Performance of Vaisala RS80 radiosonde on measuring upper-tropospheric humidity after corrections

Upper level water vapor feedback is a major source of uncertainty in global warming simulations. The primary observing system for monitoring tropospheric water vapor--the radiosonde system has reduced accuracy in the upper troposphere. NCAR/ATD has been collaborating with Vaisala to correct errors in humidity data from Vaisala RS80 radiosondes (both RS80-A and RS80-H) based on a series of laboratory tests. Errors include "chemical contamination error", "basic calibration model error", and "temperature-dependence error". Both contamination and temperature-dependence errors produce dry biases in humidity data. A new type of protective shield on the sensor boom for RS80 radiosondes was introduced by Vaisala in May 2000, and is expected to remove the contamination error completely. The temperature-dependence correction dominates total correction at temperatures below -30°C. The correction factor for relative humidity (RH) is about 1.15, 1.75 and 3.3 at -40°C, -60°C and -80°C, respectively. After correcting the radiosonde data, the data is compared with other independent measurements. The corrected radiosonde data has better agreement with the NOAA CMDL frost-point hygrometer data, and shows RH values near or above ice saturation inside cirrus clouds. The correction also reduces discrepancies between lidar and radiosonde upper-tropospheric humidity (UTH) profiles during the FIRE Cirrus-II experiments. The comparison of global UTH with TOVS satellite climatology confirms a systematic dry bias in the uncorrected Vaisala humidity data of about 10%. This dry bias introduces an artificial geographic variability to the global radiosonde data and spurious temporal variability to the time series of UTH at one station. Our correction method should reduce or eliminate the dry bias and artificial variability. Efforts in correcting the Vaisala humidity data, especially in the upper troposphere, would significantly enhance the usefulness of radiosonde data in climate study and in validating and calibrating satellite retrieval techniques for UTH. Our results show that corrected Vaisala RS80 humidity data is accurate enough in the upper troposphere to be used for many of the climate studies.