Skip to main content
zenodoopen

ACCESS-AM2 Southern Ocean cloud and radiation data and code for SHAP analysis

<p>The ACCESS-AM2 (Australian Community Climate and Earth-System Simulator - Atmospheric Model Version 2) and SHAP analysis code and data used for the study described in Fiddes et al. (2024) '<em>A machine learning approach for evaluating Southern Ocean cloud-radiative biases over the Southern Ocean in a global atmosphere model</em>' accepted in Geoscientific Model Development</p> <p>Included files:&nbsp;</p> <p>- code.zip, inc:&nbsp;</p> <p>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - pre-process_modis.ipynb: process the modis data, described in Fiddes et al. 2022 (https://doi.org/10.5194/acp-22-14603-2022)<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - pre-process.ipynb: organises model and modis data for analysis. Produces the files: COSP_vars_MODIS_2015-2019.nc, COSP_vars_cg207_2015-2019.nc and COSP_vars_bx400_2015-2019.nc<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - run_XGBoost+SHAP_control.ipynb: run the XGBoost model and SHAP analysis for the control run (bx400). Produces the files: SHAP_values_SWCRE_2015-2019_bx4002.nc, XGBoost_predicted_SWCRE_2015-2019_bx4002.nc, SHAP_interactions_bx400.nc<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - run_XGBoost+SHAP_ice.ipynb: run the XGBoost model and SHAP analysis for the ice experiment run (cg207).&nbsp;Produces the files:&nbsp;SHAP_values_SWCRE_2015-2019_cg2072.nc,&nbsp;XGBoost_predicted_SWCRE_2015-2019_cg2072.nc<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - analysis+plots_ML.ipynb: plots and stats presented in paper&nbsp;</p> <p>- COSP_vars_MODIS_2015-2019.nc</p> <p>- COSP_vars_cg207_2015-2019.nc</p> <p>- COSP_vars_bx400_2015-2019.nc</p> <p>- SHAP_values_SWCRE_2015-2019_bx4002.nc</p> <p>- SHAP_values_SWCRE_2015-2019_cg2072.nc</p> <p>- XGBoost_predicted_SWCRE_2015-2019_cg2072.nc</p> <p>- XGBoost_predicted_SWCRE_2015-2019_bx4002.nc</p> <p>- SHAP_interaction_bx400.nc</p> <p>The cloud types&nbsp;used in this work can be found at&nbsp;https://doi.org/10.5281/zenodo.6004061&nbsp;</p>

ShareScore

44/100

Overall dataset sharing score

Score breakdown

These five areas show where the dataset supports — or may limit — practical reuse.

Stewardship
8
Harmonization
4
Access
20
Reuse readiness
8
Engagement
4

Topics