Energy and CO2 Footprint of Climate Model Intercomparison Projects

Earth System Models (ESMs) used in climate research, particularly for Model Intercomparison Projects (MIPs), are increasingly reliant on High-Performance Computing (HPC) resources. Their computational demands are escalating due to finer spatial resolutions, integration of complex biogeochemical processes, and larger climate ensembles to manage uncertainty. This growth in computational intensity contributes to higher energy consumption and associated carbon footprints, as increases in peak computing performance have historically outpaced energy efficiency improvements, and the environmental costs of these projects are not yet adequately quantified.

Why it matters

Understanding and quantifying the energy and carbon footprint of computational climate modeling is crucial for sustainable research practices and effective resource allocation. The escalating demand for computing power necessitates strategic planning to mitigate environmental impacts while advancing critical climate science.

Key insights

• Earth System Models (ESMs) are highly dependent on High-Performance Computing (HPC) resources for global climate simulation.
• Computational demands of ESMs are increasing due to finer spatial grid resolutions, integration of complex biogeochemical processes (e.g., atmospheric chemistry, interactive vegetation, land use, ice sheets), and the need for larger climate ensembles.
• Historically, growth in peak computing performance (FLOP/s) has exceeded improvements in energy efficiency (FLOP/Watt), leading to increased total HPC power consumption.
• The computational and environmental costs, specifically energy and CO2 footprint, of Model Intercomparison Projects (MIPs) in climate research have not been adequately quantified.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.23509