Sea-Level Change in the Northern Mediterranean Sea from Long-Period Tide Gauge Time Series

The oldest tide gauge observations date back to the 18th century. Although, globally, they are available in limited number, these centuries-old sea level time series are the only data records providing information on the long-period rates of change of the mean ocean surface. Knowledge of the past sea level behavior can contribute key insights to the understanding of climate change impacts. We highlight the greatest importance of monitoring sea-level changes at all spatial scales, from global to local, using terrestrial and space techniques and outline the physical processes, natural and man-induced, responsible for such changes. In general, tide gauge data are made available through different archiving facilities serving both international and national developments. Tide gauges measure local sea-level relative to a benchmark on land, hence, correctly interpreting these observations is challenging since it demands, among other requirements, a proper knowledge of vertical land motions at the stations. In general, it is not easy to find well documented historical data; moreover, benchmarks were not frequently leveled. For more than two decades, space geodetic techniques, such as GNSS (Global Navigation Satellite System) and InSAR (Interferometric Synthetic Aperture Radar), have provided the opportunity to accurately position points in the surroundings of tide gauge sites, potentially giving rise to a large amount of information. However, despite the availability of these techniques, the evolution of the international efforts aiming at realizing consistent observational infrastructures for sea level networks is undergoing only a slow development. In the Mediterranean area, there are a few centennial tide gauge records. Our study focuses on the time series of Alicante, in Spain, Marseille, in France, Genoa, Marina di Ravenna (formerly Porto Corsini), Venice and Trieste, in Italy. After briefly reviewing the gauge types presently in use for sea level measurements, a comprehensive historical description is given for each time series, which may assist understanding an assessment of the problems these stations have experienced over more than one century of operations. Two Italian stations, Marina di Ravenna and Venice, are affected by both natural and anthropogenic subsidence, the latter was particularly intense during a few decades in the 20th century because of ground fluid withdrawal. For these two stations, we have retrieved leveling data of benchmarks close to the tide gauges from the end of the 19th century and, for the last couple of decades, we have evaluated GPS and InSAR heights in close proximity to the stations. The GPS (Global Positioning System) and SAR results were carefully compared. Modeling of the long-period non-linear behavior of subsidence was successfully accomplished by using an ensemble of leveling, GPS and SAR data. After removing the vertical land motions in Venice and Marina di Ravenna, and the inverted barometer effect at all the sites, the linear long period sea-level rates were estimated. The results are in excellent agreement ranging between + 1.2 and + 1.3 mm/year for the overall period from the last decades of the 19th century till 2012. The associated errors, computed by accounting for serial autocorrelation, are of the order of 0.2–0.3 mm/year for all stations, except Alicante, for which the error turns out to be 0.5 mm/year. Our estimated rates for the northern Mediterranean, a relatively small regional sea, are slightly lower than the global mean rate, + 1.7 ± 0.2 mm/year, recently published in the IPCC AR5 (Intergovernmental Panel on Climate Change 5th Assessment Report) (Church et al., 2013), but close enough, if uncertainties are taken into account. It is known that Mediterranean stations had always had lower trends than the global-average ones. Our regional results, however, are in close agreement with the global mean rate, + 1.2 mm/year, published by Hay et al. (2015) which is currently being discussed by the oceanographic community (see, for example, Hamlington and Thompson, 2015). The six time series were also analyzed by means of the EOF (Empirical Orthogonal Functions) technique over the 1934–2012 common period. As a result, about 50% of the total variance is explained by the first mode, which is characterized by a coherent behavior of the six stations.