Impact of DYNAMO Observations on the NASA GEOS-5 Reanalysis and the Representation of the MJO Initiation over Tropical Indian Ocean

While most of the DYNAMO observations are made available to the operational weather centers via GTS submissions, high vertical resolution, spatially complete and quality controlled data are not assimilated in the real-time analyses. In this study, we produce two reanalyses distinguished by different horizontal resolutions (1 and ¼ degree) that assimilate the level 4 quality controlled 5mb vertical resolution DYNAMO radiosonde data.   The reanalyses cover the extended DYNAMO period spanning 1 Jun 2011 – 30 Jun 2012 and employ the same version of the data assimilation system used in producing the new GMAO reanalysis, MERRA-2. Assimilated DYNAMO observations are from 87 stations that include enhanced observing sites, GTS resolution soundings from priority and non-priority sounding sites, pibal and dropsonde profiles^*. To aid in quantifying the impact of the DYNAMO observations, companion data-denial (control) runs are produced where all DYNAMO stations are removed. Examination of analysis tendency terms and reanalysis innovation statistics show that the DYNAMO observations considerably improve the representation of vertical profiles of temperature and winds while the impact on humidity profiles is less distinct. Known cold biases in the assimilating (GEOS-5) model at the top of the boundary layer, 700hPa and above 200hPa as well as warm biases between 400-200hPa are largely corrected by the DYNAMO observations. Furthermore, the assimilation of DYNAMO observations considerably dries model middle troposphere and moistens model upper troposphere, so that the reanalysis deep convective moist regimes occur primarily in the upper troposphere, in better agreement with the observations. Diabatic heating profiles derived as residual in the thermodynamic equation show that the bottom heavy profile during MJO preconditioning stage and top heavy profile during MJO deep convective stage are more realistic when DYNAMO observations are assimilated. Additionally, we find that finer spatial resolution largely improves the representation of convective sensitivity of environmental moisture possibly due to the stronger default minimum threshold of entrainment used in higher resolution GEOS-5 AGCM.

^*http://data.eol.ucar.edu/codiac/ds_proj?DYNAMO