Machine Learning for Stochastic Parametrisation

Hannah M. Christensen, Salah Kouhen, Greta Miller, Raghul Parthipan

Environmental Data Science · 01 January 2024 · University of Cambridge (Aardvark), University of Oxford

In one sentence. A position paper argues for combining stochastic parameterization with machine learning, reviewing early studies and outlining open challenges for data-driven stochastic subgrid schemes.

Abstract

Atmospheric models used for weather and climate prediction are traditionally formulated in a deterministic manner. In other words, given a particular state of the resolved scale variables, the most likely forcing from the sub-grid scale processes is estimated and used to predict the evolution of the large-scale flow. However, the lack of scale-separation in the atmosphere means that this approach is a large source of error in forecasts. Over recent years, an alternative paradigm has developed: the use of stochastic techniques to characterise uncertainty in small-scale processes. These techniques are now widely used across weather, sub-seasonal, seasonal, and climate timescales. In parallel, recent years have also seen significant progress in replacing parametrisation schemes using machine learning (ML). This has the potential to both speed up and improve our numerical models. However, the focus to date has largely been on deterministic approaches. In this position paper, we bring together these two key developments, and discuss the potential for data-driven approaches for stochastic parametrisation. We highlight early studies in this area, and draw attention to the novel challenges that remain.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.