Assessing the Sensitivity of A Land‐Surface Scheme to Parameters Used In Tropical‐Deforestation Experiments

A series of factorial experiments is used to elucidate the factors and multifactor interactions important for the performance of one complex land‐surface scheme in a tropical environment. the BATS (Biosphere‐Atmosphere Transfer Scheme) is found to be sensitive to atmospheric factors which are currently poorly predicted by global climate models (GCMs) and, more importantly, to combinations of factors rendering tuning for good performance when coupled into today's global models (a worthless practice if the tropical area is subjected to different conditions such as those induced by deforestation). Similarly, two‐factor interactions, between coded ecological parameters (here roughness length and short wave, <0.7 μm, vegetation albedo), mean even if the BATS characterization and its driving GCM perform well in today's tropical‐forest environment, altering either or both parameters, as will occur in a deforestation simulation, must greatly diminish confidence in representations of future climate.