PADNet constrained machine learning projections of global land fAOD and cAOD driven by six CMIP6 GCMs from 2030 to 2050

This dataset provides PADNet constrained machine learning estimates of global land fine mode aerosol optical depth (fAOD) and coarse mode aerosol optical depth (cAOD) for 2007 to 2024 and projections from 2030 to 2100 under SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5. The products were generated within a consistent machine learning framework trained with PADNet derived size resolved aerosol observations and climate related predictors. The trained models were applied to historical and future climate variables from six CMIP6 global climate models to reconstruct fAOD and cAOD for 2007 to 2024 and project their evolution under the four future scenarios. The released products are six GCM multi model mean estimates designed to provide an observation constrained representation of aerosol size mode variability across the historical and future periods. The dataset is intended for research on future aerosol change, size resolved aerosol projections, compound fine and coarse aerosol extremes, aerosol climate interactions, land atmosphere coupling and climate related aerosol risk. By providing future fAOD and cAOD separately, the dataset enables analyses of aerosol modal partitioning and compound aerosol behavior that cannot be resolved from total aerosol optical depth alone.