Modeling mixing for the giss coupled climate model climate change is one of the greatest challenges facing humanity. In order to respond to IT we must comprehend IT, therefor the further development…

Modeling mixing for the giss coupled climate model climate change is one of the greatest challenges facing humanity. In order to respond to IT we must comprehend IT, therefor the further development of climate science is an urgent matter. Due to the complexity of the climate system, numerical climate models, such as NASA gisss model E, have been developed to study IT. These numerical models require parameterizations of processes that cannot be explicitly resolved. Among these processes are the subgridscale mixing in the ocean of momentum, heat, salt and passive tracers.improved physically based parameterizations of the subgridscale mixing by smallscale turbulence, submesoscale eddies and mesoscale eddies have been developed by the giss turbulence group vittorio canuto, YE cheng, mikhail dubovikov, armando howard and anthony leboissetie]. This 4 one fourth month project seeks to implement the latest versions in the giss model E ocean component. If they perform satisfactorily in future tests they will then be available for use in the giss coupled climate model experiments and more generally to improve climate simulation. Since the giss model E is used for climate simulations, IT is particularly important that wherever possible the parameterizations in IT rely on our understanding of the physics as opposed to being based on phenomenology, as the cimate system may evolve to conditions beyond the range of data in which a phenomenological parameterization was set and into a range where IT is inapplicable. When on the other hand a subgrid model is derived from physical principles we can have some confidence that IT will be robust and correctly produce results even for a changed climate. We therefore seek in this project to improve the physical content of our ocean mixing parameterizations. With respect to diapycnal mixing by small scale turbulence , the current version of giss model E uses the K profile parameterization KPP 1, but the latter relies heavily on phenomenology and heuristic arguments. We aim to attain a more reliable parameterization of this mixing by using a model based on turbulence theory. The giss ocean vertical mixing scheme gissvm 2,3 calculates separate diffusivities for momentum, heat, salt and passive tracers throughout the water column, from the mixed layer, to the interior of the ocean to the bottom. IT uses a unified reynolds stress model approach to do so while encompassing different sources of mixing including static instability, shear instability, double diffusive instabilities which mix salt differently than heat, internal waves and the interaction of tidal currents with bottom topography. Tests with a preliminary local version of the gissvm in model ES stand alone ocean using the core protocol were promising 4, however issues remained, notably with mixedlayer depths being too shallow in the ACC. We suspect that the omission of the physical process of non local turbulent transport is one cause of our model S present difficulties. Our first objective, therefore, is to implement in the model E stand alone ocean a new version of the giss VM which includes non locality. Mixing by sub mesoscale eddies, which are a few kilometers in size, has important impacts, especially on the mixed layer, as IT can restratify the upper water column, largely undoing in some cases the destratification caused by small scale turbulence. This process is, however, completely absent in the current model E ocean. Our second objective therefore is to implement in model E a parameterization of the ocean sub mesoscale eddy effects that has been recently developed by the giss turbulence group. As for mixing by mesoscale eddies giss model E currently uses the gent mcwilliams mesoscale eddy parameterization GM 5. This parameterization has shown considerable success in many ocean models, but there is still major concern that IT is not a complete description of the eddys effects. A serious difficulty is that while the GM para