecasts matter most: in fast-responding basins, mainly in the northwest, and much less where groundwater gives a river a long memory. The authors offer that contrast as a guide to where better rainfall forecasting, or better river modeling, would pay off most.