Statistical tuning in cyclostratigraphy and astrochronology: foundations, validation, and methodological advances

Abstract. Statistical tuning has moved cyclostratigraphy from visual matching and cycle-ratio inference to quantitative tests of astronomical forcing and estimates of sedimentation rate. Yet methods answer different questions and yield rates, uncertainties, and significance measures that are not directly comparable. To clarify these differences, this review organizes the field around four goals: detecting spectral patterns unlikely to arise by chance; estimating astronomical frequencies and sedimentation rate; constructing age–depth models; and using geological cycles to investigate orbital behavior. Accordingly, Average Spectral Misfit, Bayesian inverse tuning, the TimeOpt family, modulation tests, spectral moments, correlation-coefficient (COCO) methods, AstroGeoFit, and alignment approaches are compared by assumptions, outputs, and limitations. The comparison examines false and missed detections, data reuse for modeling and testing, background models, and variable sedimentation rates. Building on this framework, COCO 2.0 adds age-dependent orbital frequencies, spectral power leakage correction, whole-record and split-record designs, and Monte Carlo tests across candidate rates. Retrospective tests on two noise controls, two known-truth synthetic records, and four geological records show that considering all rates prevents chance matches in noise from becoming detections. Synthetic records show that finding the correct rate and passing a significance test are separate outcomes, while moving-window analysis recovers rate changes missed by a whole-record estimate. Finally, geological cases emphasize the need for independent age control and stratigraphic evidence. Statistical tuning is most reliable when the astronomical target, data treatment, rate search, background model, and geological interpretation address the same question. Future progress requires geologically realistic background models, uncertainty estimates for changing rates, independent geochronology, and reproducible benchmarks.