Cited by Implications of a Pervasive Climate Model Bias for Low‐Cloud Feedback

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

Peer Nowack, Duncan Watson‐Parris

Atmospheric chemistry and physics, Journal Year: 2025, Volume and Issue: 25(4), P. 2365 - 2384

Published: Feb. 21, 2025

Abstract. Global climate change projections are subject to substantial modelling uncertainties. A variety of emergent constraints, as well several other statistical model evaluation approaches, have been suggested address these However, they remain heavily debated in the science community. Still, central idea relate future already observable quantities has no real substitute. Here, we highlight validation perspective predictive skill machine learning community a promising alternative viewpoint. Specifically, argue for quantitative approaches which each constraining relationship can be evaluated comprehensively based on out-of-sample test data – top qualitative physical plausibility arguments that commonplace justification new constraints. Building this perspective, review ideas types controlling-factor analyses (CFAs). The principal behind CFAs is use find climate-invariant relationships historical hold approximately under strong scenarios. On basis existing archives, validated perfect-climate-model frameworks. From such three reasons: (a) objectively both past and data, (b) provide more direct and, by design, physically plausible links between observations potential climates, (c) take high-dimensional complex into account functions learned constrain response. We demonstrate advantages two recently published CFA examples form constraints feedback mechanisms (clouds, stratospheric water vapour) discuss further challenges opportunities using example rapid adjustment mechanism (aerosol–cloud interactions). avenues work, including strategies non-linearity, tackle blind spots ensembles, integrate helpful priors Bayesian methods, leverage physics-informed learning, enhance robustness through causal discovery inference.

Language: Английский

Citations

Climate Models Underestimate Global Decreases in High‐Cloud Amount With Warming DOI

Sarah Wilson Kemsley, Peer Nowack, Paulo Ceppi

et al.

Geophysical Research Letters, Journal Year: 2025, Volume and Issue: 52(7)

Published: April 9, 2025

Abstract Cloud feedback has prevailed as a leading source of uncertainty in climate model projections under increasing atmospheric carbon dioxide. Cloud‐controlling factor (CCF) analysis is an approach used to observationally constrain cloud feedback, and subsequently the sensitivity. Although high clouds contribute significantly toward uncertainty, they have received comparatively little attention CCF other observational analyses. Here we use for first time ‐cloud radiative focusing on amount component owing its dominant contribution high‐cloud feedback. Globally, observations indicate larger decreases cloudiness than state‐of‐the‐art models suggest. In fact, half 16 considered here predict feedbacks inconsistent with observations, likely due misrepresenting stability iris mechanism. Despite suggested strong warming, point near‐neutral net almost canceling longwave shortwave contributions.

Language: Английский

Citations

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

Peer Nowack, Duncan Watson‐Parris

Published: June 4, 2024

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

Language: Английский

Citations

Implications of a Pervasive Climate Model Bias for Low‐Cloud Feedback DOI

Paulo Ceppi, Timothy A. Myers, Peer Nowack

et al.

Geophysical Research Letters, Journal Year: 2024, Volume and Issue: 51(20)

Published: Oct. 19, 2024

Abstract How low clouds respond to warming constitutes a key uncertainty for climate projections. Here we observationally constrain low‐cloud feedback through controlling factor analysis based on ridge regression. We find moderately positive global (0.45 W , 90% range 0.18–0.72 ), about twice the mean value (0.22 ) of 16 models from Coupled Model Intercomparison Project. link this discrepancy pervasive model mean‐state bias: underestimate response because (a) they systematically present‐day tropical marine amount, and (b) sensitivity is proportional amount. Our results hence highlight importance reducing biases in both state their environmental factors accurate change

Language: Английский

Citations