Climate models depend on evapotranspiration from models of plant stoma
tal resistance and leaf cover, and hence they depend on a description
of the response of leaf cover to temperature and soil moisture. Such a
description is derived as an addition to the Biosphere-Atmosphere Tra
nsfer Scheme and tested by simulations in a climate model. Rules for c
arbon uptake, allocation between leaves, fine roots, and wood, and los
s terms from respiration, leaf, and root turnover and cold and drought
stress, are used to infer the seasonal growth of leaf area as needed
in a climate model, and to provide carbon fluxes (assuming also a simp
le soil carbon model) and net primary productivity. The scheme is test
ed in an Il-yr integration with the NCAR CCM3 climate model. After a s
pinup period of several years, the model equilibrates to a seasonal cy
cle plus some interannual variability. Effects of the latter are notic
eable for the Amazon. Overall, drought stress has nearly as large an e
ffect on leaf mortality as cold stress. The leaf areas agree on averag
e with those inferred from Normalized Difference Vegetation Index alth
ough some individual systems are either too high (grass and crops) or
too low (deciduous needleleaf in Siberia) compared to the satellite da
ta. Evergreen needleleaf forests have significantly smaller annual ran
ge and later phase than indicated by the data. The interactive paramet
erization increases temperatures and reduces evapotranspiration and pr
ecipitation compared to the control over the extratropical Northern He
misphere summer. This interactive leaf model may serve not only to pro
vide feedbacks between vegetation and the climate model, but also to d
iagnose shortcomings of a climate model simulation from the viewpoint
of its impact on the biosphere.