Global vegetation cover changes from coarse resolution satellite data

Land cover plays a key role in various biophysical processes related to global climate and terrestrial biogeochemistry. Although global land cover has dramatically changed over the last few centuries, until now there has been no consistent way of quantifying the changes globally. In this study we used long-term climate and soils data along with coarse resolution satellite observations to quantify the magnitude and spatial extent of large-scale land cover changes attributable to anthropogenic processes. Differences between potential leaf area index (LAI), derived from climate-soil-leaf area equilibrium, and actual leaf area index obtained from satellite data are used to estimate changes in land cover. Further, changes in LAI between potential and actual conditions are linked to climate by expressing them as possible changes in radiometric surface temperatures (T )̂ resulting from changes in surface energy partitioning. As expected, areas with high population densities, such as India, China, and western Europe showed large reductions in LAI. Changes in global land cover expressed as summer, midafternoon T„ ranged from -8 ° to -F16'’C. Deforestation resulted in an increase in T̂ , while irrigated agriculture reduced the T̂ . Many of the current general circulation models (GCMs) use potential vegetation maps to represent global vegetation. Our results indicate that there are widespread changes in global land cover due to deforestation and agriculture below the resolution of many GCMs, and these changes could have a significant impact on climate. Potential and actual LAI data sets are available for climate modelers at 0.5° X 0.5° resolution to study the possible impacts of land cover changes on global temperatures and circulation patterns. Introduction Rem otely sensed data have been successfully used to study land cover changes at fine to coarse spatial scales. For example. Previous studies have demonstrated that changes in land Thematic Mapper data at 30-m resolution were used in the cover could be as important as the increase in atmospheric estimation o f urban expansion [Haack et al., 1987] and deforgreenhouse gases in climate change [Bonan et a l , 1992; Shukla [Green and Sussman, 1990; Skole and Tucker, 1993]. et al., 1990]. For example, deforestation has been found to ^and, advanced very high resolution radiometer increase land surface temperatures (T J and reduce rainfall, ( a v h R R ) data at 1and 16-km resolutions were used to suggesting a warmer and drier climate [Dickinson and Hendcontinental scales [Malingreau and T c Z : I ’ ? u r f "1 ’ Tucker, 1988; Tucker et al., 1985; Turner et al., 1993; Houghton 1990]. Although I t IS clear that global land cover has been dramatically altered by human habitation, forest clearing, and j • j r i r r i . , , , 1 .n . derived from the coarse-resolution A V H R R , therefore useful agriculture, until now there has been no means o f quantifying i . j i i i. ■ i j i j ,̂7 , , , „ 'I j 0 jQ study large-scale changes in land cover \Townshend et ^ e c anges g o a y. 1991; Goward et al,, 1993]. Earlier studies used satellite Quantitative analysis o f the incredible diversity in global , . . , u u-.. j 1-r r \ 1* data primarily for land cover classification. However, recent vegetation (species, growth habits, and life forms) IS a difficult , problem [Running et al., 1994a]. Many classification efforts f ‘^at satellite data could be used to quantify have been attempted, and their results show wide variations in biophysical variables such as leaf area index (LAI), canopy current estimates of global land cover [Townshend et al., 1991]. Absorbed Photosynthetically Active A detailed discussion o f various land cover classification Radiation [Nemani and Running, 1989; Pierce et al., 1993; schemes and their problems is given by Townshend et al. [1991] 1994, Asrar et al., 1992]. and Running et al. [1994a]. The consensus from the above Land cover is usually quantified by som e measure o f plant studies is that in the advent o f rapid anthropogenic changes, density. LAI, area of leaves per unit ground area, provides a only remote sensing data can provide accurate and repeatable simple measure o f plant canopy density. LAI varies from less means of land cover analysis and monitoring. ‘ban 1 in deserts to greater than 10 over tropical rain forests; therefore changes in LAI can also be indicative o f land cover C ln r iV r if fh it iQ Q f t hv / t h e : A m p r i r n n T T n i i^ n ------------rT ijr_ . j p between climateand soil-defined vegetation (potential) and actual vegetation. By using a biophysical variable (LAI), we assumed the influence o f the inconsistencies in current vege­ tation classification schem es could be minimized. The purpose o f this paper is to present a quantitative anal­ ysis o f the spatial distribution o f global land cover changes. Our focus is on anthropogenic changes in land cover, which have occurred over extensive areas, between potential (natural conditions) and present vegetation.