Reducing surface temperature errors at coastlines
Hogan, Robin ; Bozzo, Alessio
To limit its high computational cost, the current radiation scheme in ECMWF's Integrated Forecasting System (IFS) is run on a coarser spatial grid than the rest of the model. As a result, surface radiative fluxes computed over the ocean may be used over land, where the surface temperature ('skin temperature') and surface albedo are very different. This can lead to large near-surface temperature errors at coastal land points. For example, in Norway night-time 2-metre temperature forecasts at coastlines can sometimes be too low by more than 10 °C. The radiation scheme is also run infrequently, which leads to errors in the diurnal cycle of skin temperature. A computationally efficient solution to these problems has been developed, in which the surface longwave and shortwave fluxes are updated at every model time step and grid point according to the local skin temperature and albedo. The computational cost of performing approximate radiation updates is only 2% of the cost of the full radiation scheme, and the overall run time of the model is only increased by about 0.2%. Testing the new scheme by running daily five-day forecasts over an eight-month period has revealed a significant improvement in 2-metre temperature forecasts at coastal stations compared to observations.
Accès à la notice sur le site du portail documentaire de Météo-France