Optimal sampling resolution for water isotope records in ice cores

Water isotopes in ice cores offer a window into the climate of the past. Often the measurement of these water isotopes is done discretely, with the ice cores cut into many regularly spaced samples. Determining the optimal sampling rate for these measurements is a question of balancing high temporal resolution with processing time and effort. Furthermore, the effect of isotopic diffusion smooths the record, attenuating high frequency variability far below the measurement noise level. This results in some climate information becoming unobtainable and limits the usefulness of very high resolution data. Deconvolving (un-diffusing) the time-series can further enhance the signal but strongly depends on the precision of the isotope data that is mainly determined by the measurement error.

Here, we discuss the optimal measurement specifications in terms of sampling resolution and precision for different uses of the water isotope data, such as the estimation of the diffusion length, the direct interpretation of the time-series and the interpretation of the time-series after it has been deconvolved. We do this theoretically, based on how we model diffusion, and empirically through simulations of surrogate ice core records. Our findings can provide guidance on how to process new deep ice cores such as the Oldest Ice Core record.