On the Representation of Mediterranean Overflow Waters in Global Climate Models

Accurate representation of the Atlantic–Mediterranean exchange in climate models is important for a reliable simulation of the circulation in the North Atlantic Ocean. We evaluate the performance of 10 global climate models in representing Mediterranean Overflow Water (MOW) over the recent period 19...

Full description

Bibliographic Details
Published in:Journal of Physical Oceanography
Main Authors: Behr, L., Luther, N., Josey, S.A., Luterbacher, J., Wagner, S., Xoplaki, E.
Format: Article in Journal/Newspaper
Language:English
Published: AMS 2022
Subjects:
Online Access:https://publications.hereon.de/id/50666
https://publications.hzg.de/id/50666
https://www.hereon.de/imperia/md/content/gkss/zentrale_einrichtungen/bibliothek/journals/2022/behr_50666.pdf
Description
Summary:Accurate representation of the Atlantic–Mediterranean exchange in climate models is important for a reliable simulation of the circulation in the North Atlantic Ocean. We evaluate the performance of 10 global climate models in representing Mediterranean Overflow Water (MOW) over the recent period 1986–2005 by using various performance metrics. The metrics are based on the representation of the climatological mean state and the spatiotemporal variability of temperature, salinity, and volume transports. On the basis of analyses and observations, we perform a model ranking by calculating absolute, relative, and total relative errors Ej over each performance metric and model. The majority of models simulate at least six metrics well. The equilibrium depth of the MOW, the mean Atlantic–Mediterranean exchange flow, and the dominant pattern of the MOW are represented reasonably well by most of the models. Of those models considered, MPI-ESM-MR, MPI-ESM-LR, CSIRO Mk3.6.0, and MRI-CGCM3 provide the best MOW representation (Ej = 0.14, 0.19, 0.19, and 0.25, respectively). They are thus likely to be the most suitable choices for studies of MOW-dependent processes. However, the models experience salinity, temperature, and transport biases and do not represent temporal variability accurately. The implications of our results for future model analysis of the Mediterranean Sea overflow are discussed.