A novel data-driven method using Random Forest Regression and feature importance can identify and rank the interdependencies between climate variables, enabling the tracing of source-impact pathways from spatio-temporal climate data.
