In air–sea interaction, interfacial exchanges of momentum, heat, and mass are governing factors, yet closure models for these fluxes remain difficult to construct under strong nonlinearity. Motivated by simplified air–sea coupling models, we use gas–liquid two-phase flow simulations to investigate temporally ordered physical processes as causal pathways relevant to interfacial exchange. The present work evaluates how established information-theoretic causality tools can be used as practical components in data-driven modeling of interphase fluxes from simulation time series. We adopt two tools, transfer entropy (TE) [1] and Synergistic–Unique–Redundant Decomposition (SURD) [2], to analyze multivariate time series extracted from interface-resolved two-phase flow direct numerical simulations. For each target interphase flux, we treat its short-time tendency as the effect variable to align causal inference with directed temporal influence. TE-based causal analysis has been shown to recover physically meaningful directed processes in turbulence DNS, including energy transfer across scales in the cascade [3]. Building on our previous TE study for nonlinear modeling [4], we first use high TE values to screen candidate directed links and infer plausible causal delays among drivers (e.g., void fraction, slip velocity, pressure gradients, interfacial area, and turbulence measures) and flux tendencies. TE has difficulty quantifying nonlinear interaction effects, but SURD can interpret these as synergistic dominance in causality, which may inform the identification of basis functions for nonlinear model equations. References [1] T. Schreiber, Physical Review Letters, 85(2), 461 (2000). [2] C. Martínez-Sánchez, G. Arranz, and A. Lozano-Durán, Nature Communications, 15(1), 9296 (2024). [3] R. Araki, C. Martínez-Sánchez, and A. Lozano-Durán, Journal of Physics: Conference Series, 2753, 012001 (2024). [4] K. Kohyama, R. Irie, and M. Hisada, EGU General Assembly 2025, EGU25-3480 (2025).