Can Machines Learn How Clouds Work? The Epistemic Implications of Machine Learning Methods in Climate Science

2021/05/18 by Suzanne Kawamleh · 2 citations
Social Sciences · Psychology · Arts and Humanities · #Climate Change Communication and Perception #Mental Health Research Topics #Philosophy and History of Science

paper · doi:10.1086/714877

Abstract

Scientists and decision makers rely on climate models for predictions concerning future climate change. Traditionally, physical processes that are key to predicting extreme events are either directly represented (resolved) or indirectly represented (parameterized). Scientists are now replacing physically based parameterizations with neural networks that do not represent physical processes directly or indirectly. I analyze the epistemic implications of this method and argue that it undermines the reliability of model predictions. I attribute the widespread failure in neural network generalizability to the lack of process representation. The representation of climate processes adds significant and irreducible value to the reliability of climate model predictions.

Citations

• Making the Black Box More Transparent: Understanding the Physical Implications of Machine Learning

Cited by

• The Benchmarking Epistemology: Construct Validity for Evaluating Machine Learning Models
• The Immortal Science of ML: Machine Learning and the Theory-Free Ideal

Related

• Machine Learning for the Physics of Climate
• Machine Learning Climate Model Dynamics: Offline versus Online Performance
• Climate-Invariant Machine Learning
• Physics-informed machine learning: case studies for weather and climate modelling
• Machine learning for climate physics and simulations