The role of the land surface (soil, vegetation, snow, permafrost and land ice) was discussed in detail in the SAR. The SAR noted that improvements had occurred in our ability to model land-surface processes but that there was a wide disparity among current land-surface schemes when forced by observed meteorology. Our physical understanding of the role of land-surface processes within the climate system was discussed in Chapter 7, Section 7.4.1.
Most of the effort in trying to reduce the disparity in land-surface scheme performance has been performed in offline intercomparisons under the auspices of International Geosphere Biosphere Programme (IGBP) and the World Climate Research Programme (WCRP). Due to the difficulties in coupling multiple land-surface schemes into climate models (see Section 220.127.116.11) specific endeavours have been: the Project for the Intercomparison of Land-surface Parametrization Schemes (PILPS), the Global Soil Wetness Project (GSWP), the International Satellite Land Surface Climatology Project (ISLSCP), and the Biological Aspects of the Hydrological Cycle (BAHC) (e.g., Henderson-Sellers et al., 1995; Polcher et al., 1996; Dirmeyer et al., 1999; Schlosser et al., 2000). In comparisons between offline simulation results and observations, difficulties in partitioning available energy between sensible and latent heat and partitioning of available water between evaporation and runoff were highlighted (e.g., Chen et al., 1997; Schlosser et al., 2000). While we are far from a complete understanding of why land surface models differ by such a large degree, some progress has been made (e.g., Koster and Milly, 1997; Desborough, 1999). Significant progress has also been made in adding physical processes into land-surface models (see Chapter 7, Section 7.4). Where some observations exist (e.g., for incoming solar radiation, net radiation and soil moisture), an evaluation of the ability of current climate models to simulate these quantities suggests that significant problems remain (Wild et al., 1997; Garratt et al., 1998; Robock et al., 1998). The evaluation of surface processes in climate models tends to focus on monthly and, less commonly, daily quantities. An evaluation of the ability of climate models to simulate land-surface quantities at the diurnal scale has yet to be performed systematically, although some efforts have been initiated since the SAR (e.g., Watterson, 1997).
Figure 8.11: Illustration of the range of snow cover extent in CMIP1 model simulations listed in Table 8.3: Northern Hemisphere, DJF. The figure is constructed similarly to Figure 8.10 based on the prescribed 1 cm cutoff. The observed boundary is based on Foster and Davy (1988).
Work since the SAR has also focused on trying to identify the relative significance of land-surface processes in comparison with other components of climate models. The sophistication in the representation of the land surface in coupled models is varied (see Chapter 7 and Table 8.1). The addition of more realistic plant physiology in some land-surface models (e.g., Bonan, 1995; Sellers et al., 1996 (See Chapter 7, Section 7.4)) which permits the simulation of carbon dioxide (CO2) and gas isotope fluxes, provides the opportunity to compare these quantities with local scale, regional scale and global scale observations from flux towers and satellites.
Snow, and the snow albedo feedback, are important components of the land surface. Current climate models incorporate snow in varying degrees of sophistication and there is currently major uncertainty in the ability of land-surface schemes to simulate snow mass or cover (see Chapter 7, Section 7.5). Frei and Robinson (1998) evaluated the simulation of monthly mean snow extent from 27 AMIP AGCMs and found weaknesses in the simulation of the seasonal cycle of snow extent and a general underestimation in interannual variability. These weaknesses limit confidence in the simulation of mid- and high latitude changes simulated by current climate models, since a failure to simulate snow accurately tends to impact significantly on albedo, surface roughness length and soil moisture (and therefore precipitation on subsequent seasons).
The Northern Hemisphere snow simulations of fourteen global coupled models contributed to CMIP are summarised in Figure 8.11 and Table 8.3. Figure 8.11 (constructed similarly to Figure 8.10) provides a visual presentation of the range in simulated (land only) snow extent. The relative error in simulated snow extent is larger in summer than in winter. There is no obvious connection between either flux adjustment or land-surface scheme and the quality of the simulated snow extent.
Other components of the land surface potentially important to climate change include lateral water flows from the continents into the ocean, permafrost, land-based ice and ice sheets. Some land-surface schemes now include river routing (e.g., Sausen et al., 1994; Hagemann and Dümenil, 1998) in order to simulate the annual cycle of river discharge into the ocean. This appears to improve the modelling of runoff from some large drainage basins (Dümenil et al., 1997), although water storage and runoff in regions of frozen soil moisture remain outstanding problems (Arpe et al., 1997; Pitman et al., 1999). River routing is also useful in diagnosing the representation of the hydrological cycle in models (e.g., Kattsov et al., 2000). There has been limited progress towards developing a permafrost model for use in climate models (e.g., Malevsky-Malevich et al., 1999) although existing simple models appear to approximate the observed range in permafrost reasonably well (e.g., Volodin and Lykosov, 1998). The dynamics of ice sheets and calving are not presently represented in coupled climate models. To close the fresh water budget for the coupled system, fresh water which accumulates on Greenland and Antarctica is usually uniformly distributed either over the entire ocean or just in the vicinity of the ice sheets (e.g., Legutke and Voss, 1999; Gordon et al., 2000). The impact of these limitations has yet to be investigated.
Other reports in this collection