Using artificial intelligence for exploring the climatic drivers of the Venice Lagoon environmental variability and response to climate change.

Understanding the response to climate change of the Venice Lagoon is fundamental for the conservation and sustainable management of a vulnerable environment, with important ecological and socio-economic consequences. Deterministic dynamic models that can reproduce the behavior of the lagoon have a very high computational cost, that limits substantially their applicability, particularly considering the multiple and multidecadal simulations required to analyses climate change. This study explores the use of artificial neural networks (ANNs) to model the relationships between climate drivers and key parameters (temperature and salinity) of the Venice lagoon to understand their different dynamics within the lagoon environment. We carry on a sensitivity study on the various drivers utilized and examine the simultaneous presence of different response patterns within the lagoon. The analysis is based on the combination in situ observations of the lagoon water temperature and salinity with large-scale data from the Copernicus Marine Services’ reanalysis  to estimate how the main physical parameters of the lagoons are driven by key climatic drivers. The sensitivity analysis was conducted by excluding from the ANN or randomizing single drivers to assess their importance for describing the variability of the lagoon environment. This analysis allow to identify three clusters, defining three areas of the lagoon, whose differences that can be physically interpreted. The riverine cluster (central/northern lagoon) is influenced by the presence of small tributaries and, consequently, by local precipitation; The marine cluster is located in the part of the lagoon near the sea outlets, where salinity and temperature values are strongly influenced by marine salinity and temperature; The mixed cluster  (in the south lagoon) where both the marine and riverine regimes overlap with comparable effects on salinity and temperature.

Financial support from ICSC – Centro Nazionale di Ricerca in High Performance Computing, Big Data and Quantum Computing, funded by European Union – NextGenerationEU. Project code CN_00000033, CUP C83C22000560007 and  from NBFC – National Biodiversity Future Center, funded by European Union – NextGenerationEU. Project code CN_00000033, CUP F87G22000290001