Reproducible pipeline, code, and supporting outputs for "Record Selection and Independent Validation in Coral-Based ENSO Reconstructions"

This repository contains the reproducible analysis workflow, code, configuration files, processed input data, and supporting outputs associated with the manuscript “Record Selection and Independent Validation in Coral-Based ENSO Reconstructions.” The study evaluates how coral-record selection, network design, validation strategy, and amplitude treatment affect reconstructions of the El Niño–Southern Oscillation (ENSO). The primary analysis uses 22 monthly coral δ¹⁸O and Sr/Ca records from the CoralHydro2k compilation over 1950–1990. Record and network selection are repeated within training periods and evaluated using withheld temporal blocks. The workflow includes individual-record and pair screening, nested temporal validation of pair selection, networks of one to ten records constructed using individual-skill, site-balanced, and forward-addition strategies, variance-matching experiments, temporal-buffer sensitivity tests, seasonal and quarterly-resolution analyses, post-1990 validation, annual aggregation, and comparisons across Niño3.4, Niño3, Niño4, and a western Pacific SST target. A complementary 1900–1990 analysis evaluates century-scale record selection and the sensitivity of reconstructed ENSO variability, skewness, event characteristics, and generalized-extreme-value diagnostics to calibration, scaling, filtering, and annual-boundary definitions. The archived workflow includes the Python analysis code, configuration files, processed coral and instrumental inputs used by the analysis, record-selection and network-membership outputs, monthly reconstruction results, bootstrap and sensitivity-analysis outputs, publication figure and table manifests, and machine-readable versions of Tables S1–S33. The pipeline can be executed either from the archived processed inputs or, where the external source data are available, from the original source-data stage. The principal external datasets used in the study are the CoralHydro2k coral proxy compilation, ERSSTv5 sea-surface-temperature fields, ICOADS Release 3.0 marine-observation summaries, and the NOAA Climate Prediction Center historical Oceanic Niño Index (ONI) record. These source datasets remain attributable to their original providers and publications. The repository is intended to support transparent reproduction of the analyses and figures reported in the manuscript and to provide machine-readable outputs for inspection without rerunning the complete workflow. Citation:Deng, W. (2026). Reproducible pipeline, code, and supporting outputs for “Record Selection and Independent Validation in Coral-Based ENSO Reconstructions” [Collection]. Zenodo. https://doi.org/10.5281/zenodo.22872259