Drought‐Induced Vertical Displacements and Water Loss in the Po River Basin (Northern Italy) From GNSS Measurements

Drought is one of the most complex recurring natural disasters, defined by a deficiency of precipitations that causes prolonged water scarcity. Failure to manage drought risk has the potential to have dire consequences for people, livelihoods, economy and ecosystems.
In northern Italy, particularly in the highly productive industrial area of the Po river basin, the 2021-2022 period culminated in the most severe drought of the last two centuries.
In order to evaluate the best policies to address the problems caused by water scarcity, it is crucial to measure and monitor variations in terrestrial water storage (TWS). For drought monitoring, in fact, changes or anomalies in TWS provide direct observations of total water availability, complementing model-based measures such as drought severity indices.
To estimate the quantities and spatial distribution of TWS loss, we analyze vertical ground displacement time-series data from Global Navigation Satellite System (GNSS) stations in the Po river basin.
We use a regularization model, based on L1-norm, to reconstruct the long-term temporal evolution of vertical ground displacement trends. Next, we performed a Principal Component Analysis (PCA) on GNSS time series to extract a spatially consistent signal in vertical ground displacements. The temporal evolution of the first principal component is well-correlated with trend changes of the Po river level and with the  SPEI-12 drought index, with stations moving upward during periods of river/index level decrease and vice versa, indicating that common long-term variations in vertical ground displacements are driven by the hydrology of the area.
The inversion of the displacements associated with the first principal component allows us to estimate variations in equivalent water height (EWH) and find that between January 2021 and August 2022, the GNSS stations underwent uplift, up to 7 mm, which corresponds to ~70 Gtons of water loss. The results are compared with the Global Land Data Assimilation System (GLDAS) model and the Gravity Recovery and Climate Experiment (GRACE) data: while the temporal evolution of the three products, when averaged over the study area, is similar, the spatial distributions are different. This is likely due to the fact that GLDAS only takes surface water into account, and GRACE has a too-coarse spatio-temporal resolution.
Our results show that multi-year changes in water storage can be effectively monitored both in terms of temporal evolution and spatial distribution using space geodetic measurements, such as GNSS. This approach eliminates the need to rely solely on large-scale models or satellite measurements, which cannot reach the spatial resolution required at the scale of river basins such as the Po.