Recent advances in satellite Earth observation have enabled global monitoring of soil moisture (SM) at fine to medium resolutions, but satellite remote sensing can only capture near-surface soil moisture (SSM). It is therefore essential to evaluate the potential of satellite-based SSM measurements for inferring deeper subsurface water variations. This study compares SSM variability from the Soil Moisture Active and Passive (SMAP) satellite and the Soil Water Index (SWI) derived from SMAP SSM with subsurface SM and groundwater (GW) dynamics simulated by a high-resolution fully integrated surface–groundwater model over an agriculturally dominated watershed in eastern Canada, at two spatial scales: the SMAP grid (9 km) and the watershed scale (∼4000 km²). SMAP measurements agree well with hydrologic simulations in terms of SSM variability at both scales. Simulated subsurface SM and GW storage display smoother, lagged responses compared to SMAP SSM, with optimal delays of ∼1 d for 25–50 cm SM, ∼6 d for 50–100 cm SM, and ∼11 d for GW storage at both scales. Modeled subsurface SM dynamics align well with SMAP-derived SWI using the classic characteristic time lengths (15 d for the 0–25 cm layer and 20 d for the 0–100 cm layer), while GW storage exhibits a slightly delayed variation relative to SWI. The quantified optimal characteristic time length (Topt) for SWI estimation, obtained by matching SMAP-derived SWI with modeled root zone SM, is consistent with values reported for other agricultural regions worldwide. This study demonstrates that SMAP soil moisture measurements can serve as a valuable tool for predicting root zone SM and GW dynamics and for validating fully integrated hydrologic models across spatial scales, while also offering insights into near-surface–subsurface water interactions and the capabilities of satellite-based SM monitoring.