Prototyping Global Earth System Models at High Resolution: The Role of Comprehensiveness Touchstones Across Trade-Offs of Resolution, Comprehensiveness and Simulation Length.

Abstract:

The development of next generation Earth System Models (ESM) demands fundamental configuration decisions across three axes of computational constraint: resolution, complexity and simulation length. One of the representational aspirations driving current development is to capture the global ocean mesoscale (i.e. ocean weather) in coupled carbon-climate ESMs. This resolution constraint poses extreme limitations on both complexity (i.e. the number of biogeochemical tracers) and simulation length. The present study explores a novel experimental design in which a series of touchstones across the axes of resolution (100 km to 10 km), complexity (30 tracer full biogeochemistry to 6, 3, and 1 tracer approximations), and simulation length (millenia to decades) are explored to estimate baseline biogeochemical simulation characteristics and the biogeochemical response to climate change. Along the axis of complexity, we find biogeochemical fidelity largely reproduced with a 6 tracer version of our full 30 tracer module, which is very promising for high resolution applications. However, further reduction down to 3 tracers revealed much more pronounced biases in baseline simulation characteristics. Further, we find critical differences in simulation characteristics in the 3 tracer version, across resolutions, that challenge mechanistic attribution. Although these deficiencies may reflect specific aspects of our 3 tracer model design and are the topic of further research for future improvement, they illustrate the great challenge of adequately representing the key processes with such a small number of tracers. Nonetheless, we find that sensitivity of the biogeochemistry to climate change across the complexity bear many similarities, giving us some support that the differences we see across resolutions may indicate robust differences in biogeochemical response across resolutions.