Tropical tropospheric ozone trends (1998 to 2023): New perspectives from SHADOZ, IAGOS and OMI/MLS observations

Abstract

Tropospheric ozone trends are important indicators of climate forcing and surface pollution, yet relevant satellite observations are too uncertain for assessments. The assessment project TOAR-II has used multi-instrument, ground-based data for global trends over 2000-2022 (Van Malderen et al., 2025a, b). For the tropics, trends are derived from SHADOZ ozonesonde profiles (Thompson et al., 2021, "T21"; Stauffer et al., 2024) or combinations of satellite, SHADOZ and IAGOS aircraft measurements (Gaudel et al., 2024). We extend T21 that covered 1998-2019, analyzing SHADOZ data at five sites with a Multiple Linear Regression (MLR) model for 1998-2023 and reporting trends for two free-tropospheric (FT) segments, the lowermost stratosphere and the total tropospheric column (TrCOsonde). Trends for the Aura period, 2005-2023, are computed from OMI/MLS TrCOsatellite. We find the following: Extending SHADOZ analyses 4 years shows little change from T21; TrCOsonde trends are small (0.5-1 DU/decade) except over SE Asia. Annual trends for TrCOsonde and OMI/MLS TrCOsatellite agree within uncertainties at four of five sites, with the largest differences at Samoa. Sensitivity tests show the following: (a) Adding thousands of FT IAGOS profiles to SHADOZ yields little change in trends; SHADOZ sampling is sufficient. (b) Quantile Regression (QR) and MLR median trends are both near zero, but QR captures extremes (5th percentile, 95th percentile) with changes up to ±1 DU/decade (p< 0.10). (c) Twelve-year analyses for trends lead to uncertainty changes too large for an assessment. This study and Van Malderen et al. (2025a, b) provide the most reliable TOAR-II trends to date: over the past ∼ 25 years, tropical FT ozone changes have been modest, ∼ (-3-+3) %/decade, except over SE Asia.