This dataset contains gridded estimates of carbon dioxide (CO2) and methane (CH4) fluxes at daily resolution covering the Southern Florida region from 2000 to 2024. Gridded CO2 and CH4 flux prototype products at 500-m spatial resolution were derived from a machine learning model based on eddy covariance (EC) measurements from 1) airborne fluxes collected seasonally with the NASA Carbon Airborne Flux Experiment (CARAFE) over the region during five flight deployments and 2) regional EC tower networks representing long term wetland ecosystem fluxes since 2004. Multiscale flux measurements were upscaled with remote sensing observations of MODIS optical reflectance using a bootstrap ensemble random forest modeling approach to predict daily mean flux intensity and uncertainty from February 2000 to August 2024. Prototypes of modeled, gridded greenhouse gas fluxes were developed as part of the NASA Carbon Monitoring System (CMS) BlueFlux Project. The data are provided in netCDF format.