Coastal rivers provide freshwater that supports healthy ecosystems and supplies water for people and agriculture. However, climate change is making it increasingly difficult to predict how much water these rivers will carry in the future. Our study focuses on rivers along the Gulf of Mexico and explores how climate and land conditions influence their flow. We used an explainable machine learning model that combines satellite-based climate data with algorithms trained on historical patterns. Unlike traditional models, our hybrid approach not only predicted daily river flows more accurately but also helped identify which climate factors and geographic areas have the greatest influence. The model pointed to runoff and soil moisture as the most important drivers and highlighted regions that strongly affect streamflow. When tested with future climate scenarios, it showed a consistent decline in low flows and an increase in flow variability. These trends suggest growing risks for water availability, ecosystems, and communities that depend on these rivers. By combining high-resolution climate inputs with a transparent modeling approach, our work offers a valuable tool to support climate adaptation and water resource planning in vulnerable coastal environments.