The formation and melting of snow can be a key component of river flows in the UK, particularly in upland areas. Different configurations of the countrywide Grid-to-Grid (G2G) model are used to deliver the snowmelt component, using different approaches for Scotland; and England and Wales. The Scotland configuration used by the forecasting service uses the G2G Snow Hydrology scheme with Numerical Weather Prediction (NWP) model inputs of precipitation and air temperature at screen-height to form and melt the snowpack. In contrast the Flood Forecasting Centre (FFC) in England and Wales uses NWP parameters to split precipitation into rain and snow; and the Joint UK Land Environment Simulator (JULES) to estimate snow melt as an external input to G2G.
We have recently taken delivery of a review of the two methods, in a project jointly commissioned by the SFFS and FFC and carried out by the Centre for Ecology and Hydrology (CEH). The report reviews the snow hydrology methods, describes the operation of the two methods, looks at ways to improve them, and carries out comparisons between the two methods. Some examples of the output are shown below.
The study confirmed that modelling snow processes generally improves performance, irrespective of the particular snow formulation used. A structured comparison of the two methods currently in use, alongside other formulations, shows the NWP-JULES method currently outperforming the G2G Snow Hydrology (G2GSH) module. This difference disappears when a revised parameter set for G2GSH based on a more extensive hydrometric data record is used. There is still substantial river flow uncertainty at the catchment level, however, with peaks being over- or under-predicted.
As well as providing useful insights into how the model is performing, some recommendations were made. In the short term, new parameters for the G2G snow hydrology scheme could be implemented in time for next winter, to provide some immediate improvements. Longer term, more work is required on validation through snowpack monitoring and also on quantification of uncertainty, using NWP ensembles. The applicability of the NWP-JULES method used in England and Wales should also be considered.