Biogeomorphic River Response to an Unprecedented Hydrological Drought: Evidence from the Po River (Italy)

As extreme hydrological events become increasingly frequent and intense, there is a growing need for innovative approaches to systematically monitor their impacts on riverine landscapes. This need is especially crucial in human-modified river systems, where such events can have significant consequences on the surrounding anthropized areas. The Po River in Italy exemplifies this challenge, having experienced an unprecedented drought in 2022—its lowest streamflow in two centuries—followed in 2024 by one of the most hydrologically intense years on record. This sequence of contrasting extremes makes the Po River an ideal case study for investigating morphological adjustments and assessing river sensitivity to hydrological variability.

We leveraged Sentinel-2 satellite imagery collection, spanning 2017 to 2025, on a 130-km-long segment of the Po River. A free, globally applicable Fully Convolutional Neural Network was employed to automatically classify monthly median composite images and delineate the active channel—defined as the area encompassing both flowing water and adjacent exposed, unvegetated sediment bars. We generated a continuous, updatable time series, identifying the emergence of progressively activated areas or regions undergoing gradual vegetation colonization (“deactivated” areas). By analysing changes over multiple years rather than on a year-by-year basis, this method more effectively distinguishes areas that consistently remain active or inactive from those that fluctuate between these two states. That helps separate changes driven by varying water stage from those resulting from morphological modifications, e.g. bank erosion.

Our analysis reveals that the 2022 drought was part of an extended period of hydrological scarcity lasting nearly three years. During this time, all reaches of the Po River experienced a net loss of active channel area due to vegetation encroachment. By comparing these trends with a 2022 LiDAR-derived Relative Elevation Model, we demonstrate that vegetation encroachment expanded into topographically lower zones closer to the low-flow channel that had not previously supported vegetation. This indicates a significant shift in morphological setting and ecological dynamics. The hydrologically intense conditions of 2024 triggered unprecedented bank erosion and the widespread reactivation of previously abandoned areas, particularly those deactivated during the preceding dry years. Interestingly, not all areas reactivated in 2024 persisted into 2025. We show that patterns of reactivation and the new activation of floodplain areas depend on river configuration and the degree of artificial confinement. While some reaches restored the active channel width to pre-drought levels, others have not yet fully recovered, suggesting that changes in vegetation establishment may have induced long-lasting morphological adjustments.

This approach provides a practical and scalable tool for global river monitoring, enhancing our understanding of river sensitivity to a rapidly changing climate.