The stability and calibration of water vapor isotope ratio measurements during long-term deployments

With the recent advent of commercial laser absorption spectrometers, field studies measuring stable isotope ratios of hydrogen and oxygen in water vapor have proliferated. These pioneering analyses have provided invaluable feedback about best strategies for optimizing instrumental accuracy, yet questions still remain about instrument performance and calibration approaches for multi-year field deployments. With clear scientific potential for using these instruments to carry out monitoring of the hydrological cycle, this study examines the long-term stability of the isotopic biases associated with three cavity-enhanced laser absorption spectrometers - calibrated with different systems and approaches - at two remote field sites: Mauna Loa Observatory, Hawaii, USA, and Greenland Environmental Observatory, Summit, Greenland. The analysis pays particular attention to the stability of measurement dependencies on water vapor concentration and also evaluates whether these so-called concentration dependences are sensitive to statistical curve-fitting choices or measurement hysteresis. The results suggest evidence of monthly-to-seasonal concentration-dependence variability - which likely stems from low signal-to-noise at the humidity-range extremes - but no long-term directional drift. At Mauna Loa, where the isotopic analyzer is calibrated by injection of liquid water standards into a vaporizer, the largest source of inaccuracy in characterizing the concentration dependence stems from an insufficient density of calibration points at low water vapor volume mixing ratios. In comparison, at Summit, the largest source of inaccuracy is measurement hysteresis associated with interactions between the reference vapor, generated by a custom dew point generator, and the sample tubing. Nevertheless, prediction errors associated with correcting the concentration dependence are small compared to total measurement uncertainty. At both sites, changes in measurement repeatability that are not predicted by long-term linear drift estimates are a larger source of error, highlighting the importance of measuring isotopic standards with minimal or well characterized drift at regular intervals. Challenges in monitoring isotopic drift are discussed in light of the different calibration systems evaluated.