To facilitate the quantification of the stratospheric aerosol radiative effect, this study generates a set of aerosol direct radiative effect (ADRE) kernels based on MERRA-2 reanalysis data. These radiative kernels measure the sensitivities of ADRE to perturbations in scattering and absorbing aerosol optical depth (AOD), respectively. Both broadband and band-by-band radiative kernels are developed to account for the wavelength dependency of ADRE. The broadband kernels are then emulated by a multivariate regression model, which predicts the kernel values from a handful of predictors, including the top-of-atmosphere (TOA) insolation, TOA reflectance, and stratospheric AOD. These kernels offer an efficient and versatile way to assess the ADRE of stratospheric aerosols. The ADREs of the 2022 Hunga volcano eruption and the 2020 Australia wildfire are estimated from the kernels and validated against radiative transfer model-calculated results. The Hunga eruption induced a global mean cooling forcing of -0.46 W/m² throughout 2022, while the Australia wildfire caused a warming forcing of +0.28 W/m² from January to August. The kernel estimation can capture over 90% of the ADRE variance with relative error within 10%, in these assessments. The results demonstrate the spectral dependencies of stratospheric ADRE and highlight the distinct radiative sensitivity of stratospheric aerosols, which differs significantly from that of tropospheric aerosols.