On the Role of the Land Surface Representation and the Numerical Coupling to the Atmosphere for the Simulated Climate of the Global ECHAM4 Model

The continental surfaces replesellt an important component of the Barth's clirnate system. ÌVleteoroiogical models for climate simulations or numerical rveatirer prediction therefore re- quire a realistic description of the land surface processes. The degree of complexity needed for these land surface schemes is not yet completely deterrnined. Another aspect of the meteorological models is the numerical realization of the coupling betrveen land surface and atmosphere. This thesis investigates the sensitivity of the simulated climate to the para- meterization of land surface plocesses compared to the effect of different numerical coupling techniques betrveen land surface and atmosphere. For this reason, a detailed evaluation of the two land surface schemes BCHAM and SBCHIBA in a hierarchy of model set-ups, from off-line through one-dimensional to global three-dimensional, is conducted. In the off- line experiments, ECHAM shows deficiencies in modelling the diurnal variations of surface temperature and ground heat flux. This is due to the conceptual inconsistency in that the top soil layer temperature is both used as part of the soil ternperature flnite difference scheme and also as surface value for computing the atmospheric surface enelgy fluxes. This is improved in SECHIBA by an extrapolation of heat capacity and ground heat flux torvard the surface. The standard ECHAM4 climate model utilizes a semi-implicit coupling techniclue betrveen land surface and atmosphere in a way in which energy at the land surface-atmosphere inter- face is not conserved. This is a major deficiency. Trvo nerv model versions were developed as part of this thesis: BCHAM4/INIPI and ECHAM4/SECHIBA. They incorporate an implicit coupling technique which conserves enelgy. BCHAM4 and trCHAM4/IMPL are identical rvith respect to all physical parameterizations they apply; the only difference is the coupling. In trCHAM4/SECHIBA the ECHAIVI land surface scheme rvas replaced by StrCHIBA. The intercornparison of one-dimensional versions of these three models shows that the energy residual term in BCHAM4, which is part of the semi-implicit coupling and represents an error in the surface/atmosphere energy balance, is not negligibly small. Rather, it is of the order of the physical fluxes and therefore serves as an artifrcial (numerical) sink or source of energy at the sutface, signifrcantlv altering the surface energy balance. Biases of more than 1300 W/m2 are found due to the coupling techniclue. These are avoided in ECHANI4IIMPL, rvhich results in a more pronounced diurnal cycle of surface temperature and generally higher temperature maxima during a rvarming phase. In a global-scale intercomparison of the three models a signiflcant irnpact of the altered coupling techniclue and land surface scheme on most surface variables and atmospheric surface fields is found. The surface air temperature over large continental areas in the l{orthern Hemisphere in summer is higher by ,tp to 3-5oC in the two implicit models than in ECHAM4. For trCHAM4/IMPL this is attributed to the closure of the surface energy balance. This allows to use energy amounts of regionally more than 40 W l^'in the seasonal mean, rvhich are lost in ECHAM , for physical processes, e. g. for heating the land surface and lower atmosphere. In BCHAM4/SBCHIBA a further increased surface air temperature of up to 3oC is due to a different snorv parameterization that allows an earlier snow melt. This is in good agreement with snow observations. Due to the changed temperature stlucture a considerably improved Asian summer monsoon circulation with respect to stream patterns, wind velocities and associated precipitation distributions is simulated by the implicit models. Furthermore, the computed evaporation in StrCHIBA is more realistic, compared to measurements at some European sites, as consecluence of a more sophisticated representation of vegetation