Leveraging Contextual Cues from a Conceptual Model with Predictive Skills of Machine Learning for Improved Predictability and Interpretability in the Hydrological Processes