“Trends” and variations of global oceanic evaporation data sets from remote sensing

The variability in global oceanic evaporation data sets was examined for the period 1988-2000. These data sets are satellite estimates based on bulk aerodynamic formulations and include the NASA/Goddard Space Flight Center Satellite-based Surface Turbulent Flux version 2 （ GSSTF2）, the Japanese-ocean flux using remote sensing observations （J-OFURO）, and the Hamburg Ocean-Atmosphere Parameters and Fluxes from Satellite version 2 （HOAPS2）. The National Center for Environmental Prediction （NCEP） reanalysis is also included for comparison. An increase in global average surface latent heat flux （SLHF） can be observed in all the data sets. Empirical mode decomposition （EMD） shows long-term increases that started around 1990 for all remote sensing data sets. The effect of Mt. Pinatubo eruption in 1991 is clearly evident in HOAPS2 but is independent of the longterm increase. Linear regression analyses show increases of 9.4%, 13.0%, 7. 3%, and 3.9% for GSSTF2, J-OFURO, HOAPS2 and NCEP, for the periods of the data sets. Empirical orthogonal function （EOF） analyses show that the pattern of the first EOF of all data sets is consistent with a decadal variation associated with the enhancement of the tropical Hadley circulation, which is supported by other satellite observations. The second EOF of all four data sets is an ENSO mode, and the correlations between their time series and an SO1 are 0.74, 0.71,0.59, and 0.61 for GSSTF2, J-OFURO, HOAPS2, and NCEP in that order. When the Hadley modes are removed from the remote sensing data, the residue global increases are reduced to 2.2% , 7. 3%, and 〈 1% for GSSTF2, J-OFURO and HOAPS, respectively. If the ENSO mode is used as a calibration standard for the data sets, the Hadley mode is at least comparable to, if not larger than, the ENSO mode during our study period.