An attempt has been made to evaluate the reliability of water vapor transport estimates from atmospheric general circulation models (GCMs) by comparing the vapor export from the Atlantic basin obtained using the GISS 4-degrees x 5-degrees grid GCM with that obtained from observations of wind velocity and humidity [Oort, 1983]. We find substantial differences. The model gives an export of 0.13 Sverdrups (Sv) from the Atlantic basin compared to 0.32 Sv from the Oort data set. We note that the GCM produces far stronger easterly winds and correspondingly larger tropical vapor transports. Further, because of the model's low orography, steering of the winds and drying of air masses by mountain chains are not adequately represented. This leads to a dramatic overestimation of water vapor transport across mountain ranges. In an additional model experiment, we calculated the water vapor transports in a double-CO2 scenario. We find that atmospheric freshwater loss from the Atlantic basin is increased to 0.30 Sv compared to 0.13 Sv in the control run. Because of the deficiencies of the model mentioned above, however, we conclude that estimates of vapor transport for future greenhouse conditions and past glacial conditions made using GCMs have to be taken with great caution until models with more realistic flow patterns, especially around mountain ranges, are available.
Hg561Times Cited:28Cited References Count:40