Spatially continuous monthly precipitation stable isotope estimates across the Australian continent at 0.25° resolution from 1962–2023

A comprehensive analysis of spatio-temporal variability in Australian precipitation stable isotopes (including climatic drivers) is currently underway, based on this dataset. We expect to submit this for publication in the first half of 2026. Please contact the creator of this dataset (Georgy Falster; [e-mail]) if you are interested in hearing more about this work. Data repository for Falster et al. (2026) High resolution monthly precipitation isotope estimates across Australia from machine learning This repository contains the random forest-modelled precipitation δ²H, δ¹⁸O, and deuterium excess (dxs) datasets described in the paper High resolution monthly precipitation isotope estimates across Australia from machine learning [https://doi.org/10.5194/hess-30-289-2026]. The repository contains 27 netcdf files, which together comprise the outputs described in the paper. In all cases, please see the paper for important details on the data and how they were produced. The following files are available for each of δ²H_P, δ¹⁸O_P, and dxs_P: Netcdf files containing only the median value from the individual estimates calculated with the 50 random seeds (i.e., only latitude, longitude, and time dimensions) Monthly estimates e.g., `aus_prec.d2H_v1_196201-202312_monthly_median.nc` Monthly anomalies (relative to the long-term monthly means) e.g., `aus_prec.d2H.anoms_v1_196201-202312_monthly_median.nc` Amount-weighted annual mean estimates e.g., `aus_prec.d2H_v1_1962-2023_ann_median.nc` Anomalies in the amount-weighted annual mean e.g., `aus_prec.d2H.anoms_v1_1962-2023_ann_median.nc` Long-term annual mean (with only latitude and longitude dimensions) e.g., aus_prec.d2H_v1_1962-2023_long-term-annual-mean_median.nc In all cases, the data are provided at 0.25° resolution across the entire Australian continent, with the following two options: Longer models trained over 1962–2023, using the reduced set of predictor variables (without weather objects) Shorter models trained over 1980–2019, using the full set of predictor variables (including weather objects) Please contact the lead author if you would like the following outputs (assuming fewer if any people will be interested in these): Netcdf files including the individual estimates calculated with the 50 random seeds (i.e., latitude, longitude, time, and ensemble dimensions) Monthly estimates e.g., `aus_prec.d2H_v1_196201-202312_monthly_ensemble.nc` Amount-weighted annual mean estimates e.g., `aus_prec.d2H_v1_1962-2023_ann_ensemble.nc` How to cite this repository If using this data, please cite the original publication, available from https://hess.copernicus.org/articles/30/289/2026/