An Improved Water Vapor Trajectory Clustering Method and Its Application Analysis

In atmospheric water cycle research, water vapor trajectory analysis is a crucial tool for understanding the sources and transport pathways of precipitation water vapor. As a mainstream Lagrangian trajectory model, the HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) model provides water vapor trajectory data. However, its built-in trajectory clustering method has drawbacks, including long computation time and the loss of source point information. To address these issues, this study proposes an improved clustering method that incorporates a group-based computational optimization strategy and a weighted trajectory clustering approach to enhance computational efficiency and better capture dense water vapor source information. The study focuses on the cloud water resource high-value areas in Northwest China during the spring seasons from 2005 to 2015. Using the HYSPLIT model, backward water vapor trajectory tracking was conducted, followed by trajectory clustering analysis. The results demonstrate that the improved method reduces computational time costs, with experiments demonstrating an optimal reduction of up to 95.8%, while still preserving key source point information along water vapor transport pathways. Additionally, it enhances the identification of major water vapor transport routes. This improved method provides a more efficient and accurate data processing approach for large-scale water vapor trajectory analysis, offering valuable support for studying water vapor pathways in the atmospheric water cycle.