Continuous Gap-filled Atmospheric N2O Record for the Past 800,000 Years Using Machine Learning Techniques: Dataset

This dataset includes a recent compilation of (1) the published Antarctic ice core greenhouse gas (GHG) records over the last 800,000 years and (2) a simulated continuous gap-filled nitrous oxide (N₂O) record over the same period. Existing ice core records contain numerous gaps for the atmospheric N₂O variations, particularly during glacial periods, while CO₂ and CH₄ data have relatively fewer gaps compared to N₂O. Machine Learning (ML) models were applied to simulate the continuous time series of N₂O based on existing CO₂, CH₄, and N₂O data after carefully aligning their time scales. Several ML algorithms were compared and multiple simulations were iterated to decide the best-performing ML model and outcome for the final simulation of continuous time series. Examining the simulated gap-filled N₂O data, we identified millennial-scale N₂O variations associated with known climate orbital forcing in different periodicities, which could be a subject for future studies. The simulated records here, combined with existing measurements of N₂O concentrations, present a benchmark for future studies to improve our understanding of atmospheric N₂O control mechanisms.<b>Version history:</b><b>DataSet_20241016.xlsx</b>: Original dataset submitted with the manuscript before peer review.<b>EDC_timescale.xlsx</b>: Revised dataset reflecting changes made during peer review; used for all figures in the published article.<b>AICC2012_timescale.xlsx</b>: Supplementary dataset applying the same gap-filling methods to GHG records, based on the updated AICC2012 timescale.<b>Apply_offset.xlsx</b>: Contains the calculation of the CO₂ offset value used in the analysis.<br>