Machine Learning-Based Quantification of Archean Crustal Thickness (Moho Depth)

The Earth's crust, an incredibly thin layer of rock encompassing the planet's outermost solid shell, plays a crucial role in sustaining human existence by housing vital natural resources. Moreover, actively engaged in a multitude of geological processes such as plate tectonics, volcanic activities, and erosion, it significantly molds landscapes and influences natural phenomena, thereby serving as a cornerstone in maintaining Earth's habitable conditions. However, substantial debate persists regarding the mechanisms governing the growth of continental crust, and the broader implications of its role within Earth's tectonic framework remain elusive. In this study, we sought to predict the historical evolution of crustal thickness during the Archean era using machine learning based on global geochemical data of igneous rocks. The model predicted that crustal thickness decreased progressively from approximately ~35km during the Paleoarchean period to a nadir of roughly ~33km in the Early Mesoarchean era. Subsequently, there was a trend of thickening observed from the Early Mesoarchean to the Neoarchean period. Our findings reveal a dynamic history of Archean crustal thickness characterized by an initial phase of thinning followed by subsequent thickening from the Paleoarchean to the Neoarchean. These fluctuations align cohesively with the transition from a pre-plate tectonic to a regime of sustained plate tectonics. The observed phenomena were attributed to the initiation of ultrathick primary crust formation, accompanied by a tectonic regime primarily influenced by Rayleigh–Taylor instabilities. The instabilities arising from an excessive buildup of primary crust likely played a pivotal role in causing the subsequent thinning of the continental crust, marking the transition phase towards a predominantly modern-style subduction.