Effect of sampling error on ozone partial pressure trends within a unified ozonesounding dataset

Abstract. This work discusses the impact of the sampling frequency on ozone partial pressure trends, estimating its impact at various latitudes and vertical layers in the upper troposphere/lower stratosphere (UT/LS) region. The trends are estimated in the periods 1978–1999 and 2000–2022, using a new unified dataset combining the ozonesounding profiles provided by SHADOZ (Southern Hemisphere ADditional OZonesondes – https://tropo.gsfc.nasa.gov/shadoz/), NDACC (Network for the Detection of Atmospheric Composition Change – https://ndacc.larc.nasa.gov/) and WOUDC (World Ozone and Ultraviolet Radiation Data Centre – https://woudc.org/). These datasets are combined to offer adequate coverage at various latitudes and to enhance the estimation of anomalies and trends in ozone concentration on a global scale. The available measurements are classified into three groups based on the temporal coverage of historical time series. Some regression approaches are utilised to estimate trends and the related difference to quantify structural uncertainty. Significant trends for the period 1978–1999 are estimated for the Northern Hemisphere mid-latitude (NH), which shows a negative trend of 5 % per decade in the layer 50-1 hPa and a negative trend of 10 % per decade at 100-50 hPa, and for the Tropics (TR), which shows a positive trend of about 5 % per decade at 50-1 hPa and 7 % per decade at 100-50 hPa, respectively. Furthermore, the sampling error between the clusters was investigated, revealing a small effect of less than 2 % at 100-50 hPa and 1.5 % at 50-1 hPa for NH and about 3 % at 100-50 hPa and 3.5 % at 50-1 hPa for TR, as well as the structural uncertainty between the regressors used, 1.05 % at 100-50 hPa and 1.15 % at 50-1 hPa for NH and about 2 % at both 100-50 hPa and 50-1 hPa for TR.