Comparison of COSMIC-2 radio occultation retrievals with RS41 and RS92 radiosonde humidity and temperature measurements

  • Evaluate the quality of the RS41/RS92 data and two COSMIC-2 wet profile retrievals
  • Comprehensive evaluations of the bias and uncertainty of RAOB data versus RO data
  • Investigate height and day-night dependence of temperature and humidity biases

Understanding the bias and uncertainty between radio occultation (RO) retrievals and radiosonde observations (RAOBs) impacts climate studies and numerical weather predictions. In this study, the temperature and humidity data independently retrieved by University Corporation for Atmospheric Research (UCAR) and NOAA Center for Satellite Applications and Research (STAR) from COSMIC-2 RO data are compared with in-situ Vaisala RS41 and RS92 RAOB data. Collocated 7-month of COSMIC-2 RO and RS41/RS92 RAOB data are analyzed to investigate the height and day-night dependence of temperature and humidity biases. It is found that UCAR and NOAA/STAR COSMIC-2 temperature retrievals are consistent above 12.5 km. There are warm biases in RS92 data compared to RS41 data over the height region above 17.8 km, mainly due to the warm daytime bias in RS92 data. The main temperature difference between UCAR and NOAA/STAR retrievals is ~0.1-0.2 K over 8-11 km, due to differences in the variational retrieval algorithms. Over 8-11 km, the relative temperature difference between COSMIC-2 retrievals and RS41/RS92 RAOBs are more significant than other heights. The UCAR and NOAA/STAR COSMIC-2 humidity retrievals generally consist in the troposphere, especially above 4.8 km. There are systematic wet biases below 4.2 km in the RO retrievals relative to RAOB humidity data. The COSMIC-2 retrieval and RS92 RAOB comparison show a clear day-night humidity bias difference below 4.2 km due to slight dry biases in the daytime RS92 data. The RO versus RAOB comparison helps quantify the temperature and humidity biases among different radiosonde sensor types and different RO retrieval algorithms.

Read 105 times