Models and Data for Reproducing the Result in "A Geographically Weighted Gaussian Process Regression Emulator of the GCHP 13.0.0 Global Air Quality Model"

The frozen version of GW-GPR glolbal air quality emulator, and related data for publication. 

"Input" folder contains the emulator input for AerChemMIP and IGSM-GAINS-EPPA scenarios, processed AerChemMIP output,and data for calculating health impacts ("shared")

"models" folder contains the GP models at each gridcell for all versions the GW-GPR models (including the random subsampling) tested in the paper

"output" folder contains the data (csv and netcdf) used to create the plots in the paper, including those from GCHP-CAM, GW-GPR and other intermediate data.

"Plots" and "plot scripts" are self-explanatory

"prediction" folder contains the code for executing GW-GPR emulation

"prediction" folder contains the code for performing sensitivity analysis

"TAPS_input" folder contains the input for running TAPS , including the processed IGSM/GAINS/FAO data

"TAPS_output" folder contains the TAPS-generated emission data for each IGSM-GAINS-TAPS scenarios

"training" folder contains the code for training GW-GPR models, and the "data" subfolder contains the training data (annual mean PM2.5 change of 120 GCHP-CAM LHS-configured simulations and the corresponding emission fields). 

The raw GCHP-CAM output of the training and testing set are too large to be hosted. Instead, we provide the annual mean PM2.5 data from each GCHP-CAM simulations performed in our study (training set: training/data, testing set: output/)

The raw GCHP-CAM output, CAM forcing data are available via contacting the authors, to arrange the transfer of large data sets. The actively developed version of the emulator on GitHub.