The accurate determination of surface-layer turbulent fluxes over urban areas is critical to understanding urban boundary layer (UBL) evolution. In this study, a remote-sensing technique using a large aperture scintillometer (LAS) was investigated to estimate surface-layer turbulent fluxes over a highly heterogeneous urban area. The LAS system, with an optical path length of 2.1 km, was deployed in an urban area characterized by a complicated land-use mix (residential houses, water body, bare ground, etc.). The turbulent sensible heat (Q H) and momentum fluxes (τ) were estimated from the scintillation measurements obtained from the LAS system during the cold season. Three-dimensional LAS footprint modeling was introduced to identify the source areas (“footprint”) of the estimated turbulent fluxes. The analysis results showed that the LAS-derived turbulent fluxes for the highly heterogeneous urban area revealed reasonable temporal variation during daytime on clear days, in comparison to the land-surface process-resolving numerical modeling. A series of sensitivity tests indicated that the overall uncertainty in the LAS-derived daytime Q H was within 20%–30% in terms of the influence of input parameters and the nondimensional similarity function for the temperature structure function parameter, while the estimation errors in τ were less sensitive to the factors of influence, except aerodynamic roughness length. The 3D LAS footprint modeling characterized the source areas of the LAS-derived turbulent fluxes in the heterogeneous urban area, revealing that the representative spatial scales of the LAS system deployed with the 2.1 km optical path distance ranged from 0.2 to 2 km2 (a “micro-a scale”), depending on local meteorological conditions.