Coupling Different Hierarchies of Climate Models with the CliMA Coupler

The CliMA model is an Earth system model (ESM) which is being developed to incorporate data-driven parameter estimation techniques with complex physical models. It is designed to be modular and flexible to seamlessly couple models of different spatial and temporal timescales, different prognostic variables, and different approaches of estimating shared parameters and their uncertainties. The connecting component, the CliMA Coupler, manages the communication and transformation of this information between components and controls the time evolution of the coupled system. It is designed to allow for flexible substitution of model components of varying complexity and to accurately and conservatively account for component models with different spatial discretizations and timestepping methods. Drawing inspiration from existing coupling frameworks and libraries, the CliMA Coupler is being written from the ground up using the Julia programming language. The coupler will be used to couple both high resolution large eddy simulations (LES) and the large-scale CliMA ESM. Ensembles of coupled turbulence-resolving LES simulations of atmosphere-ocean and atmosphere-land interactions will serve as synthetic training data for the CliMA ESMs Bayesian learning machinery. We present our design with preliminary results of the CliMA Coupler, its current and future functionality, and the value of this design for hierarchical modelling from idealized toy models to complex Earth systems within one modelling framework.