Understanding the spatiotemporal characteristics of terrestrial water storage (TWS) variation is crucial for evaluating climate change and promoting sustainable water resource management. Numerical hydrological models and satellite data, particularly information obtained from gravity variations through the Gravity Recovery and Climate Experiment (GRACE) mission, are widely utilized for the analysis of TWS fluctuations. Even though GRACE holds excellent promise for large-scale exploration, these measurements are too coarse for localized quantification. Therefore, enhancing the resolution of GRACE measurements holds significant potential for application at the local to regional scale. The objective of this study is to produce higher-resolution images capable of depicting the local-scale TWS anomalies (TWSA) in the South America continent using a modified version of the Generative Adversarial Network (GAN). Our results demonstrate that the conditional WGAN-GP model can effectively replicate the structures of the TWSA signals with a reasonably close resemblance. This indicates that the model can successfully execute the downscaling task and harness high-resolution spatiotemporal structures within the low-resolution satellite observation for unseen temporal data.