Impacts of Assimilation of Satellite and Rawinsonde Observations on Southern Hemisphere Baroclinic Wave Activity in the NCEP–NCAR Reanalysis

Abstract

In this study, the impacts of the assimilation of satellite and rawinsonde observations on Southern Hemisphere (SH) baroclinic wave activity in the NCEP?NCAR reanalysis are examined by comparing analyses made with and without the assimilation of satellite data (SAT and NOSAT, respectively) for the year 1979, as well as by comparing analyses to the corresponding first guesses from 1958 to 1999. Comparing the eddy kinetic energy (EKE) computed based on the SAT and NOSAT analyses, it is found that the assimilation of satellite data generally decreases the EKE in the SH, which is the opposite of the findings for the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) in previous studies. The decrease of EKE by satellite data in the NCEP?NCAR reanalysis can be traced back to a low bias in retrieved satellite temperature (SATEMP) variance. The eddy available potential energy (EPE) is decreased even more than the EKE with the assimilation of SATEMP, making the waves more barotropic in the SAT analysis. The EKE analysis increment, that is, the difference between the EKE based on analysis and first guess, is a good quantity to indicate the impacts of all observations assimilated. In the NOSAT analysis, positive EKE analysis increments are found around the SH rawinsonde stations, indicating that the assimilation of rawinsonde data increases EKE significantly from the first guess. This also suggests that the NCEP?NCAR first guess is probably biased low. Positive analysis increments around the rawinsonde stations become even larger in the SAT analysis compared with the NOSAT, suggesting that with the assimilation of low-biased SATEMP data, the EKE in the analysis (the initial condition for next time) and hence the first guess is reduced, therefore the rawinsonde observations have to further increase the EKE from the first guess. The patterns of EKE increment from the presatellite (1958?77) and satellite (1979?99) eras show high degrees of similarities to the NOSAT and SAT reanalysis patterns, respectively, lending further support to these findings. The impact of the assimilation of satellite data on the trend of SH baroclinic wave activity is discussed. Positive trends in the SH EKE are found in both the NCEP?NCAR and ERA-40 reanalyses during 1958?99. After taking the impacts of satellite data into account, the EKE trend in the NCEP?NCAR reanalysis gets stronger, while that in the ERA-40 is largely weakened, which adds complications to assessing the real trend in SH baroclinic wave activity. Comparisons among the variances based on the two reanalyses, NCEP?NCAR first guess, SATEMP, and rawinsonde observations are presented to substantiate some of the findings discussed above, such as the low bias in energy in NCEP?NCAR first guess and SATEMP variance.