Accurate monitoring of ozone (O3) concentrations by remote sensing is essential for achieving pollution control and ecological protection. However, the existing O3 remote sensing data with a low spatial resolution do not facilitate fine-grained studies of small-scale urban clusters. In this study, the multiscale geographically weighted regression kriging (MGWRK) method was used to spatially downscale O3 remote sensing products (10 km × 10 km). Downscaling factors were selected from meteorological factors and vegetation, aerosol optical thickness (AOD), and air pollutant emission inventory data. Spatial heterogeneity and scale differences among the factors were considered and compared via multiple regression kriging (MLRK) and geographically weighted regression kriging (GWRK) to generate 1-km annual and seasonal O3 remote sensing products. The results showed that I) the downscaling accuracy of each model can be expressed as MGWRK > GWRK > MLRK; the local downscaling model yields data that are more consistent with the actual spatial distribution of O3 after considering the spatial heterogeneity of the influencing factors; and the downscaled annual and seasonal data exhibit satisfactory spatial texture characteristics and consistency with the original spatial distribution of O3, while the distribution boundary problem of image elements is eliminated. II) Nitrogen oxide (NOx) and volatile organic compound emissions and temperature exhibit strong positive correlations with O3, while wind speed, humidity, the normalized difference vegetation index, and AOD indicate weak positive correlations with O3. Moreover, precipitation exhibits a weak negative correlation with O3. III) The coefficient of determination (R2) of the 1-km resolution annual O3 concentration data after downscaling based on the MGWRK model reaches 0.93, while the RRMSE and MAE values are only 3% and 1.86, respectively, with a coefficient of variation of 9.55%; the downscaling accuracy of the seasonal O3 concentration data is higher in summer and winter than during the other seasons, with R2 greater than 0.85, further confirming the spatial and temporal downscaling advantages of the MGWRK model for O3 in the Chang-Zhu-Tan city cluster. This further corroborates the feasibility of the MGWRK model for spatial and temporal O3 downscaling in the Chang-Zhu-Tan urban area.