Propagation of radiosonde pressure sensor errors to ozonesonde measurements

Abstract. Several previous studies highlight pressure (or equivalently, pressure altitude) discrepancies between the radiosonde pressure sensor and that derived from a GPS flown with the radiosonde. The offsets vary during the ascent both in absolute and percent pressure differences. To investigate this problem further, a total of 731 radiosonde/ozonesonde launches from the Southern Hemisphere subtropics to northern mid-latitudes are considered, with launches between 2005 and 2013 from both longer term and campaign-based intensive stations. Five series of radiosondes from two manufacturers (International Met Systems: iMet, iMet-P, iMet-S, and Vaisala: RS80-15N and RS92-SGP) are analyzed to determine the magnitude of the pressure offset. Additionally, electrochemical concentration cell (ECC) ozonesondes from three manufacturers (Science Pump Corporation; SPC and ENSCI/Droplet Measurement Technologies; DMT) are analyzed to quantify the effects these offsets have on the calculation of ECC ozone (O₃) mixing ratio profiles (O_3MR) from the ozonesonde-measured partial pressure. Approximately half of all offsets are > ±0.6 hPa in the free troposphere, with nearly a third > ±1.0 hPa at 26 km, where the 1.0 hPa error represents ~ 5% of the total atmospheric pressure. Pressure offsets have negligible effects on O_3MR below 20 km (96% of launches lie within ±5% O_3MR error at 20 km). Ozone mixing ratio errors above 10 hPa (~ 30 km), can approach greater than ±10% (> 25% of launches that reach 30 km exceed this threshold). These errors cause disagreement between the integrated ozonesonde-only column O₃ from the GPS and radiosonde pressure profile by an average of +6.5 DU. Comparisons of total column O₃ between the GPS and radiosonde pressure profiles yield average differences of +1.1 DU when the O₃ is integrated to burst with addition of the McPeters and Labow (2012) above-burst O₃ column climatology. Total column differences are reduced to an average of −0.5 DU when the O₃ profile is integrated to 10 hPa with subsequent addition of the O₃ climatology above 10 hPa. The RS92 radiosondes are superior in performance compared to other radiosondes, with average 26 km errors of −0.12 hPa or +0.61% O_3MR error. iMet-P radiosondes had average 26 km errors of −1.95 hPa or +8.75 % O_3MR error. Based on our analysis, we suggest that ozonesondes always be coupled with a GPS-enabled radiosonde and that pressure-dependent variables, such as O_3MR, be recalculated/reprocessed using the GPS-measured altitude, especially when 26 km pressure offsets exceed ±1.0 hPa/±5%.