Summary: | Current and past records of hydroperoxides in air, snow and firn of West Antarctica were investigated over 3 field seasons covering >5000 km and 24 sites. Concentrations of methylhydroperoxide (MHP), the only important organic atmospheric hydroperoxide found, were ten times the levels expected based on past photochemical modeling. Between 76 and 90 °S, the snow pack is a net source for hydrogen peroxide (H₂O₂) but not for MHP in summer. There is strong evidence that low stratospheric ozone leads to increases of H₂O₂ at the surface. The modeled sensitivity of H₂O₂ and particularly MHP to nitrogen oxide (NO) shows that atmospheric hydroperoxides help constrain the NO background and thus estimate the past atmospheric oxidation capacity using ice cores. Century-scale H₂O₂ ice core records from 24 locations across the West Antarctic Ice Sheet (WAIS) were used to develop a regional semi-empirical deposition model, with accumulation rate and temperature as parameters. The model fit using long-term average H₂O₂ concentrations is consistent with an existing physically based atmosphere-to-snow transfer model and predicts ‘effective’ annual mean atmospheric H₂O₂ mixing ratios of 1-3 pptv across the region, comparable to observations. Modeled sensitivities suggest that recent Antarctic temperature changes have no noticeable effect on the H₂O₂ record in the interior of WAIS and that accumulation rates dominate the interannual variability of H₂O₂ under the current temperature regime. This leads to the imprint of a large-scale climate signal in the H₂O₂ record, with significant spectral peaks at ENSO-like periodicities (2-7 yr). Validation of results from the new ECMWF reanalysis (ERA-40) using accumulation records from the same cores showed that simulated accumulation rates are too low at 21 of the 22 ice core sites, averaging only 66% (range 34–122%) of the observed accumulation over 1958-2001. Compared to the pre-satellite era (1958-1977) total amounts of accumulation matched better and linear correlations improved after 1978, showing statistical significance at 10 of 22 core sites (0.11 < r < 0.73, p<0.1).
|