(2011) and Edman and Omstedt (2013) and classify the values of C = 0–1 and (1 − r) = 0–1/3 as indicating good agreement and strong correlation, the values of C = 1–2 and (1 − r) = 1/3–2/3 as indicating reasonable agreement and moderate correlation, and the values of C > 2 and (1 − r) > 2/3 as indicating poor agreement

and weak or negative correlation. The baroclinic equations (Eqs. (6) and (7)) and the water balance equations (Eqs. (1) and (2)) were Romidepsin used to model the water exchange through the Gibraltar Strait and Sicily Channel and the results are illustrated in Fig. 2. Surface and deeper flows through the Gibraltar Strait were calculated and the long-term means were estimated to be 0.65 × 106 m3 s−1 and 0.63 × 106 m3 s−1, respectively. The surface and deep flows through the Sicily Channel were calculated as long-term means to be 0.95 × 106 m3 s−1 and 0.93 × 106 m3 s−1, respectively, almost 40% greater than the Gibraltar Strait flows. There are clear annual variations in the flows through the Gibraltar Strait but no strong annual variability in the flows through the Sicily Channel. The flows through the Gibraltar Strait and Sicily Channel displayed positive significant trends of 0.0009 × 106 m3 s−1 yr−1

and 0.0004 × 106 m3 s−1 yr−1, respectively. The present paper uses various reanalysis datasets instead of direct observations to validate the model results. Reanalysis data give a superior learn more state estimate, produced by combining models with observations covering large spatial and temporal scales. By contrast, observations do not cover the Mediterranean Sea spatial distribution and are valid only over a specific range of times. The current study uses three of the best relevant datasets to validate the modelling results. The NCEP dataset was used to validate weather variables (Jakobson et al., 2012) MEDAR and NODC Pyruvate dehydrogenase lipoamide kinase isozyme 1 datasets were used to validate oceanic

variables (Rixen et al., 2005 and Shaltout and Omstedt, 2012). Validations of the PROBE-MED version 2.0 model were performed for surface temperature, surface salinity, evaporation, net heat loss, solar radiation, and total heat loss through the two sub-basins. Fig. 3 classifies the results by dividing the statistics into three fields: an inner field (good agreement between reanalysed and modelled results), middle field (reasonable agreement between reanalysed and modelled results), and outer field (poor agreement between reanalysed and modelled results). In both the WMB and EMB, five of the six studied parameters are well modelled. However, monthly average sea surface salinities are not modelled satisfactorily over the two studied sub-basins (Fig. 3). There is an insignificant bias of less than 0.2% between the PROBE-MED version 2.0 model calculations and the reanalysed monthly averaged sea surface salinity data, but the resolution of the observed and modelled data differ greatly (see discussion below). Generally, the PROBE-MED version 2.