Supplementary Files for "Secular Changes in the Occurrence of Subduction During the Archean"

This supplementary file contains data and python code for in the manuscript "Secular Changes in the Occurrence of Subduction During the Archean" by Chun-Tao Liu, Chen-Yang Ye, J ZhangZhou.

Contents:

1.Tables:

Table-S1_Parental training dataset: The parental training dataset was compiled from the GEOROC and the PetDB databases comprising arc and non-arc basalts (see Materials and Methods in the main text).

Table-S2_Dataset for training model: We extracted 27 elements and elemental ratios from the "Parental training dataset", and then it was applied to train the XGBoost machine learning model.

Table-S3_Application dataset: The application dataset was collected from the EarthChem database. Our team of ten experienced geochemists has collaborated for two years to meticulously filter a dataset of crust compositions spanning the Archean. This effort has resulted in a substantial increase in both the quantity and quality of the dataset (see Materials and Methods in the main text).

Table-S4_Application dataset for prediction: We extracted featrues from the "Application dataset", which can be predicted by the well-trained XGBoost model.

Table-S5_Validation dataset: We compiled validation dataset to test our model further (see details in the main text and Figure S6).

Table-S6_Melting P-T for modern and predicted Archean arc basalts: We calculate the melting P-T conditions following the method of Lee et al. (2009) (see details in the main text, Text S5, and Figure S7). 

2.Code

Python_arc_evolution_ctliu2023 for using Table-S2, Table-S4 and Table-S5.