DESCRIPTION:

This repository contains the source data for graphs and charts used in the paper "A machine learning-based estimator for real-time earthquake ground-shaking predictions in Southern California" submitted and accepted at "Communications Earth & Environment journal" 

Marisol Monterrubio-Velasco, Scott Callaghan, David Modesto, Jose Carlos Carrasco, Rosa M. Badi , Pablo Pallares, Fernando Vázquez-Novoa, Enrique S. Quintana-Ortı́, Marta Pienkowska, and Josep de la Puente

DATA FOR FIGURES: 

Figure 1 : 

The data used in this figure comes from the CyberShake Study 15.4.  Seismogram, intensity measure, and duration data from CyberShake Study 15.4 is available through the SCEC CyberShake Study 15.4 Globus Collection, served by the University of Southern California's Center for Advanced Research Computing.  Direct link:https://g-46eaba.a78b8.36fe.data.globus.org."

Figure 2:

Model evaluation on the validation dataset for T = 2s

1. Random Forest predictions using the optimized hyperparameters depth=30, n_estimators=30 for the validation dataset at T=2s

y_pred_dislib_T2s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10.dat

2. Artificial Neural Network predictions using the optimized hyperparameters 9layer and 256 neurons for the validation dataset at T=2s

Prediction_NN_BS_256_9capas_DataSet_AllRuptureVariations_OneRuptureID_Period2.0_ALL_Validation_log10.csv

3. True values for the validation dataset at T=2s

y_true_dislib_T2s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10.dat

Figure 3:

Error metrics obtained for each simulated scenario using:

- Artificial Neural Networks

Test_Results_metrics_paper_2.0_NN.csv, Test_Results_metrics_paper_3.0_NN.csv, Test_Results_metrics_paper_5.0_NN.csv, Test_Results_metrics_paper_10.0_NN.csv

- Random Forest regressor

Test_Results_metrics_paper_2.0_RF.csv, Test_Results_metrics_paper_3.0_RF.csv, Test_Results_metrics_paper_5.0_RF.csv, Test_Results_metrics_paper_10.0_RF.csv

- ASK14 GMPE

Results_metrics_paper_2.0_GMPE.csv, Results_metrics_paper_3.0_GMPE.csv, Results_metrics_paper_5.0_GMPE.csv, Results_metrics_paper_10.0_GMPE.csv

Figure 4:

MLESmap RotD50 predictions on a validation event of magnitude 6.85

RF_predictions_T2s_map_2748.csv,  ASK_14_prediction_T2s_map_2748.csv , ANN_predictions_T2s_map_2748.csv

RF_predictions_T3s_map_2748.csv,  ASK_14_prediction_T3s_map_2748.csv , ANN_predictions_T3s_map_2748.csv

RF_predictions_T5s_map_2748.csv,  ASK_14_prediction_T5s_map_2748.csv , ANN_predictions_T5s_map_2748.csv

RF_predictions_T10s_map_2748.csv,  ASK_14_prediction_T10s_map_2748.csv , ANN_predictions_T10s_map_2748.csv

Figure 5:

Spatial configuration of five historical earthquakes and BBP stations also including the coordinates of synthetic stations  from the CS_15_4 study

SyntheticStationsCoordinates_CS_15.4.csv, Whittier_BBP_sites.csv, Northridge_BBP_sites.csv, North_Palm_Springs_BBP_sites.csv, Landers_BBP_sites.csv, Hector_Mine_BBP_sites.csv

Figure 6:

RotD50 predictions for real events for the ‘inside’ stations 

NN_layers9North_Palm_Springs_IN_event_metrics-T_10.csv, NN_layers9Northridge_IN_event_metrics-T_10.csv, NN_layers9Landers_IN_event_metrics-T_10.csv, NN_layers9Hector_Mine_IN_event_metrics-T_10.csv, NN_layers9Whittier_IN_event_metrics-T_10.csv

NN_layers9North_Palm_Springs_IN_event_metrics-T_5.csv, NN_layers9Northridge_IN_event_metrics-T_5.csv, NN_layers9Landers_IN_event_metrics-T_5.csv, NN_layers9Hector_Mine_IN_event_metrics-T_5.csv, NN_layers9Whittier_IN_event_metrics-T_5.csv

