Modeling of the Total Alkalinity and the Total Inorganic Carbon in the Mediterranean Sea

Abstract

Measurements of the CO 2 system parameters in the Mediterranean Sea are relatively scarce and not representative for all its sub-basins. High quality data collected on May 2013 during the 2013 MedSeA cruise covering the whole basin were used to provide for the first time linear relationships estimating the total alkalinity (A T) and the total dissolved inorganic carbon (C T) from salinity in each Mediterranean basin and sub-basin at different depth layers. These correlations show that a substantial quantity of alkalinity is added to the seawater during its residence time in the Mediterranean Sea, whereas the biological processes, the air-sea exchange and the high remineralization rate are responsible of the high C T concentrations in this sea. Moreover, these fits could be used to estimate the A T and C T from salinity where there are not available measurements of the carbonate system parameters.

A Regional Neural Network Approach to Estimate Water-Column Nutrient Concentrations and Carbonate System Variables in the Mediterranean Sea: CANYON-MED

“…Generally, CANYON-MED neural networks are more accurate than other methods with the same inputs. This is not (Sauzède et al, 2017), CANYON-B (Bittig et al, 2018), CANYON-MED (this paper), CONTENT (Bittig et al, 2018), A T -S and C T -S (Hassoun et al, 2015), LIARv2, LINR, and LIPHR (Carter et al, 2018) for NO − 3 , PO 3− 4 , Si(OH) 4 , A T , C T , pH T . surprising as it was built specifically for the Mediterranean Sea whereas CANYON, CANYON-B, CONTENT, and the LIRs were developed for the global ocean.…”

“…CONTENT has very similar errors in the prediction of carbonate system parameters compared to CANYON-B. The equations from Hassoun et al (2015), predict A T and C T with errors up to four times the errors of CANYON-MED. Moreover, the LIRs clearly stand out, with errors up to ten times higher than the other methods.…”

“…surprising as it was built specifically for the Mediterranean Sea whereas CANYON, CANYON-B, CONTENT, and the LIRs were developed for the global ocean. However, high MAE and RMSE were obtained using A T -S and C T -S from Hassoun et al (2015). This can be explained by the fact that, while having been developed specifically for the Mediterranean Sea, these equations were derived using data collected only in May whereas our validation dataset covers as much as possible the whole year.…”

“…Furthermore, deep A T ranges between 2590 and 2610 µmol.kg −1 (Figure 4D) corresponding to the deep waters of the Western and Eastern Mediterranean Sea respectively. The difference between the two basins stems from an eastward increasing trend for A T mirroring the increase in salinity and an eastward increase in pH T (Hassoun et al, 2015). However, 2 | Number of points in the training and validation datasets and statistics between in situ measurements from the validation database and the values predicted by CANYON-MED's "best topology" (ANN-1) and CANYON-MED (ANN-E) for NO − 3 , PO 3− 4 , Si(OH) 4 , A T , C T and pH T .…”

Temporal variability of the carbonate system and air-sea CO2 exchanges in a Mediterranean human-impacted coastal site

“…In the Med-Sea, several linear relationships between A T and salinity in the surface waters have been proposed for different sub-basins (e.g., Copin- Table 3 Distributions of salinity measured at the SOLEMIO site, the Rhone River discharge water (m 3 .h −1 ), the precipitation rate (mm), the wind direction (°) and speed (m.s −1 ) and maximum and minimum salinity values measured by the four buoys of the STPS network (see Fig. 1 Wimart-Rousseau et al Estuarine, Coastal and Shelf Science xxx (xxxx) xxx-xxx Montégut and Bégovic, 2002;Hassoun et al, 2015;Ingrosso et al, 2016;Rivaro et al, 2010;Schneider et al, 2007). However, in the BoM, over the two-year time-series considered, no significant linear relationship between surface A T and surface salinity has been observed ( Fig.…”