Machine Learning and Earth System Modeling - From Parameter Calibration to Feature Detection

Explore machine learning applications in Earth system modeling through this conference talk from the Kavli Institute for Theoretical Physics. Delve into parameter calibration and feature detection techniques used to advance climate science understanding. Learn about carbon cycle modeling, land models, and atmospheric river detection. Discover how machine learning is applied to climate model outputs, including front detection, seasonal patterns, and extreme precipitation events. Gain insights into the Climate Net project and the Earth System Data Science Initiative. Understand the challenges and opportunities in using big data and machine learning algorithms to inform society about future climate changes at regional and local scales.

Introduction
Carbon Cycle
Land Models
Basic Setup
Training Data
Input Parameters
Feature Input Parameters
parameter distributions
parameter uncertainty
leverage
motivation
machine learning in earth science
climate net project
climate model output
atmospheric river detection
single input field
front detection
labeled data
seasonal front crossing
validation
delta
jet response
precipitation extremes
dipole response
seasonal response
extreme precipitation
SmartSim
Earth System Data Science Initiative
Summary
Parameters