Assessing the Intensity Decay of US Landfalling Tropical Cyclones: Long Term Trends and Prediction Model Evaluations

Tropical cyclones (TCs) have caused significant damages in the U.S. after making landfall. Understanding and predicting changes in inland TC intensity is crucial for improving TC forecasting and hazard mitigation. This study aims to assess long-term trends in TC intensity changes and the predictive skills of two statistical models. To characterize the changes in TC intensity (measured by maximum wind speeds), we estimated the decay timescale of 50 TCs using the HURDAT2 dataset and a simple exponential model. We find a significant increase in the decay timescale of inland TCs by 1.6 h decade −1 from 1967 to 2023, indicating slower intensity decay. This increase is especially pronounced after 1994, coinciding with warmer temperatures in the North Atlantic basin. To understand the impacts of the increased decay timescale on TC intensity models, we compared the wind speeds from two prediction models with the observations. The first model is a simple exponential model (hindcast model), and the second was developed by Kaplan and DeMaria in 1995 (KD95 model). Both models demonstrate reasonable performance in predicting inland TC wind speeds, with the hindcast model showing better accuracy. While both models show slight decreases in predicted wind speed errors from 1967 to 2023, these trends are not statistically significant. This suggests that the models remain effective for predicting inland TC wind speeds despite the increased decay timescale. To further reduce wind speed errors, the KD95 model should account for varying TC decay rates when correcting errors associated with proximity to the coastline.