Evaluation of surface water and energy cycles in the Met Office global NWP model using CEOP data

Components of the surface water and energy balance are evaluated in the Met Office global Numerical Weather Prediction (NWP) 00-36 hour forecasts using (i) GEWEX Continental Scale Experiment (CSE) reference site data provided by the Coordinated Enhanced Observing Period (CEOP) project and (ii) other earth observing datasets from satellites and in-situ measurements. The global hydrological cycle in the model is in reasonable balance in recent model versions. However, comparison against available observations suggest that both precipitation and evaporation are overestimated over both land and ocean, with largest errors over the tropical oceans. Comparison of the model's seasonal and diurnal cycles of surface fluxes, temperature, precipitation, and moisture against the GEWEX/CEOP CSE sites during October 2002 to September 2003 reveals several issues concerning model parametrization performance. For the high latitude (Arctic) sites low cloud is overestimated during boreal winter leading to an enhanced greenhouse effect and too warm near surface temperatures in the model. Snow melt occurs too early in boreal spring in the model forecasts which currently have no snow analysis component. Possible reasons for early snow melt are discussed including the specification of albedo over heterogeneous terrain and excessive sublimation of snow in the model. The mid-latitude CEOP sites all show excessive evaporation and precipitation during boreal spring and early summer and comparison with the International Satellite Cloud Climatology (ISCCP) products show systematic under prediction of cloud cover over mid-latitude land in summer. Some of the largest biases in mid-latitude surface fluxes in summer are during daytime on non-precipitating cloudy days where the model overestimates the downward SW radiation, latent and sensible heat fluxes. Finally, over tropical land we see a tendency to underestimate precipitation which is in direct contrast to tropical oceans where precipitation accumulations are too large. We have also used the CEOP data to investigate structural errors in the vertical profiles of tropical humidity which suggest deficiencies in the convection scheme.