Abstract
The performance of numerical weather forecasts made with the Met Office Unified Model is evaluated using near‐surface measurements made at a comprehensively instrumented site over the course of the annual cycle, with the aim of characterizing the forecasting system's generic behaviour in the surface layer. The study is focused on the seasonal diurnal cycles of the screen‐level and surface temperatures and the surface flux budget.
High‐resolution versions of the model simulate seasonal composite diurnal cycles of the screen temperature to an accuracy of about 1 K, but appear to show a cold bias in the surface skin temperature by day and a warm bias at night, while underestimating the magnitude of the difference in temperature between the screen level and the surface, particularly in conditions of light winds and strong surface cooling.
The consistency of the measured surface fluxes is assessed by deriving the full annual cycle of the monthly mean atmospheric surface fluxes. The measured surface flux budget at this site is closed to an accuracy better than 5 W m^−2^. Comparison of the modelled and measured fluxes shows that in spring and summer the modelled downward shortwave fluxes are too large, while the latent heat fluxes are overestimated. Copyright © 2011 British Crown Copyright. Published by John Wiley & Sons Ltd.