The impacts of the introduction of satellite data on the EKE trends in the two reanalyses are discussed. For the NCEP/NCAR reanalysis, comparing the EKE computed from analyses with and without satellite data assimilated, it is found that the assimilation of satellite data decreases the SH EKE amplitude by about 7%. Therefore, after correcting the pre-satellite period for this bias, the EKE trend in NCEP/NCAR reanalysis becomes more significant. However, results from a study by Bengtssen et al. suggested that the assimilation of satellite data increases the SH EKE by up to 15% in ERA-40, which makes the trend in ERA-40 much smaller or even non-existent after bias-correction. These opposite impacts of the assimilation of satellite data can also explain the larger amplitude difference while better correlation between the SH EKE in the two reanalyses after late 1970s compared to before late 1970s.
The energetics of EKE has also been examined. Although the EKE shows upward trends in the two reanalyses, both the baroclinic and barotropic conversion terms show downward trends, which suggests the inconsistency of the EKE energetics in terms of the secular trend. However, after taking into account the impacts of the assimilation of satellite data on the conversion terms, the downward trends in the conversion terms in the NCEP/NCAR reanalysis are largely removed, somewhat mitigating the inconsistency in the energetics.
Due to the large impacts of the assimilation of satellite data, it is at present difficult to ascertain whether the upward trends in SH EKE is real or is instead a reanalysis artifact due to changes in the observing system. Currently, rawinsonde observations at SH stations close to the storm track regions are being examined to explore whether these observations support an upward trend in SH EKE. These results will be compared with those from the reanalyses and presented at the conference.