Dataset and code for the publication: "Connecting the forward problem to the inverse problem in uncertainty quantification of Earth system models using fast emulators"

This Zenodo record contains the dataset and code to reproduce figures in the publication: "Connecting the forward problem to the inverse problem in uncertainty quantification of Earth system models using fast emulators." Gaussian process (GP) emulator–based global sensitivity analysis (GSA) and uncertainty quantification (UQ) is performed for the YSU PBL scheme and revised-MM5 scheme in the Weather Research and Forecasting (WRF) Model v4.6.1 (https://github.com/Howland-Lab/WRF-4.6.1.git). A high-level overview of the content is provided below. Details on content can be found in metadata.csv.

Contents:

data/Processed data used in the study:

posteriors/: Posterior distributions inferred from emulator-aided inverse UQ

sobol_indices/: Sobol’ sensitivity indices computed in emulator-aided GSA

wrf_outputs/: Model outputs (i.e., time-averaged vertical profile outputs from the WRF model)

notebooks/Jupyter notebooks used to generate figures:

plot_diagnostic_measures/: Visualizes nondimensional diagnostic measures of Sobol' sensitivity indices

plot_posteriors/: Visualizes marginal posterior distributions

plot_sobol_indices/: Visualizes Sobol' sensitivity indices

plot_wrf_profiles/: Visualizes outputs from the WRF model

scripts/Scripts for running emulator-aided GSA and inverse UQ:

gp_based_gsa.py: Workflow for GP-based GSA

gp_based_inverse_uq.py: Workflow for GP-based inverse UQ

utils/Plotting utilities:

plot_settings.py: Plotting configuration for consistent figure styling