Advancing tropical cyclone tracking by incorporating dynamical constraints on motion

Understanding the climatological characteristics of tropical cyclone (TC) activity is a topic of substantial scientific and socio-economic importance. As a key tool for this purpose, TC tracking algorithms detect cyclonic structures in a gridded dataset and link them together into a trajectory. However, many existing algorithms rely on simple geometric rules to connect TC positions, which do not account for the physics of TC motion and limit the reliability of the tracking results. To address this, we develop the Vorticity-based Dynamic Tracking (VDT) algorithm, whose key novelty lies in incorporating the dynamical constraints on TC motion, specifically, environmental steering flow and beta effect. By connecting successive TC positions relative to a dynamically predicted location, VDT improves the physical consistency of tracking results. Benchmarks against widely used algorithms also show that the VDT algorithm achieves strong performance, recording the highest Critical Success Index, the smallest track matching offset, and the greatest spatiotemporal coverage. Furthermore, the VDT algorithm most accurately reproduces key characteristics of observed TCs in both spatial distribution and temporal variability. Regional analyses further indicate that this improvement is evident near complex topography, where terrain-induced vorticity anomalies can mislead geometric linkage and produce spurious track deflections. These results demonstrate that the newly proposed algorithm overcomes a key shortcoming of previous methods by incorporating the physics of TC motion, thus providing a more reliable basis for analyzing historical TC activity and assessing future risks under climate change.