South American Summer Monsoon (SASM) Index Reconstruction (1850–2000) Using Bayesian Optimal Information Extraction

This dataset provides an ensemble reconstruction of the South American Summer Monsoon (SASM) index from 1850 to 2000. The SASM index is defined as the area-weighted mean of December–February (DJF) precipitation over the core monsoon domain (5°N–15°S, 50°W–80°W), a region characterized by peak austral summer rainfall. To extend the index prior to the instrumental era, we applied a Bayesian Optimal Information Extraction (OIE) approach that integrates multiple annually resolved, high-resolution proxy records. The reconstruction assimilates 94 proxy time series, including oxygen isotope records from ice cores and tree rings, historical documentary sources, tree-ring width series, long-term precipitation station data, Amazon River water-level observations, and ice accumulation records. Proxies were selected based on statistically significant correlations with the instrumental SASM index and consistent data availability from 1850 onward. The OIE method combines elements of local regression, Bayesian inference, and Generalized Likelihood Uncertainty Estimation (GLUE), producing 1,500 ensemble members that reflect uncertainties in proxy selection and calibration period. Ensemble weighting is based on cross-validated root-mean-square error (RMSE), enabling robust reconstruction of both high- and low-frequency variability in the monsoon system. This dataset includes the full ensemble as a NetCDF file (sasmi variable with dimensions [series, year]), as well as summary statistics such as the ensemble mean. It is intended to support research on past hydroclimate variability, monsoon dynamics, and model-data comparison.