Evidence for Missing Geomagnetic Reversals From Geomagnetic Reversal Frequency Model Using Adaptive Kernel Density Estimation

Abstract The existence of missing geomagnetic reversals has been proposed, with potential for new magnetostratigraphic age controls. We estimate geomagnetic reversal frequency from 0 to 155 Ma using adaptive‐bandwidth kernel density estimation (AKDE) to evaluate data sparseness and to assess how reversal frequency changes when recently identified geomagnetic reversals are incorporated into the geomagnetic polarity time scale (GPTS) data set. AKDE is a two‐stage procedure that uses an initial density estimator based on an initial (pilot) bandwidth. We found that the pilot bandwidth determined using cross‐validation is stable with respect to data set length. The AKDE results obtained based on the cross‐validated pilot bandwidth reveal four troughs after the Cretaceous Normal Superchron, spaced 13.5–15.0 Myr apart and corresponding to relatively long chrons (>0.8 Myr). One trough near 32 Ma becomes less distinct after the four recently identified reversals are added to the data set. This sensitivity suggests that troughs in the frequency curve may indicate missing geomagnetic reversals.