The rapid development of the scientific research agenda calls for a different GLASS organizational structure. The difference between local and global coupling is not straightforward, as many processes that play a role in land-atmosphere interaction have both local and regional aspects. Many feedbacks shown on global maps are partially an expression of local land-atmosphere coupling, which cannot be clearly defined without consideration of interactions with the moving, advecting atmosphere. In addition, it was realized that the realm of data assimilation should get a renewed incentive, as model development and verification increasingly make use of modern statistical techniques and new data. Finally, the development of land models raises the question as to which level the available observations contain enough degrees of freedom to adequately attribute single processes to the overall land state or flux.
In analyzing the structure of GLASS, it was noted that a critical assessment of the expected performance and intrinsic predictability of land processes in models was missing. In addition, GLASS model parameterization and evaluation activities were missing that are necessary for proper assessment of the true added value of land model updates. For this purpose, a protocol needed to be developed in which comparisons of models and data against benchmarks was a key component. To accomplish these new requirements, three new working groups were created: (1) Benchmarking Land-Surface Models and Observations; (2) Model Data Fusion; and (3) Land-Atmosphere Coupling.
The goal of this working group is to develop a protocol for evaluating experiments to address the central question, “Does my land-surface model describe the processes in the climate system sufficiently well?” This implies that one needs to define what is “sufficiently well.” Recent studies have compared the skill of state-of-the-art land surface models to statistical models or neural networks calibrated on forcing data alone. This is one example of a test in which we can understand the information content actually added by the land-surface models.
Activities in land data assimilation and procedures to calibrate and optimize models by systematic confrontation with observations are the focus of this working group. It will aim at the development of an optimal system to create global land-surface data sets in which information is extracted from both land-surface models and sophisticated observations.
An important activity of this working group is the estimation of the contribution of memory in the land system to the overall predictability of regional atmospheric phenomena at seasonal time scales. In addition, changes in this predictability under changed climate conditions, and the creation of a systematic hierarchy of levels of land-atmosphere coupling, will be coordinated by this working group.