A Simple Ocean Data Assimilation analysis of the global upper ocean 1950-95. Part I: Methodology

The authors describe a 46-year global retrospective analysis of upper-ocean temperature, salinity, and currents. The analysis is an application of the Simple Ocean Data Assimilation (SODA) package. SODA uses an ocean model ba sed on Geophysical Fluid Dynamics Laboratory MOM2 physics. Assimilated data includes temperature and salinity profiles from the World Ocean Atlas-94 ( MBT, XBT, CTD, and station data), as well as additional hydrography, sea su rface temperature, and altimeter sea level. After reviewing the basic methodology the authors present experiments to ex amine the impact of trends in the wind field and model forecast bias (refer red to in the engineering literature as "colored noise"). The authors belie ve these to be the major sources of error in the retrospective analysis. Wi th detrended winds the analysis shows a pattern of warming in the subtropic s and cooling in the Tropics and at high latitudes. Model forecast bias res ults partly from errors in surface forcing and partly from limitations of t he model. Bias is of great concern in regions of thermocline water-mass for mation. In the examples discussed here, the data assimilation has the effec t of increasing production of these water masses and thus reducing bias. Additional experiments examine the relative importance of winds versus subs urface updating. These experiments show that in the Tropics both winds and subsurface updating contribute to analysis temperature, while in midlatitud es the variability results mainly from the effects of subsurface updating.