Extracting geologic data from text: a big data approach to the dolomite problem

Abstract:

Dolomite (CaMg(CO₃)₂) is a ubiquitous carbonate mineral throughout the rock record yet does not precipitate readily in modern oceans; two end-member hypotheses have been invoked to explain this discrepancy. On one end, ancient ocean chemistry may have been sufficiently different from the modern to promote the precipitation of primary (or early diagenetic) dolomite in deep time. On the other hand, ancient dolomites may have formed secondarily, through the diagenetic alteration of limestone (CaCO₃), implying ancient seawater chemistry similar to the modern. Primary/earlyand secondary/late dolomites can be differentiated on the basis of petrographic fabric. Individually analyzed Precambrian (older than 541 Myr) dolomites have fabrics associated with early formation more often than younger rocks, but this observation has not been robustly shown for dolomites across space and throughout geologic time. Understanding whether and when dolomite formed by precipitation from ancient seawater has great bearing on our understanding of the evolution of ocean chemistry and carbonate-producing environments throughout Earth history.

By coupling the high temporal resolution of Macrostrat with the machine reading capabilities of geoDeepDive, we construct a high-resolution record of the prevalence of dolomitic fabrics throughout the past 2 billion years of Earth history. Our work uses text mining and natural language processing to identify geologic units in space and time, link units to their lithology, and extract petrographic descriptions of carbonate rocks. Machine reading allows us to differentiate between early and late dolomite on the basis of described carbonate fabrics throughout the geologic record and across the globe. We find that the relative abundance of dolomite is constant across the Precambrian-Cambrian boundary (541 Ma) despite evidence for a change to more fabric-destructive dolomite in the Phanerozoic, suggesting that seawater chemistry has changed throughout geologic time.