Dominant spatiotemporal structures in total water storage anomalies

This study employs Dynamic Mode Decomposition (DMD) to derive a global-scale linear model for the temporal evolution of Total Water Storage Anomaly (TWSA) measured by GRACE satellite missions, with the goal of extracting and analyzing the dominant spatiotemporal structures governing TWSA variability. Our analysis differentiates modes associated with a periodic dynamic – linked to precipitation-driven seasonal cycles and multi-year variations – from those incorporating trend effects indicating, on average, a progressive TWSA decline. Focusing on the latter, we examine patterns associated with extreme TWSA values and their intensification over time. In regions experiencing significant TWSA changes over the past decade, DMD effectively distinguishes natural variability from trends, aligning with previous findings that identify climate change and human impact effects in the same regions. This study underscores DMD’s potential in capturing essential hydrological dynamics in data, thus supporting the interpretation of these dynamics at the scale of the analysis.