LAND-SURFACE CHARACTERIZATION IN GREENHOUSE CLIMATE SIMULATIONS

A simplified Holdridge-type vegetation prediction scheme has been coup led to a version oi the NCAR community climate model (CCM1-Oz) that in cludes the biosphere-atmosphere transfer scheme (BATS) and a mixed-lay er ocean. This interactive vegetation climate model has been used to c onduct two complementary CO2-doubling experiments: an instantaneous 2 x CO2 simulation (15 years in total) and a fast, transiently increasin g CO2 simulation (45 years in total). There are some differences in th e predicted vegetation distributions and areas. However, there is agre ement that in a warmed world the vegetation type termed 'agriculture' increases in area at the expense of deciduous needle-leaf trees and sh ort grass; and the tundra extent, already underestimated, decreases fu rther whereas deserts and the deciduous broadleaf tree areas expand. T he overall vegetation areas predicted are not particularly sensitive t o initialization, although effects of different initialization can be monitored for 1-2 years. On the other hand, when the sensitivity of th e modelled climate to the inclusion of some aspects of an interactive biosphere is examined, it is found that annually updated continental c haracteristics do not disrupt the climate simulation but do modify zon al temperatures and precipitation and increase continental evaporation . The latter intensifies the Hadley circulation, especially in July, a nd, thus, leads to increased evaporation globally. These results, if c orroborated by other similar studies, indicate that simple, post facto application of vegetation diagnostic schemes once climatic equilibriu m is achieved may be diagnosing vegetation from an incorrect climatic state.