UC San Diego

I present airborne in situ and flask measurements of the atmospheric oxygen-nitrogen ratio--[delta](O₂/N₂)--and CO₂ over the Southern Ocean made at five points in the seasonal cycle between 2009 and 2011. The flights, which were part of the HIPPO global project, were conducted with repeated profiling along north-south tracks from near the ocean's surface to the lower stratosphere. I use these measurements to define seasonal variations in meridionally and vertically averaged atmospheric potential oxygen (APO=O₂+1.1*CO₂) over the Southern Ocean--the "meridional curtain average"--and relate these to the underlying oceanic O₂ and CO₂ fluxes. I find that a 1- harmonic fit to the integrated seasonal cycle of APO over the latitudes 65-45°S is better constrained than a 2-harmonic fit, and has an amplitude of 36.4 ±5.6 per meg with a seasonal peak/trough centered at year day 58/260 ±10.8 days. Comparison with nearby station APO records at Cape Grim Observatory and Palmer Station suggest this constitutes roughly 60% of the surface seasonal cycle. I use TM3 and ACTM atmospheric transport model output to compare observations and surface fluxes, and to estimate uncertainties in my approach. The lower boundary condition for these runs is set by flux estimates derived variously from O₂ and CO₂ measurements ("dissolved climatologies") and physical ocean models with biogeochemistry modules. ACTM and TM3 dissolved climatology runs, and TM3 runs with CCSM3, CESM, MOM4, NEMO-PISCES-T, NEMO-CNTRL, and NEMO- WSTIR ocean model simulations overestimate (+) or underestimate (-) the observed amplitude by +49%, +44%, + 17%, +45%, +13%, +1%, +4%, -7% respectively, with 1- harmonic phasing that leads (+) or lags (-) observations by +13, +12, -32, -1, -10, -51, -13, and -28 days, respectively. No single model falls within HIPPO error bars by all metrics: amplitude, peak and trough timing, and 2-harmonic trough-peak asymmetry, but MOM4 and NEMO- CNTRL are the most consistently good performers. Analysis of heat fluxes suggests both physical and biological forces are behind differences between observed and modeled seasonal phasing and amplitude. I find that Garcia and Keeling [2001] O₂ flux fields anticipate observations by about two weeks, and need to be re-evaluated based on new gas exchange scaling parameters. I also present analyses of gradients and seasonal cycles of atmospheric Ar/N2. I find a measurable vertical Ar/N₂ gradient in the lower stratosphere that corresponds with N₂O data and height. I also find resolvable, phase-opposite seasonal cycles at surface elevations in both hemispheres, and an interhemispheric gradient between the latitudes of Cold Bay Alaska (55°N) and Cape Grim Observatory (41°S) of - 5.6 ±4.6 per meg (CBA-CGO) which matches mean annual differences between the two station records well