UQ Framework for E3SM Land Model (ELM) — Code to Reproduce Figures

This repository contains scripts and notebooks used to train Gaussian-process (GP) emulators, perform Bayesian calibration, and reproduce figures for the manuscript:

“A framework for parametric and predictive uncertainty quantification in the E3SM Land Model: Assessing site and observable generalizability”
(Journal of Advances in Modeling Earth Systems)

Repository Structure

• *.ipynb: Jupyter notebooks for emulator training, calibration, post-processing, and figure generation
• emulation_functions.jl: Julia helper functions used by selected workflows
• fig2_ENF_5sites.ncl: NCL script to generate the site-location/topography map

Note: Large intermediate artifacts (e.g., trained emulator binaries *.jld2 and large ensemble outputs *.nc) are not tracked in GitHub.

---

Figure-to-Script Map (Main Text & Supporting Information)

Maps / site overview figure

• fig2_ENF_5sites.ncl
  → Site locations + topography map; reproduce Fig. 2

Emulator training & evaluation

• ESM_emulator_train_evaluation_GPP_26param_fig3.ipynb
  → Train and evaluate emulator (26-parameter case); reproduce Fig. 3

• ESM_emulator_train_GPP_4par_US_Me2.ipynb
  → Example workflow for training a 4-parameter emulator (illustrated for US-Me2 and GPP); the same script can be adapted to other sites and observables.

Global sensitivity analysis (GSA)

• sobol_GSA_heatmap_fig4_5.ipynb
  → Sobol-based GSA post-processing; reproduce Figs. 4–5

Bayesian calibration / parameter estimation (PE)

• PE_3sites_4par_be_fig6_7_S2_S5.ipynb
  → Posterior estimation using synthetic (best-estimate) data and FLUXNET data across sites; reproduce Figs. 6–7 and SI Figs. S2 & S5

• PE_3sites_4par_fluxnet_fig9_S7.ipynb
  → Probabilistic prediction using calibrated parameters across sites; reproduce Fig. 9 and SI Fig. S7

• PE_4par_be_4qoi_Me2_1be_fig8_S3_S6.ipynb
  → Posterior estimation using synthetic (best-estimate) data and FLUXNET data across observables for specific site; reproduce Fig. 8 and SI Figs. S3 & S6

• PE_4par_be_4qoi_Me2_figS1.ipynb
  → Posterior estimation using synthetic (best-estimate) data across observables for specific site; reproduce SI Fig. S1

• PE_4par_fluxnet_4qoi_Me2_fig10_S4_S8.ipynb
  → Probabilistic prediction using calibrated parameters across observables for specific site; reproduce Fig. 10 and SI Figs. S4 & S8

---