delta(18)oxygen of atmospheric carbon dioxide: Towards the development of an artifact free database from the NOAA/ESRL Carbon Cycle cooperative global air sampling network

The stable oxygen isotopic ratio of atmospheric carbon dioxide (δ 18O-CO2) can potentially track fundamental indicators of environmental change such as the balance between photosynthesis and respiration on regional to global scales. The Stable Isotope Laboratory (SIL) at the Institute of Arctic and Alpine Research (INSTAAR), University of Colorado at Boulder, has measured the stable isotopes of atmospheric carbon dioxide from more than 60 NOAA/Earth System Research Laboratory (ESRL) Carbon Cycle cooperative global air sampling network sites from 1991 to the present. If air is sampled without drying, oxygen can exchange between carbon dioxide and water vapor in the flasks, in some cases entirely masking the desired signal. The effects of water vapor on δ18O-CO2 flask samples are investigated by comparing the SIL measurements with specific humidity, calculated from the National Climatic Data Center (NCDC) global integrated surface hourly temperature and dew point database. Comparison of NOAA/ESRL sites with two collection methods, ‘wet’ and ‘dry’, are analyzed in conjunction with the humidity data. Samples that initially appear isotopically unaltered, in that their δ18O-CO 2 values are within the expected range, actually display subtle influence by water vapor in the air. At tropical and semi-tropical NOAA/ESRL sites, the ‘wet’ sampling δ18O-CO2 values display a seasonal cycle that is strongly anti-correlated to the specific humidity, while the ‘dry’ δ18O-CO2 values display a seasonal cycle that occurs 1-2 months earlier than the specific humidity seasonal cycle. The latter phasing is expected given the seasonal phasing between climate over the ocean and land, while the former is consistent with a small, but measurable isotope exchange in the flasks. A potential threshold of specific humidity, below which ‘wet’ air flask samples can be considered unaltered, has been determined and applied to the δ 18O-CO2 database. This approach segregates suspect and reliable data, providing an improved record of δ18O-CO 2 over the past two decades. Modeling studies that apply the improved δ 18O-CO2 database to investigations will have increased confidence in the results.