Modeling flood inundation in a changing environment is a challenging task, as it requires computationally intensive simulations of large-scale hydrologic processes and detailed hydraulic responses driven by ensembles of downscaled future meteorological forcings. With advances in computing power, data availability, and modeling tools, we now have the capability to overcome many of these challenges. On the hydroclimate modeling side, various downscaling and bias-correction techniques have been developed to transform raw Earth system model outputs into actionable inputs for ensemble hydrologic and water management simulations, thereby improving long-term water resources planning. On the hydraulic modeling side, new software frameworks that fully leverage modern GPU-centric high-performance computing systems have been developed, enabling high-resolution hydrodynamic simulations of flood propagation with fewer simplifications. In this talk, I will present ongoing research efforts at Oak Ridge National Laboratory focused on developing a modeling framework for future runoff and streamflow projections, a GPU-accelerated two-dimensional hydrodynamic model – TRITON (https://triton.ornl.gov/), and their integrated application for studying flood inundation under changing environmental conditions. Several publicly available datasets and tools will be introduced, along with discussions of key challenges, emerging opportunities, and the potential role of artificial intelligence and machine learning in advancing this modeling framework.