The Gravity Recovery and Climate Experiment (GRACE) and its Follow-On Mission (GRACE-FO) have transformed our understanding of global terrestrial water storage (TWS) dynamics by revealing previously unknown mass changes. However, persistent data gaps induced by sensor breakdowns, orbital constraints, and processing aberrations significantly limit their utility in crucial hydrological applications that necessitate a continuous, uninterrupted time series. This study examines four different imputation methods: two traditional statistical approaches (linear interpolation and mean substitution) and two sophisticated machine learning techniques (Random Forest regression and K-nearest neighbors). We use 18 years of GRACE/GRACE-FO measurements spanning 2002-2020 from 15 globally distributed river basins covering varied hydroclimatic regimes (including tropical, temperate, and dry systems) to execute a rigorous evaluation framework. The methodology uses three controlled gap situations (10%, 15%, and 20% missing data) to model realistic data interruption patterns, as well as two complementing performance metrics (Root Mean Square Error and R-squared) to evaluate both accuracy and variance explanation. Our thorough investigation shows that, while traditional methods perform well for short gaps, machine learning approaches outperform them when dealing with bigger data discontinuities. Specifically, the Random Forest approach outperforms all investigated cases by effectively retaining both long-term trend components and seasonal amplitudes in the rebuilt time series. These findings provide hydrologists and water resource managers with an evidence-based framework for selecting effective gap-filling strategies that are suited to individual application requirements. The improved data continuity improves the reliability of GRACE/GRACE-FO datasets for critical applications such as groundwater monitoring, drought early warning systems, and climate change impact assessments, especially in data-scarce regions with few alternative observations.