The stable boundary layer (SBL) over land exhibits pronounced vertical decoupling and structural heterogeneity that remain challenging to diagnose using routine operational monitoring. Here, we utilize a three-year continuous multi-level tower dataset ($N=\mathrm{40,630}$ quality-controlled 15-min records at 2, 10, 20, and 50 m) over the Horqin Grassland to characterize stability-dependent modifications of surface-layer wind structure and near-surface kinetic energy. Using the bulk Richardson number ($R{i}_b$) and raw thermal gradients ($\Delta T_v/\Delta z$), we identify a stability-driven reduction in vertical wind coupling, with a statistical change-point cluster centered near $R{i}_b\approx 0.30$ (95% CI: 0.09–0.52). Cross-level correlation analysis empirically localizes a structural transition zone to the 10–20 m interval, separating a faster, shear-driven upper layer from a dynamically suppressed near-surface flow under strong stability. Concurrently, near-surface horizontal kinetic energy ($HK{E}_{02}$) undergoes a 47% (nighttime) to 55% (full-record) median reduction, while horizontal wind-direction variability (${\sigma }_{\theta \_02}$) broadens due to low-wind meandering, and sub-hourly vertical velocity variance (${\sigma }_{w,\mathrm{LF}}^2$) is suppressed by 57–72% across all heights. Multidimensional scaling (MDS) reveals a statistically significant phase-space compaction under high stability (${\tilde{d}}_{\mathrm{high}}=2.13$ vs. ${\tilde{d}}_{\mathrm{low}}=2.67$, $p<0.001$), with adjusted odds ratios highlighting thermal gradient ($\mathrm{OR}=5.80$), wind shear ($\mathrm{OR}=1.52$), and directional variability ($\mathrm{OR}=1.34$) as dominant positive predictors. These empirical patterns remain robust under temporal-holdout validation (Year-1 calibration vs. Years 2–3 validation; consensus breakpoint $R{i}_b=0.161$ vs. $0.156$) and seasonal/wind-speed stratifications. The results demonstrate that routine 15-min multi-level tower networks provide valuable observational constraints on surface-layer structural transitions, informing boundary-layer parameterizations in numerical weather prediction models and near-surface dispersion assessments.