Assimilation of GPS radio occultation data for weather prediction

Atmospheric limb sounding technique making use of the radio signals transmitted by Global Position System (GPS) has emerged as a promising approach for global meteorological observations. As demonstrated by the proof-of-concept GPS Meteorology (GPS/MET) experiment, the GPS radio occultation (RO) sounding data are of high accuracy and high vertical resolution. Results from recent GPS RO missions of CHAMP and SAC-C confirm this assessment. In early 2006, the joint U.S.-Taiwan COSMIC mission will be launched and is expected to collect approximately 2,500 RO soundings per day. These data will be available in near real-time for global weather analysis and prediction. However, the GPS RO soundings are non-traditional atmospheric measurements. In order to effectively assimilate the data into atmospheric models, one needs to properly model the observables (through “observation operator”) and carefully consider the observational error characteristics. Moreover, different approaches of assimilation (e.g., 3D-Var, 4D-Var, and ensemble Kalman Filter) have their advantages and disadvantages. In this paper, we will review recent work on the assimilation of GPS RO soundings for the neutral atmosphere. In particular, we will discuss key factors affecting an effective assimilation of GPS RO data for the prediction of important weather systems, such as hurricane and intense Antarctic cyclones. These discussions will also be relevant to the assimilation of ionospheric data assimilation.