The Unified Model, used at the UKMO for operational forecasting as well as all climate research, is a fairly recent development and the version used for the first AMIP run was one of the earliest. Since then there has been steady progress in improving many aspects of the physical parametrization schemes, and it is planned to repeat the run with an updated version of the model. Many systematic errors in the simulated climate have been identified, and amongst them a few stand out as being of particular importance:
- A cold bias throughout most of the troposphere
- Excessively strong flow in the surface westerlies in middle latitudes, the trade-wind easterlies, and at jet-stream level
- Too much cloud over the northern continental land masses in summer and too little stratocumulus over the tropical oceans
- An over strong summer monsoon flow over South East Asia which, coupled with excessive rainfall, extends too far into the West Pacific
Radiative fluxes. A new radiation scheme has been written which more accurately calculates the gaseous transmissions in the longwave and provides a more consistent treatment of clouds. Long-wave cooling is less than in the original scheme and the cold bias is reduced by about half.
Cloud physics. Although the original cloud scheme gave radiation fluxes in good agreement with ERBE data, several aspects of the formulation were unsatisfactory. Improvements in the parametrization include a better diagnosis of cloud liquid water and changes to the ice-water ratio in mixed-phase clouds. In runs with the modified scheme, cloud amounts in middle latitudes and the top-of-atmosphere albedo are both closer to observed values.
Gravity-wave drag and orographic roughness. The parametrization of drag due to the sub-grid scale effects of mountains has been modified to include anisotropy of the orography, the effects of trapped lee waves, and a better representation of high drag states where flow blocking and hydraulic jumps occur. Roughness lengths for momentum have been redefined to include the effect of orographic form drag. This scheme applies more drag in the lower troposphere, resulting in some reduction in the overall westerly bias.
Convection. An improved tropical simulation was achieved when an error in the evaporation of convective rainfall was corrected. In particular the excessive strength of the Asian summer monsoon was reduced.
Other items. High diffusion in the original run was leading to excessive smoothing of synoptic-scale features and causing local precipitation bulls-eyes over mountains. Levels of diffusion have been reduced for all variables, and in the case of moisture it has been made more scale selective. Corrections for negative moisture where adjustments were applied globally were found to be causing a steady drying in the stratosphere; use of a scheme which borrows moisture from neighboring points was found to reduce the problem.