Systematic errors in air temperature are believed to be caused primarily by the heating of the ship superstructure, causing biases of 0.4°-0.8° C in some regions of the tropics (Isemer and Hasse 1987 and references within). In view of this, some authors (Folland et al. 1984) have preferred to discard daytime observations. Although night time observations may be adequate to study long term trends in air temperature, we feel that it significantly reduces the number of observations and tends to bias the data toward a generally lower night time temperature.
Kent et al. (1993b) devised a method of correction for daytime
surface air temperature based on wind speed and the incoming short
wave radiation. Preliminary calculations by these authors indicate
correcting daytime air temperature results in increases in mean
sensible and latent heat fluxes of 3.3 and 1.0 W/m, respectively,
in the North Atlantic Ocean (Kent and Taylor 1995). We
intend to include such a correction in future versions of our data
set, but we do not do so at this time.