Reproducible pipeline for the integrated downscaling–BMA–BiLSTM framework: CMIP6 GCM evaluation, two-track downscaling (LARS-WG 8 + DQM), EM-based Bayesian Model Averaging, deep-ensemble streamflow projection, and drought / ETCCDI / compound-extreme analysis. Demonstrated on the Haraz basin, northern Iran.
deep-learning reproducible-research iran hydrology drought uncertainty-quantification cmip6 climate-change bilstm bayesian-model-averaging statistical-downscaling quantile-mapping deep-ensemble compound-extremes streamflow-prediction lars-wg
-
Updated
Aug 12, 2026 - Python