Geophysics and Tectonics Seminar fall-2021

The mid-ocean ridges represent a very visible consequence of mantle convection and plate tectonics, and exhibit distinct large-scale geophysical and geochemical patterns. Previous studies try to understand the differences by examining the correlation among basin-distinct spreading rate, ridge depth, and geochemistry of MORB. (Klein and Langmuir 1987, Gale et al. 2014, Yaoling Niu 2016, Brandl et al. 2013), but they usually focus on shallow processes associated with partial melting and melt transport to explain the observations. However, MORB observations, especially geochemistry and associated estimates of mantle temperatures are strongly affected by shallow melting processes, this makes it difficult to understand any potential contributions of the deep mantle flow (Krein et al. 2021). In this study, we examine the role of the deep mantle and large-scale tectonics on the mid-ocean ridge system, using just the seismically-inferred (Bao et al. 2021) deep (260-600 km) upper mantle temperatures. Using a robust machine learning model (Random Forest), we show that it is possible to predict the ocean basin of each ridge segment (data: Gale et al. 2014) up to 90% accuracy, with the mantle temperature at depth alone, in the absence of any other information. We find that the two features that provide more than half of the discriminative power in the random forest model are the temperature difference between the mid-layer (320-500 km) and other depths, as well as average temperature over all depths. Our result implies that the temperature of the upper mantle may record the 100s Ma (mantle overturn timescale) convection and tectonic history. These processes have long-term impact on the upper mantle flow, and the (thermal) heterogeneity in each basin. We posit that the mantle convection and tectonic processes play a significant role in explaining the large-scale geophysical and geochemical differences in the mid-ocean ridge system.