Disagreement in Detected Heatwave Trends Resulting From Diagnostic Methods

Abstract

Heatwaves pose increased risk to ecosystem and society. Advanced event-based detection methods offer novel insights into the spatiotemporal dynamics of heatwaves. However, robust assessments of heatwave trends remain challenging due to the sensitivity of results to parameter selection, including connected structure, area threshold, and overlap ratio. Here, we employed 3-dimensional connected detection algorithm (3DCDA) to explore the dependency of heatwave identifications on various parameter combinations. The results indicate that heatwave metrics are sensitive to 3DCDA parameters, with the appropriate combination for large-scale heatwaves being overlap ratio of 40%–50%, area threshold of 10⁶ km², and 10-connected structure. Based on this configuration, we discovered significant increasing trends in heatwave frequency, projection area and total magnitude, while observed a decrease in maximum intensity and area mean intensity both annually and during summer. These findings highlight the critical importance of parameter selection in 3DCDA for robust analysis of heatwaves and other extreme events.