An optimal transformation method for inferring ocean tracer sources and sinks

Abstract

The geography of changes in the fluxes of heat, carbon, freshwater and other tracers at the sea surface is highly uncertain and is critical to our understanding of climate change and its impacts. We present a state estimation framework wherein prior estimates of boundary fluxes can be adjusted to make them consistent with the evolving ocean state. In this framework, we define a discrete set of ocean water masses distinguished by their geographical, thermodynamic and chemical properties for specific time periods. Ocean circulation then moves these water masses in geographic space. In phase space, geographically adjacent water masses are able to mix together, representing a convergence, and air–sea property fluxes move the water masses over time. We define an optimisation problem whose solution is constrained by the physically permissible bounds of changes in ocean circulation, air–sea fluxes and mixing. As a proof-of-concept implementation, we use data from a historical numerical climate model simulation with a closed heat and salinity budget. An inverse model solution is found for the evolution of temperature and salinity that is consistent with “true” air–sea heat and freshwater fluxes which are introduced as model priors. When biases are introduced into the prior fluxes, the inverse model finds a solution closer to the true fluxes. This framework, which we call the optimal transformation method, represents a modular, relatively computationally cost-effective, open-source and transparent state estimation tool that complements existing approaches.