Stratified, low-Richardson number shear flows, which are of crucial importance in scientific and practical contexts, place high demands on turbulence closures. A turbulence closure model that can realistically simulate such flows will be developed by taking, and if necessary modifying, existing closures and to subjecting them to a three-step screening process. Step 0 is a demand that the closures have a realistic steady state for homogeneous, stratified shear flows such as in the laboratory experiments of Van Atta's group. In Step 1 the closures are tested in isolation in a strict quantitative comparison of simulated and observed turbulent dissipation rates and length scales. The observational basis of Step 1 are extensive microstructure measurements from the tidally driven Hudson River estuary. The demand on the closures in Step 1 is that, given the observed shear and buoyancy frequency, the closures produce a close match of the levels and tidal depth-time structure of the observed turbulent mixing. In Step 2 the closures are embedded in 2D and 3D numerical models of the outflow of saline water from the Red Sea. The demand on the closure here is that it allow the numerical models to reproduce the peculiar structure of the outflow which combines a core layer at the bottom that undergoes little entrainment and dilution over a distance of over 100 km while a thick shear zone above the core layer undergoes highly energetic turbulent mixing. The difference in performance of the closures in a non-hydrostatic 2D model and a coarser-resolution, hydrostatic 3D model will be explored. The 3D model allows investigating possible effects of rotation and of the outflow channel topography. The closure schemes which will be considered include the new closure by Baumert and Peters, the Canuto et al. closure which passes Step 0 after modification, another closure previously employed by Peters et al. and Mellor/Yamada which is being used ubiquitously. The broader impacts of the proposed research will be through improved management of coastal and estuarine waters as well as research into climate and climate change as a result of improvement of turbulence closures embedded in most widely used models. An educational component is given through the proposed participation of a graduate student.