CHEMOSTRATIGRAPHIC CORRELATION FROM A REPRODUCIBLE DYNAMIC PROGRAMMING ALGORITHM (Invited Presentation)

Dynamic programming algorithms objectively align time-series of Plio–Pleistocene paleoclimate data subject to uncertainties in temporal overlap and relative accumulation (e.g., Lisiecki and Lisiecki, 2002; Channell et al., 2009); applications of this technique to pre-Cenozoic correlation problems are more limited (Sadler, 2004). We describe a dynamic programming method for obtaining least-squares optimal alignments between chemostratigraphic records (here δ¹³C_carbonate ) for the purpose of extrapolating an age model developed at a radiometrically calibrated stratigraphic section to one lacking an absolute chronology (see Hay et al., Geology, 2019, for the open-source algorithm). For case studies undertaken to date, including Cambrian Series 1–2 δ¹³C_carb records (Maloof et al., 2010), dynamic programming produces a library of plausible δ¹³C_carb alignments arising from various assumptions about the insertion of depositional hiatuses and the total temporal overlap of the two records (Hay et al, 2019). Since each dynamic programming-determined δ¹³C_carb alignment produces a sub-variant on the reference age model, this method commonly yields a range of plausible ages for a reference event (datum), such as a fossil first/last appearance or the age of a stratigraphic bed/surface, or rates of geologic processes. As such, geological ages (or rates) derived from dynamic programming algorithms should be reported as a confidence interval.