Can Transfer Learning be Used to Identify Tropical State-Dependent Bias Relevant to Midlatitude Subseasonal Predictability?

TL;DR

This paper explores the use of transfer learning with explainable neural networks to identify tropical state-dependent biases in Earth system models, highlighting challenges in data requirements for improving subseasonal predictability.

Contribution

It introduces a machine learning framework employing transfer learning to detect model biases affecting subseasonal climate predictability, with insights into data limitations.

Findings

Transfer learning may need more data than current reanalysis datasets.

Neural networks can identify tropical biases in climate models.

Caution is advised for future transfer learning applications in climate prediction.

Abstract

Previous research has demonstrated that specific states of the climate system can lead to enhanced subseasonal predictability (i.e., state-dependent predictability). However, biases in Earth system models can affect the representation of these states and their subsequent evolution. Here, we present a machine learning framework to identify state-dependent biases in Earth system models. In particular, we investigate the utility of transfer learning with explainable neural networks to identify tropical state-dependent biases in historical simulations of the Energy Exascale Earth System Model version 2 (E3SMv2) relevant for midlatitude subseasonal predictability. Using a perfect model framework, we find transfer learning may require substantially more data than provided by present-day reanalysis datasets to update neural network weights, imparting a cautionary tale for future transfer learning approaches focused on subseasonal modes of variability.