The exchange, or flux, of heat between the oceans and atmosphere is an important driver of the global oceanic and atmospheric circulations but remains poorly quantified. Direct measurement of heat flux remains a research activity and so global heat flux datasets are generated using observations of winds, air and sea temperatures, and humidity as input to heat flux parameterizations known as “bulk formulas.” We remain dependent on the observations from merchant ships in the Voluntary Observing Ships (VOS) program, which are archived in the International Comprehensive Ocean-Atmosphere Dataset (ICOADS); measurements from buoys are sparse and satellites cannot accurately recover all the variables required for heat flux calculation.
Careful analysis of VOS data is necessary to produce gridded datasets of meteorological variables and fluxes with the accuracy required for climate research. Past in situ flux datasets have averaged observations on monthly timescales to reduce random uncertainty. It has therefore been hard to understand the contributions to observed variability from measurement errors, poor sampling, or natural variability. The new dataset, which covers the period 1973 to 2006, avoids this problem by first constructing daily mean fields using optimal interpolation. This allows each component of variability to be handled correctly and, for the first time, uncertainty estimates to be produced. New bias adjustments have also been developed and applied. The new dataset is described and a preliminary comparison with flux estimates from moored buoys, satellites, and atmospheric reanalysis models is presented.
National Oceanography Centre Southampton, Southampton, Southampton, United Kingdom