A detailed description of the fourth-generation ECHAM model is presented. Compared to the previous version, ECHAM-3, a number of substantial changes have been introduced in both the numerics and physics of the model. These include a...
moreA detailed description of the fourth-generation ECHAM model is presented. Compared to the previous version, ECHAM-3, a number of substantial changes have been introduced in both the numerics and physics of the model. These include a semi-Lagrangian transport scheme for water vapour, cloud water and trace substances, a new radiation scheme (ECMWF) with modifications concerning the water vapour continuum, cloud optical properties and greenhouse gases, a new formulation of the vertical diffusion coefficients as functions of turbulent kinetic energy, and a new closure for deep convection based on convective instability instead of moisture convergence. Minor changes concern the parameterizations of horizontal diffusion, stratiform clouds and land surface processes. Also, a new dataset of land surface parameters have been compiled for the new model. The climatology of the model, derived from two extended AMIP simulations at T42L19 resolution, is documented and compared with ECMWF operational analyses. Some of the biases noted for the previous model version remain virtually unchanged. For example, the polar upper troposphere and lower stratosphere is much too cold, and the zonal wind errors become very large above the 200 hPa level. Furthermore, the low-frequency variability is still too small but the errors are reduced by about 50% compared to ECHAM-3. The tropospheric temperature and zonal wind errors are generally smaller than in the previous model, except for the tropics, where the overestimation of Walker-type circulations in the equatorial plane is even more pronounced in the new model and the simulation of the Indian summer monsoon is less realistic. The most substantial improvements, compared to ECHAM-3, are found for the land surface climate. The temperature and precipitation errors are generally smaller than before, and the biome distributions derived from these parameters are more realistic in ECHAM-4. These improvements can be attributed to an improved represention of surface radiation fluxes via larger absorption of solar radiation in the atmosphere due to both water vapour and clouds.