Preprints
https://doi.org/10.5194/egusphere-2024-1636
https://doi.org/10.5194/egusphere-2024-1636
04 Jun 2024
 | 04 Jun 2024
Status: this preprint is open for discussion and under review for Atmospheric Chemistry and Physics (ACP).

Opinion: Why all emergent constraints are wrong but some are useful – a machine learning perspective

Peer Nowack and Duncan Watson-Parris

Abstract. Global climate change projections are subject to substantial modelling uncertainties. A variety of emergent constraints, as well as several other statistical model evaluation approaches, have been suggested to address these uncertainties. However, they remain heavily debated in the climate science community. Still, the central idea to relate future model projections to already observable quantities has no real substitute. Here we highlight the validation perspective of predictive skill in the machine learning community as a promising alternative viewpoint. Building on this perspective, we review machine learning ideas for new types of controlling factor analyses (CFA). The principal idea behind these CFA is to use machine learning to find climate-invariant relationships in historical data, which also hold approximately under strong climate change scenarios. On the basis of existing data archives, these climate-invariant relationships can be validated in perfect-climate-model frameworks. From a machine learning perspective, we argue that such approaches are promising for three reasons: (a) they can be objectively validated both for past data and future data, (b) they provide more direct – by design physically-plausible – links between historical observations and potential future climates and (c) they can take higher dimensional relationships into account that better characterize the still complex nature of large-scale emerging relationships. We demonstrate these advantages for two recently published CFA examples in the form of constraints on climate feedback mechanisms (clouds, stratospheric water vapour), and discuss further challenges and opportunities using the example of a climate forcing (aerosol-cloud interactions).

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.
Peer Nowack and Duncan Watson-Parris

Status: open (until 16 Jul 2024)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
Peer Nowack and Duncan Watson-Parris
Peer Nowack and Duncan Watson-Parris

Viewed

Total article views: 154 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
106 41 7 154 10 7
  • HTML: 106
  • PDF: 41
  • XML: 7
  • Total: 154
  • BibTeX: 10
  • EndNote: 7
Views and downloads (calculated since 04 Jun 2024)
Cumulative views and downloads (calculated since 04 Jun 2024)

Viewed (geographical distribution)

Total article views: 151 (including HTML, PDF, and XML) Thereof 151 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 12 Jun 2024
Download
Short summary
In our Opinion article, we review uncertainties in global climate change projections and current methods using Earth observations to constrain them, which is crucial for climate risk assessments and for informing society. We then discuss how machine learning can advance the field, discussing recent work that provides potentially stronger and more robust links between observed data and future climate projections. We further discuss challenges of applying machine learning to climate science.