The practice of collecting surface meteorological (surface met) data has been in use by ATD since the launching of balloon borne radiosondes began. This surface data is collected by several independent sensors that measure temperature, humidity, pressure, wind velocity, and wind direction. The temperature and humidity sensors are aspirated, using an electric aspirator, and protected with a radiation shield. The accuracy of the measurements is approximately +/- 4.0 degrees Celsius for the temperature gauge, and +/- 2% for the humidity sensor. The surface met data is integrated into the radiosonde data and can be used as a reference in determining the accuracy of pre-launch surface radiosonde data.
Data quality control, after the completion of a project, is a three step process which includes: (1) running the data through Atmospheric Sounding Processing Environment (ASPEN) software, (2) visually evaluating the data, and (3) comparing the pre-launch PTU radiosonde data with the surface data to check for problems that may have gone undetected by the first two quality control procedures. We began to pay more attention to the third procedure earlier this year. This has helped us to identify several problems associated with pre-launch radiosonde data, and/or surface met data. In several experiments, when comparing the data, we found the pre-launch radiosonde data was warmer and drier as a result of sensor arm heating by solar radiation. In another experiment, it was found that the surface met data was warmer due to poor ventilation of the temperature and humidity sensors. The comparisons also revealed the humidity sensor contamination and resulting dry bias in the Vaisala RS 80 humidity sensor, and the moist bias in pre-launch data that occurs at night in warm and humid tropical regions due to condensation on the radiosonde sensor arm. Our response to such errors has been to either correct the data sets, or simply to warn users about the presence of errors and/or bias found. The comparison of pre-launch radiosonde and surface met data has shown us the importance of recording pre-launch data, and has prompted us to improve system operations and performance in the field.