The variability of sea surface temperature and the impact it has upon climate modelling

A set of in situ meteorological and radiometric measurements is acquired to facilitate the investigation of skin sea surface temperature (SSST) variability. The Tasco THI-500L radiometer produces SSST accurate to 0.5K when deployed in-shore and calibrated every 82 hours. Radiometer performance is enhanced if the instrument is insulated from, and characterised for, the effect of solar heating, allowing operational deployment to improve the availability of in situ SSST.;Two statistical tests for similarity of populations are evaluated as quantifiers of sea surface temperature variability. Violations of the parametric requirements of the analysis of variance F-Test produce unreliable results. A dependable measure of SSST variability is generated using the non-parametric Mann-Whitney U-Test. Applying this test to imaging radiometer measurements shows that U depends on the variation between radiometric images in the physical factors that govern the ocean-atmosphere heat fluxes. The implications of this dependency for satellite SSST validation and climate modelling are considered. Analysis over metre- to kilometre- scales produces three trends of SSST variability against measurement interaction time. No dependency of these trends on meterological conditions is found.;Applying the U-Test to large-scale ATSR-1 ASST data of the Tropical Pacific Ocean finds less variability than in small-scale SSST. Large-scale SSST are shown to be inhomogeneous with co-incident bulk sea surface temperature (BSST), implying that the use of ATSR-1 SSST in climate models will impact on the model results.;Forcing a model of the Tropical Pacific Ocean separately with SSST and BSST shows a divergence in model output for both seasonal and ENSO SST variability. The BSST output is 0.7K warmer than the SSST output in the central equatorial Pacific during the Spring warming period.