Skip to content

Latest commit

 

History

History
66 lines (46 loc) · 5.43 KB

README.md

File metadata and controls

66 lines (46 loc) · 5.43 KB
Raw

DOI

GEORGIA: a Graph neural network based EmulatOR for Glacial Isostatic Adjustment

This repository contains the Python code base for Lin et al., 2023: "GEORGIA: a Graph neural network based EmulatOR for Glacial Isostatic Adjustment".

Paper in GRL: https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2023GL103672

Project abstract:

Glacial isostatic adjustment (GIA) modeling is not only useful for understanding past relative sea-level change but also for projecting future sea-level change due to ongoing land deformation. However, GIA model predictions are subject to a range of uncertainties, most notably due to uncertainty in the input ice history. An effective way to reduce this uncertainty is to perform data-model comparisons over a large ensemble of possible ice histories, but this is often impossible due to computational limitations. Here we address this problem by building a deep-learning-based GIA emulator that can mimic the behavior of a physics-based GIA model while being computationally cheap to evaluate. Assuming a single 1-D Earth rheology, our emulator shows 0.54 m mean absolute error on 150 out-of-sample testing data with <0.5 s emulation time. Using this emulator, two illustrative applications related to the calculation of barystatic sea level are provided for use by the sea-level community.

If you have any questions, comments, or feedback on this work or code, please contact Yucheng

Installation

Video Guide

Because GEORGIA contains some data files that are larger than 25 Mb, you should download it from Zenodo.

Once you installed GEORGIA and extract it, generate a virtual environmental for GEORGIA through conda:

conda create --name GEORGIA python=3.9
conda activate GEORGIA
cd Gerogia_GIA/
pip install -r requirements.txt
conda config --add channels conda-forge
conda install healpy

It should be noted that GEROGIA was written and tested with Python 3.9.7 and Jupyter Notebook 5.4.0.

GEORGIA API

We created a GEORGIA API that allows you to easily generate a North American Ice Sheet model of your own choice and calculate the corresponding global sea-level change.

API video tutorial: https://www.youtube.com/watch?v=J2rAPckk0yU&t=10s

Note the web based app might be broken due to large calculation amount, the more stable choice is installing GEORGIA on a local environment.

File Descriptions

  • GEORGIA_Tutorial.ipynb - A notebook contains the tutorial to use GEORGIA to emulate RSL. It includes information about loading and preparing data for GEROGIA, visulising spatial temporal emulation error (Figures 2, 3 and S1, S4, S5 in the paper). Two illustrative examples of using GEORGIA to investigate palaeo sea-level problems are also provided.
  • data/healpix16_coord.csv - A csv file contains coordinate for 16-degree Hierarchical Equal Area isoLatitude Pixelation (Healpix) of a sphere.
  • data/heal16_input_mean.npy - Mean ice history used to normalise input data ($\mu_{I}$).
  • data/heal16_input_std.npy - Standard deviation of ice history that is used to normalise input data ($\sigma_{I}$).
  • data/heal16_output_mean.npy - Mean relative sea-level change hisotry used to normalise output data ($\mu_{RSL}$).
  • data/heal16_output_std.npy - Standard deviation of relative sea-level change history that is used to normalise output data ($\sigma_{RSL}$).
  • data/healpix_test_input_norm.npy - 150-member normalised input data from the testing set.
  • data/healpix_test_output_norm.npy - 150-member normalised output data from the testing set.
  • data/ice_0_healpix16.npy - Modern ice thickness expressed in 16-degree Healpix.
  • data/modern_topo_healpix16.npy - Modern topography data expressed in 16-degree Healpix.
  • data/ice_model_input_norm.npy - Normalised synthetic ice history for emulation.
  • data/ice_model_output_norm.npy - Normalised RSL output history from physics-based GIA model for comparison.

Co-authors

  • Pippa Whitehouse
  • Andrew Valentine
  • Sarah Woodroffe

License

This project is licensed under the MIT License - see the LICENSE file for details

Acknowledgements

We thank Roger Creel and Lev Tarasov for their constructive comments that significantly improve this paper. The authors thank Glenn A. Milne for providing the code used to perform the GIA modelling, Parviz Ajourlou and Ryan Love for useful discussion. Y.L. was supported by a China Scholarship Council - Durham University joint scholarship. A.P.V acknowledges support from the Australian Research Council under grant DP200100053. This work made use of the facilities of the N8 Centre of Excellence in Computationally Intensive Research (N8 CIR) provided and funded by the N8 research partnership and EPSRC (Grant No. EP/T022167/1). The Centre is co-ordinated by the Universities of Durham, Manchester and York. The authors acknowledge the collaborative research opportunities created by PALSEA, a working group of the International Union for Quaternary Sciences (INQUA) and Past Global Changes (PAGES), which in turn received support from the Swiss Academy of Sciences and the Chinese Academy of Sciences.