next up previous
Next: The energy balance equation Up: A proposal for Previous: Introduction

Tasks of a land-surface scheme coupled to a GCM

This attempt to define a general interface between land-surface processes and the atmosphere will be limited to fluxes of energy (radiation, sensible and latent heat) and momentum. This restriction is motivated by the known importance of these fluxes for atmospheric processes [Polcher, 1997] but it should not present any difficulty to include, at a later stage, into such an interface the fluxes of passive tracers or chemical components. This will be required when LSSs close the carbon cycle in coupled ocean atmosphere models or when they provide the source terms for aerosols and chemical species.

Before defining the interface, the tasks of the LSS in determining the surface energy fluxes must be decided. In current GCMs the tasks of the LSS have evolved over time and they are not the same in all models. For instance, albedo calculations have been considered part of the radiation scheme in the past. The albedo computations have shifted towards the land surface scheme in more recent models [Chalita and Le Treut, 1994, Douville et al., 1995], following the recognition that albedo depends on fast surface processes, such as the presence of snow on vegetation. Thus some GCMs will include these calculations in the radiation scheme while others include them in the LSS. These differences are partly historical, as new developments in land-surface modeling had to be put into the older context of the GCM, avoiding costly rewriting of code. The coupling between the surface and the atmosphere has very often been guided by numerical and computational constraints rather than theoretical consideration on the physical system to be solved.

For this interface we will define the tasks of the LSS based on theoretical consideration on the closure of the atmospheric processes which interact with the surface. The LSS should provide the lower boundary conditions for all atmospheric processes but there are two ways of achieving this, the Neumann and the Dirichlet closures [Richter, 1978]. In the former case the fluxes have to be given to the atmosphere, thus leading to more complex computations within the surface scheme than would be needed if only state variables were provided as is the case for the Dirichlet closure. In the cases discussed here, the Dirichlet and Neumann closures are equivalent from an analytical point of view but not in their numerical implementation. When the discretized equations are considered, both closures will induce discontinuities in the systems but at different locations. In some cases a mixed Dirichlet Neumann closure is also possible and leads to a simultaneous solving of the atmospheric processes and the surface conditions. Physical considerations on the role of the fluxes considered and the characteristic time-scales will allow us to decide which solution is best when modeling the interactions between the surface and the atmosphere. Once this is decided the role of the surface scheme will be well defined.

While defining the tasks of the LSS and its interface to the GCM attention will be paid to energy conservation. The number of operations needed on fluxes of energy will be held to a minimum, thus reducing the risk of errors on either side of the interface. We will also be guided by a set of more practical considerations :

Their role is to ensure that the resulting interface is robust and easy to use.

In defining such a coupling scheme there is always the temptation for variables computed in the GCM, but which could also be recomputed in the LSS to postpone the issue and allow both solutions in the interface. This can be achieved by increasing the redundancy in the interface but at the cost of major problems in its practical use. The interface will be more complex, more difficult to use and increase the risk of errors. Thus an effort will be made to reduce redundancies.

The resulting interface will be just another compromise but perhaps one built on better theoretical grounds and which can be used to couple with a minimum of work and a maximum likelihood of success any GCM to any other LSS.




next up previous
Next: The energy balance equation Up: A proposal for Previous: Introduction

POLCHER Jan
Fri Mar 6 16:09:11 MET 1998