Quantifying soil moisture (SM)-soil temperature (ST) coupling is crucial for advancing modeling of energy balance at the land surface and agricultural management optimization. While previous studies have investigated SM-ST relationships at discrete temporal scales (e.g., daily or weekly) driven by natural factors (e.g., topography, climate), multiscale temporal dynamics and anthropogenic drivers (e.g., land management) are poorly quantified. To address this, we applied the Ensemble Empirical Mode Decomposition to multi-year SM and ST time series from an irrigated field, isolating intrinsic mode fluctuations from minutes to years. The experimental design compared land cover (cropped vs. forested) and irrigation regimes (irrigated vs. non-irrigated). Results reveal strong scale-dependent SM-ST coupling, with peak correlations occurring at matched temporal scales (r = 0.67 vs. 0.26). Sub-weekly fluctuations exhibited weak positive associations (r < 0.3, p < 0.1), contrasting with strongly negative correlations at longer scales (r < -0.6, p < 0.01). Irrigation attenuated sub-weekly SM-ST coupling (r = 0.12 at irrigated sites vs. 0.28 at non-irrigated sites), while land cover mediated longer-term interactions through synergistic environmental effects (r = -0.37 in forests vs. 0.09 in cropland). Seasonally, inverse SM-ST coupling at weekly-to-monthly scales was stronger in summer than winter. These scale-specific mechanisms reconcile contradictory SM-ST patterns reported across studies, improving predictions of subsurface hydrothermal processes in managed ecosystems.