Optimizing groundwater monitoring is necessary to overcome the inherent challenges of limited accessibility, subsurface complexity, and high infrastructure costs. Optimization can be achieved by reducing redundant wells (e.g., those with similar water-level dynamics, water quality and/or hydrostratigraphic units) and by strategic additions of monitoring to complement existing networks to better capture spatial and temporal variability. We aim to develop a framework that integrates groundwater level dynamics, aquifer geology, and groundwater-surface water interactions as a foundation to optimize monitoring networks. Normalized water level time-series are used to cluster wells with a hierarchical clustering algorithm, with flexibility to apply different linkage methods. Dynamic Time Warping (DTW) is employed as a similarity metric to capture temporal alignment between wells, enabling identification of potential physical connections and time-lag relationships—an improvement over conventional point-by-point Euclidean distance. Hydrostratigraphic factors (e.g., lithology, screen depth, aquifer characterization) are incorporated post-clustering to aid further interpretation. Additionally, dynamic patterns of groundwater level and streamflow from nearby rivers are examined using cross-correlation and hysteresis analysis to investigate groundwater-surface water interactions, and subsequently classify wells as dominantly streamflow-driven or recharge-driven. Water levels from 116 wells in Alberta’s Groundwater Observatory Well Network (GOWN; 1990 to 2024) are used to demonstrate the potential of this framework to support efficient groundwater monitoring and management in Alberta. A particular focus on 14 GOWN wells in the Canadian Rockies reflects their dominant role in generating >90% of river flow in southern Alberta.