Our new publication led by Emma Kahle (University of Washington) deals with the technical challenges the new continuous melting techniques for water isotope analysis have introduced to the study of the firn isotope diffusion. Improvements in precision and resolution have resulted in sizeable effects that hinder our ability to accurately infer diffusion lengths from high resolution water isotope records. In this paper we consider various new approaches for describing the power spectral densities of high resolution data sets and discuss their implications for temperature reconstructions based on firn isotope diffusion.
Kahle, E. C., Holme, C., Jones, T. R., Gkinis, V., & Steig, E. J. (2018). A generalized approach to estimating diffusion length of stable water isotopes from ice‐core data. Journal of Geophysical Research: Earth Surface, 123. https://doi.org/10.1029/2018JF004764
Diffusion of water vapor in the porous firn layer of ice sheets damps high‐frequency variations in water‐isotope profiles. Through spectral analysis, the amount of diffusion can be quantified as the“diffusion length,” the mean cumulative diffusive displacement of water molecules relative to their original location at time of deposition. In this study, we use two types of ice‐core data, obtained from either continuous‐flow analysis or discrete sampling, to separate diffusional effects occurring in the ice sheet from those arising through analytical processes in the laboratory. In both Greenlandic and Antarctic ice cores, some characteristics of the power spectral density of a data set depend on the water‐isotope measurement process. Due to these spectral characteristics, currently established approaches for diffusion estimation do not work equally well for newer, continuously measured data sets with lower instrument noise levels. We show how smoothing within the continuous‐flow analysis system can explain these spectral differences. We propose two new diffusion‐estimation techniques, which can be applied to either continuously or discretely measured data sets. We evaluate these techniques and demonstrate their viability for future use. The results of this study have the potential to improve climate interpretation of ice‐core records as well as models of firn densification and diffusion.