Correction of high-frequency (> 0.3 mHz) air pressure effects in gravity time-series

Correcting atmospheric effects in measurements of vertical acceleration is challenging, particularly at frequencies above 0.3 mHz. Corrections based on 4-D atmosphere models should be the most effective, but are limited in temporal and spatial resolutions. So these models are commonly combined with a constant admittance between air pressure and gravity, though the correction can be further improved by allowing a frequency-dependent admittance. This paper studies the effectiveness of such procedures given that air pressure admittance varies with time because of variability in the underlying local atmospheric process, and finds that admittances estimated by cross-spectral analysis gives better results than a constant admittance does. The best results are achieved by combining a 4-D model with admittances estimated from time-series that cover the event of interest.