92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Wednesday, 25 January 2012: 9:30 AM
An Investigation of Tropical Atlantic Bias in a High-Resolution Coupled Regional Climate Model
Room 355 (New Orleans Convention Center )
Christina M. Patricola, Texas A&M University, College Station, TX; and M. Li, Z. Xu, P. Chang, R. Saravanan, and J. S. Hsieh

Coupled atmosphere-ocean general circulation models (AOGCMs) commonly fail to simulate the boreal summer cold tongue in the eastern equatorial Atlantic and produce a westerly equatorial trade wind bias. This tropical Atlantic bias problem is investigated with a high-resolution (27-km atmosphere represented by the Weather Research and Forecasting Model, 9-km ocean represented by the Regional Ocean Modeling System) coupled regional climate model.

Uncoupled atmospheric simulations conducted to test climate sensitivity to cumulus, land-surface, planetary boundary layer, microphysics, and radiation parameterizations reveal that the radiation scheme has a pronounced effect in the tropical Atlantic. The CAM radiation simulates a severe dry precipitation (up to -90%) and cold land-surface temperature (up to -8 K) bias over the Amazon related to an over-representation of low-level clouds, and an almost basin-wide westerly trade wind bias. The RRTM and Goddard radiation simulates a doubling of Amazon and Congo basin precipitation, but a relatively weak trade wind bias confined to the eastern Atlantic.

High-resolution regional coupled model experiments indicate that the warm eastern equatorial Atlantic sea surface temperature (SST) bias is more sensitive to the local rather than basin-wide trade wind bias and to a wet Congo basin instead of a dry Amazon – a finding that differs from previous AOGCM studies. Comparisons between coupled and uncoupled simulations suggest that a local Bjerknes feedback confined to the far-eastern equatorial Atlantic amplifies the SST, wind, and deepened thermocline bias in this regional coupled model. The SST bias in one regional model configuration is similar to the typical AOGCM bias indicating that increasing resolution is not a simple solution to this problem.

Supplementary URL: