Investigating deconvolution techniques for deep ice core water isotope profile reconstruction

Deconvolution has been frequently applied to measured water isotope records, enabling reconstructions of the timeseries before diffusion. The most commonly used method, Wiener deconvolution, aims to optimise the frequencies that are restored while minimising the amplification of the measurement noise, which dominates the high frequency variability in deep ice cores. We investigate the effectiveness of Wiener deconvolution, along with modified approaches, at recovering the original climate signal, with a focus on deep, old ice. We use both real data and surrogate timeseries and apply our methods to the bottom of the Dome C d¹⁸O record, with the aim of retrieving millennial timescale variability.