Bradley, Kelley A (2015-05). Analyzing El Nino- Southern Oscillation Sensitivity in a Large Ensemble of Ocean Reanalyses. Master's Thesis. Thesis uri icon


  • A 56-member ensemble of ocean reanalyses is used to explore strong El Nino events in two 5-year periods, 1916 to 1920 and 1996 to 2000, that have markedly different quantities of observations. To generate the 56 forcing fields, we use a 56 member atmospheric reanalysis (20CRv2 system). Prescribed as boundary conditions were 8 different sea surface temperature (SST) estimates from an ocean reanalysis system, SODA with sparse input (SODA.si1), resulting in 8 sets of 7 ensemble members each. The 56 atmospheric reanalyses were used to force an ocean reanalysis for the same two time periods. The ocean reanalyses, SODA_XP, are used to explore ENSO sensitivity in the tropical Pacific Ocean. Results from the two periods show two sources of uncertainty in the reanalyses. One source is the inherent atmospheric noise that partially causes the representation of the same ENSO event to vary widely in strength, duration, and location among the 56 ensemble members. For example, warming during the 1918/1919 event in some members is far in the eastern equatorial Pacific Ocean while in other members the major warming is in the central Pacific. The other source of uncertainty comes from prescribing SST to the atmosphere, and is primarily responsible for differences seen among ensemble members. During the well-observed 1996-2000 period, the ensemble variance is considerably smaller than that of the 1916-1920 period, thus a markedly reduced level of uncertainty. Similarities among the results of each atmospheric reanalysis set generated with the same SODAsi.1 SST suggest that the state estimates are strongly dependent upon the SST boundary condition. The results add to what is previously known about ENSO in order to improve ENSO predictability, as well as highlight the importance of loosely coupling ocean and atmosphere reanalyses to adequately represent the range of possible climate states in periods of few observations.

publication date

  • May 2015