Skills of different hydrographic networks in capturing changes in the Mediterranean Sea at climate scales

Climate Research

G. Jord๠ J. Llasses¹  D. Gomis¹ 

关键词: Climate variability;    Mediterranean;    Monitoring;    Observing systems;    Climate change;   

DOI  :  10.3354/cr01270

来源: Inter-Research Science Publishing

【 摘 要 】

ABSTRACT: The skills of 5 observational networks are explored in the context of the monitoring of climate signals in the Mediterranean Sea. Namely we explore the capabilities of hydrographic surveys and ships of opportunity, of Argo buoys, of a (virtual) regularly distributed mooring network, of the present-day observational system (which makes use of the 3 kinds of observations) and of a targeted future system. The skills of each observational network are quantified as follows: first, the output of a realistic regional circulation model (considered here as the virtual truth) is sampled at the same time and location of the actual observations gathered by each observational network. An objective analysis scheme based on Optimal Statistical Interpolation is then applied to the pseudo-observations to obtain gridded products, which are compared to the model output in order to infer the capability of each sampling to capture the true fields. We do it for different periods (for 1962-2000 and for the whole 21st century) and for different parameters (temperature, salinity and the rate of deep water formation in the Western Mediterranean). Results indicate that the skills to reproduce large scale climatic signals depend on the depth and variable, ranging from >90% of explained monthly variance and <5% relative trend errors for the upper (0-100 m) and intermediate layer (100-400 m) temperature fields, to <60% of variance and 30% relative trend errors for the upper layer salinity field. When averaging temperature and salinity over the whole basin volume, both annual values and long term trends are properly captured by all the networks, though the deep water formation rate in the Western Mediterranean is largely overestimated. Conversely, regional features are missed by all the sampling networks, since none of them has an adequate spatial distribution to capture small scale processes.

【 授权许可】

Unknown