A global-regional hierarchy approach to exploring upscale processes in km-scale Earth System models

Exploitation of more powerful supercomputers has unlocked the potential to run kilometre-grid scale global simulations. Reaching convection-permitting resolution has been highlighted as a means to both transform local-scale weather prediction and reduce long-standing biases in global climate models. The dynamical downscaling benefits of delivering convection-permitting predictions, have been long established for weather and climate applications. Explicitly representing many of the key dynamical convective processes leads to better representation of several aspects of the mesoscale phenomena that lead to high impact weather than is achievable in coarser grid-scale models which require convection to be fully parametrized. GSRM potentially unlock representation of this upscale interaction within models, not currently simulated in global models in which the influence of convection is parametrized, or in nested limited-area models in which smaller scales do not feedback onto the general circulation. By doing so, it is hypothesized that long-standing model biases, such as in large-scale circulations and their effect on global precipitation patterns, might be resolved or reduced.

In the UK, Met Office and university partner K-Scale research has been focussed on assessment of this upscale hypothesis. A traceable global-regional model hierarchy has therefore been established, exploiting the Unified Model seamless modelling framework and model development foundations. The hierarchy spans global and limited area atmosphere-only simulations across a range of grid resolutions and model physics. We exploit the hierarchy to demonstrate the influence of upscale processes on the predicted strength and variability of upper-level winds. Enabling upscale growth in our simulations results in a relative strengthening of the tropical easterly jet. Over S. America, there is evidence of a weakening of the westerly jet over the eastern Pacific and stronger easterlies over the tropical Atlantic in vicinity of the Atlantic ITCZ. Over SE Asia, there is a general strengthening of upper-level easterly winds.

We describe the further evaluation of the hierarchy, including its sensitivity to model physics choices, in the context of new year-long simulations adopting the DYAMOND3 protocol, and discuss plans and challenges of how the Met Office is looking to apply Earth system models at km-scale in the context of evolving operational NWP, climate research and machine learning capability and service development.