Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

7.4.2 Land-Surface Change

Figure 7.5:
Schematic showing relationships between a simulation of the Atmospheric Boundary Layer (ABL), a Land-Surface Parametrization (LSP), vegetation and soil properties, and anthropogenic change. Interactions are shown by broad white arrows marked with capital letters, fluxes by grey arrows, and dependencies by dotted lines. (A) Diurnal-seasonal interactions between the ABL and the LSP; the ABL variables of air temperature, humidity, downward short-wave radiation, downward long-wave radiation, wind speed and precipitation (T, q, S ,L , u, P) are used to force the LSP which calculates net radiation minus ground heat, sensible heat, and latent heat fluxes ( Rn -G, H, lE), which in turn feed back to the atmosphere. Three surface parameters in the LSP are critical to these calculations: Albedo and surface roughness (a,Z0 ) determine the radiative balance and turbulent exchange regime, and in third generation LSPs, the canopy conductance term, gc (equivalent to the summation of all the leaf stomatal conductances) determines the vegetation evapotranspiration rate (lE) and net photosynthetic rate (Pnet ). On time-scales of minutes to hours, gc is a direct function of T, q, S ,CO2 concentration and soil moisture (W). Increasing CO2 concentration can act to signifi-cantly reduce gc and hence limit lE. The maximum value of gc is determined by parameters related to vegetation density or leaf area index (LAI), and biochemical capacity (Vmax ). Long-term climatic forcing (B) and land-use change (C) can alter the vegetation type and density, soil properties and ecosystem respiration rates, Rd , by which carbon is returned to the atmosphere from the vegetation and soil. (D) Changes in vegetation properties affect Vmax and LAI, and changes in soil properties affect soil moisture (W) and runoff (R0 ).

Climate and carbon cycle simulations extending over more than a few decades must take account of land-surface change for two main reasons. First, changes in the physical character of the land surface can affect land-atmosphere exchanges of radiation, momentum, heat and water (see Figure 7.5 and the simulation studies discussed below). These effects must be allowed for within climate simulations or analyses to avoid confusion with the effects of global warming. Second, changes in vegetation type, density and associated soil properties usually lead to changes in terrestrial carbon stocks and fluxes that can then directly contribute to the evolution of atmospheric CO2 concentration. Therefore, any historical analysis of the atmospheric CO2 record must estimate these contributions to avoid inaccurate attribution of carbon sinks or sources. Similarly, model simulations extending over the next 50 to 100 years should allow for significant perturbations to the atmospheric carbon budget from changes in terrestrial ecosystems (Woodwell et al., 1998; see also Chapter 3).

There are two types of land-surface change; direct anthropogenic change, such as deforestation and agriculture; and indirect change, where changes in climate or CO2 concentration force changes in vegetation structure and function within biomes, or the migration of biomes themselves. With respect to direct anthropogenic change, population growth in the developing countries and the demand for economic development worldwide has led to regional scale changes in vegetation type, vegetation fraction and soil properties (Henderson-Sellers et al., 1996; Ramankutty and Foley, 1998). Such changes can now be continuously monitored from space, and the satellite data record extends back to 1973. Large-scale deforestation in the humid tropics (South America, Africa and Southeast Asia) has been identified as an important ongoing process, and its possible impact on climate has been the topic of several field campaigns (Gash et al., 1996), and modelling studies (for example, Nobre et al., 1991; Lean et al., 1996; Xue and Shukla, 1996; Zhang et al., 1996a; Hahmann and Dickinson, 1997; Lean and Rowntree, 1997). Some significant extra-tropical impacts have also been identified in several model experiments (e.g., Sud et al., 1996; Zhang et al., 1996b). Replacement of tropical forest by degraded pasture has been observed to reduce evaporation, and increase surface temperature; these effects are qualitatively reproduced by most models. However, large uncertainties still persist about the impact of large-scale deforestation on the hydrological cycle over the Amazon in particular. Some numerical studies point to a reduction of moisture convergence while others tend to increase the inflow of moisture into the region. This lack of agreement occurs during the rainy season and reflects our poor understanding of the interaction of convection and land-surface processes (Polcher, 1995; Zhang et al., 1997a), in addition to the effects of differences between the formulations in the land-surface schemes, their parameter fields, and the host GCMs used in the studies.

Other simulation work has indicated that the progressive cultivation of large areas in the East and Midwest USA over the last century may have induced a regional cooling of the order of 1 to 2°C due to enhanced evapotranspiration rates and increased winter albedo (Bonan, 1999). Snow-vegetation albedo effects significantly influence the near-surface climate; assignment of an open snow albedo value to the winter boreal forest in an NWP led to the prediction of air temperatures that were 5 to 10°C too low over large areas of Canada (Betts et al., 1998). Work has also been done on the interaction between Sahelian vegetation and rainfall that suggests that the persistent rainfall anomaly observed there in the 1970s and 1980s could be related to land-surface changes (Claussen, 1997; Xue, 1997). All these studies indicate that large-scale land-use changes can lead to significant regional climatic impacts. However, it is unlikely that the aggregate of realistic land-use changes over the next 50 to 100 years will contribute to global scale climate changes comparable to those resulting from the warming associated with the continuing increase in greenhouse gases.

Changes to the land surface resulting from climate change or increased CO2 concentration are likely to become important over the mid- to long term. For example, the extension of the growing season in high latitudes (Myneni et al., 1997) will probably result in increases in biomass density, biogeochemical cycling rates, photosynthesis, respiration and fire frequency in the northern forests, leading to significant changes in albedo, evapotranspiration, hydrology and the carbon balance of the zone (Bonan et al., 1992; Thomas and Rowntree, 1992; Levis et al., 1999). There have been several attempts to calculate patterns of vegetation type and density as a function of climate (e.g., Zeng et al., 1999); most of these have made use of climate predictions to calculate the future steady-state distribution of terrestrial biomes but some have attempted to model transitional cases (Ciret and Henderson-Sellers, 1998).

However, over the next 50 to 100 years, it is more likely that changes in vegetation density and soil properties within existing biome borders will make a greater contribution to modifying physical climate system and carbon cycle processes than any large-scale biogeographical shifts. In some cases, soil physical and chemical properties will limit the rate at which biomes can "migrate"; for example, colonisation of the tundra by boreal forest species is likely to be slowed by the lack of soil. Climate-vegetation relations are discussed further in Chapter 8, Section 8.5.5 with respect to past climates.

At present, only limited global data sets for LSPs are available and these need to be further improved. A comprehensive land-use/land cover data set, providing a global time-series of vegetation and soil parameters over the last two centuries at GCM resolution, would be a very useful tool to separate land-use change impacts on regional climate from global scale warming effects. Additionally, for both historical analyses and future projections, there is a need for interactive vegetation models that can simulate changes in vegetation parameters and carbon cycle variables in response to climate change. These proposed fourth generation models are just beginning to be designed and implemented within climate models.

Other reports in this collection