CycloneMAE: A Scalable Multi-Task Learning Model for Global Tropical Cyclone Probabilistic Forecasting

TL;DR

CycloneMAE is a scalable multi-task deep learning model that leverages multi-modal data and a masked autoencoder structure to improve probabilistic tropical cyclone forecasting across multiple variables and global basins.

Contribution

It introduces a novel multi-task, probabilistic forecasting framework using a structure-aware masked autoencoder for transfer learning in tropical cyclone prediction.

Findings

Outperforms leading NWP systems in pressure and wind forecasting up to 120 hours.

Achieves better track forecasting accuracy up to 24 hours.

Provides physically interpretable learning dynamics through attribution analysis.

Abstract

Tropical cyclones (TCs) rank among the most destructive natural hazards, yet their forecasting faces fundamental trade-offs: numerical weather prediction (NWP) models are computationally prohibitive and struggle to leverage historical data, while existing deep learning (DL)-based intelligent models are variable-specific and deterministic, which fail to generalize across different forecasting variables. Here we present CycloneMAE, a scalable multi-task forecasting model that learns transferable TC representations from multi-modal data using a TC structure-aware masked autoencoder. By coupling a discrete probabilistic gridding mechanism with a pre-train/fine-tune paradigm, CycloneMAE simultaneously delivers deterministic forecasts and probability distributions. Evaluated across five global ocean basins, CycloneMAE outperforms leading NWP systems in pressure and wind forecasting up to 120 hours and in track forecasting up to 24 hours. Attribution analysis via integrated gradients reveals physically interpretable learning dynamics: short-term forecasts rely predominantly on the internal core convective structure from satellite imagery, whereas longer-term forecasts progressively shift attention to external environmental factors. Our framework establishes a scalable, probabilistic, and interpretable pathway for operational TC forecasting.