In this paper we have identified the variables which should be exchanged between a general circulation model and its land surface scheme. An effort was made not to take into account practical considerations, such as the current state of GCMs and LSSs, but rather to rely on physical and numerical arguments with the aim of being as general as possible. The resulting list of variables to be exchanged are presented in tables 1 and 2.
The authors are not aware of any reason why this general interface could not be implemented in all current GCMs and LSSs with a minimal effort. In some GCMs it will mean splitting the calculations of the planetary boundary and some LSSs will need to gather all computations related to the surface and which are now distributed throughout the GCM. This effort will make it possible to exchange land-surface scheme between GCMs thus opening new doors to intercomparison projects such as AMIP [Gates, 1992] and PILPS [Henderson-Sellers et al., 1996]. It has to be expected that in the years to come progress will be made on the radiation or vertical diffusion schemes used in GCMs. This may lead to a rethinking of certain aspects of the coupling and eventually to an extension of the interface.
The interface defined here is general enough to be also applied to the coupling of the atmosphere to the ocean and sea-ice. Beside offering the plug-compatibility of surface parameterizations to the GCM it also has the advantage of making the lower boundary conditions a truly autonomous model within the GCM. This opens a new range of possibilities for the development and validation of the schemes at a global scale.
This proposal is only a first step towards a ``plug compatibility'' of land-surface schemes. It will certainly evolve as LLSs and GCMs evolve and as it is implemented. In order to give the interested reader the opportunity to follow the evolution a Web page exists at the following location: http://www.lmd.jussieu.fr/pilps4c.html. It will give informations on the evolution of the interface, its current status, where it is used and which points are currently being discussed.