Water vapor is fundamental to the thermodynamic and dynamical coupling of the tropical ocean and atmosphere. Therefore, water vapor observations are critical to understanding and predicting modes of variability such as the El-Nino-Southern Oscillation (ENSO) that dominates atmosphere-ocean climate system variations at low latitudes. We present initial results of a study on ENSO focusing on free tropospheric water vapor derived from Global Positioning System Radio Occultation (GPSRO) observations. We are using vertical profiles of water vapor derived from two sources: the CHAllening Minisatellite Payload (CHAMP) and the Constellation Observing System for Meteorology, Ionosphere and Climate (COSMIC).

The CHAMP dataset, which extends back to 2002, allowed us to discover the signature of El Nino episodes, where the wettest portions of the free troposphere over the central Pacific at the peak of the El Nino are as wet as the wettest free troposphere profiles over India and SE Asia during the Indian-Asian monsoon which are generally the wettest profiles on Earth. Another intriguing pattern of behavior, suggestive of a predictive skill of the La Nina phase, is in the dryness of the free troposphere. The free troposphere becomes unusually dry in the preceding January followed by very wet profiles in the Southern Pacific convergence zone (SPCZ) in March/April followed by a La Nina in November-January; this sequence has happened in each of the two La Ninas captured by CHAMP (2005 and 2007).

The COSMIC dataset, which spans mid-2006 to the present, with order of magnitude higher sampling density provides far denser coverage than previous GPSRO datasets, enabling a more in-depth study. While far too short for a climatology, the tropical system has experienced both warm (fall 2006- spring 2007) and cold (fall 2007-spring 2008) ENSO phases over this short interval. The contrast between the two phases has allowed us to extract spatio-temporal patterns associated with ENSO. To expose links between water vapor and other key ENSO variables such as SST, clouds and OLR and search in particular for those that may improve predictive skill, we have created a new vertically-resolved, gridded water vapor dataset via a cluster analysis. We will summarize our approach and some of the key spatio-temporal patterns that we have found in this study.