Advection fog remains difficult to simulate because microphysics, radiation, and turbulence are tightly coupled in the shallow fog layer. At the fog top, longwave cooling promotes condensation and liquid-water accumulation, which further strengthens radiative cooling and forms a positive feedback. Although two-moment schemes predict liquid-water mass and droplet number, aerosol activation is commonly driven by updrafts, whereas fog supersaturation often arises from radiative cooling. How radiative-cooling-induced activation affects droplet sedimentation and this fog-top feedback remains unclear. We incorporate radiative-cooling-induced activation into the Thompson aerosol-aware microphysics scheme in the Weather Research and Forecasting model and simulate a Yellow Sea advection-fog event. We compare the modified scheme with the original Thompson scheme and the one-moment Lin scheme. Lin produces excessive liquid water and an overly deep fog layer, whereas the original Thompson scheme removes liquid water too efficiently and underestimates liquid water path and fog depth. The modified scheme better reproduces visibility and liquid water path, reducing the liquid water path bias to −3.14 g m^−2 and the root-mean-square error to 20.28 g m^−2. Liquid-water budget analyses show that cooling-induced activation increases droplet number and reduces droplet size near the fog top, weakening gravitational sedimentation and sustaining the feedback. Sensitivity experiments confirm that sedimentation constrains this feedback and that its strength depends on background aerosol loading. These results suggest that radiative-cooling-induced activation regulates marine-fog liquid water by weakening size-dependent droplet sedimentation, highlighting the need to represent both processes in two-moment microphysics schemes.