Deep Learning Imputation of Missing Radius of Maximum Winds (Rmax) Values in Tropical Cyclone Best-Track Data

2026-08-10

AI Summary

This study addresses the frequent absence of the radius of maximum wind (Rmax) in tropical cyclone best-track datasets, which limits the accuracy of probabilistic coastal hazard assessments. To overcome this, the authors propose a data-driven imputation method combining one-dimensional convolutional neural networks (1D-CNNs) and long short-term memory (LSTM) networks, enhanced with physically informed input features and explicit modeling of temporal dependencies. The model is pretrained on synthetic datasets (RAFT, STORM) and fine-tuned on observational data (IBTrACS). Results show that incorporating the 34-kt wind radius (R34) significantly improves performance; temporal models outperform non-temporal approaches and better preserve inter-storm Rmax variability, especially under limited sample sizes. When R34 is unavailable, it can partially substitute for storm-scale information. However, transfer learning yields no improvement due to distributional discrepancies in Rmax between synthetic and real-world data.

Abstract

Probabilistic coastal hazard assessments require accurate characterization of tropical cyclone (TC) parameters, yet datasets often contain missing records for the radius of maximum winds (Rmax), a key variable in Joint Probability Method analyses. This study evaluates data-driven approaches for Rmax imputation, including one-dimensional Convolutional Neural Networks (1DCNNs), Long Short-Term Memory (LSTM) networks, and conventional machine learning models. We examine physics-informed input augmentation, temporal modeling, and transfer learning using synthetic RAFT and STORM datasets for pre-training and observational IBTrACS data for fine-tuning. Including the radius of 34-knot winds (R34) substantially improves performance across all model types. Temporal models achieve higher average correlations than non-temporal models despite using approximately an order of magnitude fewer samples, indicating better preservation of relative Rmax variability across storms. This advantage is more pronounced when R34 is unavailable, suggesting temporal information can partially compensate for missing storm-size predictors. Transfer learning does not improve performance, likely because synthetic datasets have lower and less variable Rmax distributions than IBTrACS. These findings demonstrate the potential of temporal deep learning for reconstructing incomplete TC records and highlight the importance of physics-informed inputs, observational data availability, and distributional consistency in coastal hazard assessment.

Problem

• Rmax imputation
• tropical cyclone
• missing data
• coastal hazard assessment
• best-track data

Innovation

• temporal deep learning
• Rmax imputation
• physics-informed inputs
• tropical cyclone best-track data
• transfer learning