Introduction

Abstract

The pioneering work on statistical turbulence closure schemes performed in the 1970s by many different researchers has been concisely reviewed by Rodi [1980]. In those days, complex turbulence schemes were mainly applied in the field of computational fluid dynamics with its well-defined model problems. This restriction to technical and engineering applications can be explained by several arguments: clearly defined model domains and forcing, higher relevance of complex physical features such as non-hydrostatic pressure, bigger computer resources due to better funding e.g. from nuclear research programs, and maybe some more reasons. Apart from technical and engineering applications, higher order turbulence closure schemes were more and more applied in atmospheric sciences as well, see e.g. the famous article by Mellor and Yamada [1974]. There, the demand for better turbulence closure models came mainly from meteorological institutes which wanted to improve their weather prediction models. In oceanography however, complex turbulence models were introduced rather late, see e.g. Rosati and Miyakoda [1988]. The reason might be that simple bulk models or length scale parametrisations gave reasonable answers and the scientific focus (understanding of general circulation) was not so much related to turbulence closure modelling. The limited computational resources were mainly used for achieving fine spatial resolution and long model runs, see e.g. the recent study by Demirov and Pinardi [2001] for the Mediterranean Sea.