The determination of high-quality mean dynamic topography (MDT) over island areas faces multiple challenges, due to the error in a global geopotential model (GGM) as well as the contaminated waveforms of satellite altimetry data and scarcity of surveyed gravimetric observations. We study the potential to enhance regional quasi-geoid/MDT over island areas by using recently released satellite altimetric gravity data and airborne gravimetric observations. Numerical experiments over the Paracel Islands in the northwestern South China Sea show that the combination of airborne gravity data enhances the local gravity field, and reduces the errors in the associated quasi-geoid, in comparison to the result directly derived from a GGM. The results demonstrate that the direct application of a GGM may cause significant oscillations in MDT determination over island areas, by a magnitude exceeding 6 cm. The incorporation of altimetric gravity data slightly reduces these errors; however, the bubble-like errors are prominent, suggesting that altimetry data alone may be insufficient for high-accuracy MDT recovery. In contrast, combining airborne gravity data dramatically reduces these oscillations, by a magnitude of several centimeters, indicating that airborne gravimetric surveys may be necessary for MDT determination with centimeter-level accuracy over coastal/island areas. Further comparison with independent ocean data shows that the geostrophic velocities computed by merging a high-degree GGM and local airborne gravity data can reconstruct real surface ocean circulation; and the GOCE-based (Gravity Field and Steady-State Ocean Circulation Explorer) GGMs may be preferable as the reference models in MDT determination.