The first aim of this research, which began in 1992 as part of the GISP2 program, was to determine the relation between concentrations of hydrogen peroxide (H2O2) in the atmosphere and corresponding levels in snow and shallow firn at Summit, Greenland. The work involved: i) atmospheric and snow measurements at Summit, Greenland, ii) modeling, and iii) laboratory studies. From this we developed a conceptual understanding and model of processes affecting H2O2 in surface snow, firn and ice cores. H2O2 in near surface snow equilibrates with atmospheric levels over a time scale of weeks to months. Once buried it is preserved, however redistribution continues throughout the firn. In shallow firn, gas-phase H2O2 concentrations are in equilibrium with those at the surface of the ice grains, but not the bulk ice grain. Equilibration times in the firn are limited by grain-scale redistribution of H2O2 between grain interiors and surfaces rather than by diffusion in the open pore space. Equilibration was slowest at lower temperatures. This equilibrium is temperature dependent below about -12 C (greatest uptake at lowest temperature studied), and influences the annual pattern of H2O2 in snow. Both seasonal and year-to-year H2O2 levels in snow (and thus ice) do respond to changes in atmospheric H2O2. However, degassing of H2O2 from surface snow also influences atmospheric concentrations.

The second aim of this research, which continued after completion of the GISP2 deep drilling, was to develop a better conceptual understanding of the processes controlling post-depositional fluxes of H2O2, formaldehyde (HCHO) and other species between the atmosphere and surface snow or shallow firn at Summit. The work involved: i) field studies at Summit, ii) development of mathematical models for transfer processes, and iii) model calibration and evaluation with field and laboratory data. Photochemical modeling showed that homogeneous processes alone cannot account for the high daytime H2O2 levels measured at Summit. Daytime highs and nighttime lows differed by about 0.5-1.0 ppbv, indicating heterogeneous loss to the snow at night and possible degassing from the snow in the daytime. HCHO concentrations in the atmosphere were typically about 0.4 ppbv, and failed to exhibit a distinct diel variation. H2O2 concentration in the surface snow layer showed small increases in parallel with increases in atmospheric concentrations. During most days there was evidence of hoar or rime deposition, and a few nights of diamond dust or fog deposition. There was low spatial variability in surface snow H2O2 on the scale of 1 m. A direct measurement indicated that a peak in summer H2O2 concentration in 1993 occurred in August, well after the summer solstice. Measurement of the near-surface snow gradient in hydrogen peroxide at three hourly intervals during a number of intensive surface sampling periods gave evidence of firn degassing; the concentration gradient shifted with solar radiation and temperature. Snow-pit H2O2 profiles adjacent to three automatic snow-depth gages from Summit, Greenland were used to estimate parameters and evaluate the performance of a lumped parameter model to relate concentrations in the atmosphere with those in surface snow and shallow firn (i.e. a transfer function). Three of the model parameters define an equilibrium partitioning coefficient between snow and atmosphere as a non-linear function of depositional temperature. Model parameters yielded a function that closely matched previous laboratory estimates. A fourth parameter reflects the disequilibrium that may be preserved during periods of rapid accumulation. The final model parameter describes the exchange of H2O2 between near-surface snow and the atmosphere, allowing already buried snow to either take up or release H2O2 as conditions in and above the snowpack change. We simulated snow-pit profiles by combining this transfer function model with a finite difference model of gas-phase diffusion in the snowpack. Two applications for this transfer function are: i) to estimate the local seasonal or annual atmospheric H2O2 concentration in the past from snow-pit and ice core-records, and ii) to invert snow-pit and ice-core H2O2 profiles to obtain estimates of the seasonal or annual accumulation time series. We then applied an extension of this model to invert year-round surface snow concentrations at South Pole to give atmospheric concentrations. Inverted concentrations matched both photochemical model output and spot measurments.