An intercomparison between the surface heat flux feedback in five coupled models, COADS and the NCEP reanalysis
MetadataShow full item record
The surface heat ﬂux feedback is estimated in the Atlantic and the extra-tropical Indo-Paciﬁc, using monthly heat ﬂux and sea surface temperature anomaly data from control simulations with ﬁve global climate models, and it is compared to estimates derived from COADS and the NCEP reanalysis. In all data sets, the heat ﬂux feedback is negative nearly everywhere and damps the sea surface temperature anomalies. At extratropical latitudes, it is strongly dominated by the turbulent ﬂuxes. The radiative feedback can be positive or negative, depending on location and season, but it remains small, except in some models in the tropical Atlantic. The negative heat ﬂux feedback is strong in the mid-latitude storm tracks, exceeding 40 W m–2 K–1 at place, but in the Northern Hemisphere it is substantially underestimated in several models. The negative feedback weakens at high latitudes, although the models do not reproduce the weak positive feedback found in NCEP in the northern North Atlantic. The main differences are found in the tropical Atlantic where the heat ﬂux feedback is weakly negative in some models , as in the observations, and strongly negative in others where it can exceed 30 W m–2 K–1 at large scales, in part because of a strong contribution of the radiative ﬂuxes, in particular during spring. A comparison between models with similar atmospheric or oceanic components suggests that the atmospheric model is primarily responsible for the heat ﬂux feedback differences at extra-tropical latitudes. In the tropical Atlantic, the ocean behavior plays an equal role. The diﬀerences in heat ﬂux feedback in the tropical Atlantic are reﬂected in the sea surface temperature anomaly persistence, which is too small in models where the heat ﬂux damping is large. A good representation of the heat ﬂux feedback is thus required to simulate climate variability realistically.
CitationClimate Dynamics 22(4): 373-388
Copyright Springer-Verlag 2004