1. COSMIC Data Analysis and Archival Center
Status and Challenges
C. Rocken

2. Key Achievements Over 3 Million profiles
All six satellites continue to operate
Switched to NOAA ground stations (and added McMurdo & Wallops)
Daily count holds at ~1800 in spite of slowly degrading satellite performance
Advanced understanding of refractivity biases
First new GPS “L2C” test results - no soundings lost due to “bad L2”
Agreement to obtain global fiducial data from EUMETSAT

3. 3.2 Million Profiles pass QC
1.5 Million Neutral Atmosphere
1.7 Million Ionosphere
952 registered data users & major operational weather centers

4. Data Latency: ~90% < 3hours

5. Mission Overview: Total Count by satellite

6. Statistical comparison of COSMIC-retrieved N to ECMWF
SNR (60-80km) > 800 V/V
SNR (60-80km) < 600 V/V
Positive N-bias (~0.7% between 2-6 km) - due to antenna gain limitation (only low SNR soundings affected)
Negative N-bias (~0.5% in bottom 2 km) - due to antenna gain limitation and possibly atmospheric super refraction

7. RO Biases and Solar Activity
BA Bias
F10.7 solar flux
6-year km bending angle bias from CHAMP
F10.7 Flux vs. Bending Angle bias regression
Changes in solar activity(diurnal and 11-year solar cycle) introduce changes in the RO bias.
These changes are significant for the RO climate record > 20 km
Discussion of magnitude, causes, correction methods in COSMIC Workshop on Wednesday afternoon (last talk)

8. A remaining bias puzzle ….
Compared to ECMWF rising occultations have a larger N bias > 20 km than setting occs.
We do not see this bias compared to NCEP/GFS
This bias difference is not understood yet.

9. Tracking the new L2C GPS signal (height where L2 data stop)
Only 25% of L2 is tracked < 10 km - compared to 90% of L2C.
30% of L2 tracking data cannot be processed compared to ~1% of L2C.

10. Tasks for next year Continue to improve COSMIC products
Work with JPL on firmware: L2C open loop, “SNR splitting”, rising occultation count improvement, …
Modify CDAAC processing to:
Restructure processing for climate products
Higher resolution vs. lower noise processing mode tradeoff & decision (talk on Wed. 1:45)
Increase occultation count by relaxing L2 QC
Improve space weather processing (new scientist will come on board)
Support development of follow-on mission(s)

11. Summary
Goals of data quality, latency, and reliability are met or exceeded for over 3.2 million profiles
Daily occultation count is currently ~1800 profiles
Increasing count proved difficult against backdrop of degrading hardware - but we have one more “card to play”: relaxing our L2 QC
Current research into small RO biases yields interesting insight and helps with planning for future mission