Identification and classification of kinematically effective early flight postures in ski jumping

This study aimed to characterise early flight posture and classify movement strategies in ski jumping of trunk, hip, and knee kinematics. Using a 10-camera markerless motion capture system, three-dimensional motion data were collected from 30 ski jumpers of different competitive levels. During the early flight phase, time-series waveforms of trunk anterior tilt and hip and knee extension angles were extracted and time-normalised. Principal component analysis (PCA) was applied to reduce waveform dimensionality while preserving dominant coordination features. The first two principal components from each variable were subjected to hierarchical clustering. Based on clustering indices, a three-cluster solution was statistically supported, revealing three distinct movement strategies. To evaluate cluster-specific kinematic differences over time, statistical parametric mapping (SPM) one-way ANOVA with post hoc tests was applied to the original waveforms. Significant main effects of cluster were observed for trunk, hip, and knee kinematics. Post hoc analyses showed that trunk and knee differences were mainly driven by contrasts between Clusters 2 and 3, whereas hip kinematics primarily distinguished Cluster 1. Together, these results indicate that PCA-based clustering supported by SPM enables an objective classification of early flight movement strategies and a descriptive interpretation of inter-individual kinematic differences.