A comprehensive global root database is used to derive vertical root distri
bution and rooting depth for various vegetation categories in one of the mo
st widely-used land models; i.e., the Biosphere-Atmosphere Transfer Scheme
(BATS). Using a variety of observational datasets, observed root distributi
on is found to significantly improves the offline simulation of surface wat
er and energy balance. Global climate modeling further demonstrates that ob
served root distribution primarily affects latent heat flux and soil wetnes
s over tropical and midlatitude land, respectively.