NN_layers9North_Palm_Springs_IN_event_metrics-T_3.csv, NN_layers9Northridge_IN_event_metrics-T_3.csv, NN_layers9Landers_IN_event_metrics-T_3.csv, NN_layers9Hector_Mine_IN_event_metrics-T_3.csv, NN_layers9Whittier_IN_event_metrics-T_3.csv

NN_layers9North_Palm_Springs_IN_event_metrics-T_2.csv, NN_layers9Northridge_IN_event_metrics-T_2.csv, NN_layers9Landers_IN_event_metrics-T_2.csv, NN_layers9Hector_Mine_IN_event_metrics-T_2.csv, NN_layers9Whittier_IN_event_metrics-T_2.csv

Supplementary material:

Supplementary Fig 2

Boxplots comparing ML models and "true" values

-  DNN

Prediction_NN_BS_256_9capas_DataSet_AllRuptureVariations_OneRuptureID_Period2.0_ALL_Validation_log10.csv

Prediction_NN_BS_256_9capas_DataSet_AllRuptureVariations_OneRuptureID_Period3.0_ALL_Validation_log10.csv

Prediction_NN_BS_256_9capas_DataSet_AllRuptureVariations_OneRuptureID_Period5.0_ALL_Validation_log10.csv

Prediction_NN_BS_256_9capas_DataSet_AllRuptureVariations_OneRuptureID_Period10.0_ALL_Validation_log10.csv

- RF

y_pred_dislib_T2s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

y_pred_dislib_T3s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

y_pred_dislib_T5s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

y_pred_dislib_T10s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

- TRUE VALUES FROM CYBERSHAKE SIMULATIONS

y_true_dislib_T2s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

y_true_dislib_T3s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

y_true_dislib_T5s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

y_true_dislib_T10s_depth-30_n_estimators-30_val_MODEL_ORIGINAL_Dataset_ORIGINAL_ALL_log10_8Feat_Plus_RealData.dat

Supplementary Fig 3

Error metrics obtained for each simulated scenario:

Artificial Neural Networks:

Test_Results_metrics_paper_2.0_NN.csv, Test_Results_metrics_paper_3.0_NN.csv, Test_Results_metrics_paper_5.0_NN.csv, Test_Results_metrics_paper_10.0_NN.csv

 Random Forest regressor:

Test_Results_metrics_paper_2.0_RF.csv, Test_Results_metrics_paper_3.0_RF.csv, Test_Results_metrics_paper_5.0_RF.csv, Test_Results_metrics_paper_10.0_RF.csv

ASK14 GMPE

Results_metrics_paper_2.0_GMPE.csv, Results_metrics_paper_3.0_GMPE.csv, Results_metrics_paper_5.0_GMPE.csv, Results_metrics_paper_10.0_GMPE.csv

Supplementary Fig 4

Predictions for the Synthetic event of magnitude 7.45

RF_predictions_T2s_map_3.csv,  ASK_14_prediction_T2s_map_3.csv , ANN_predictions_T2s_map_3.csv

RF_predictions_T3s_map_3.csv,  ASK_14_prediction_T3s_map_3.csv , ANN_predictions_T3s_map_3.csv

RF_predictions_T5s_map_3.csv,  ASK_14_prediction_T5s_map_3.csv , ANN_predictions_T5s_map_3.csv

RF_predictions_T10s_map_3.csv,  ASK_14_prediction_T10s_map_3.csv , ANN_predictions_T10s_map_3.csv

Supplementary Fig 5

Predictions for the Synthetic event of magnitude 8.05

RF_predictions_T2s_map_1240.csv,  ASK_14_prediction_T2s_map_1240.csv , ANN_predictions_T2s_map_1240.csv

RF_predictions_T1240s_map_1240.csv,  ASK_14_prediction_T1240s_map_1240.csv , ANN_predictions_T1240s_map_1240.csv

RF_predictions_T5s_map_1240.csv,  ASK_14_prediction_T5s_map_1240.csv , ANN_predictions_T5s_map_1240.csv

RF_predictions_T10s_map_1240.csv,  ASK_14_prediction_T10s_map_1240.csv , ANN_predictions_T10s_map_1240.csv

Supplementary Fig 6

NN_layers9North_Palm_Springs_OUT_event_metrics-T_10.csv, NN_layers9Northridge_OUT_event_metrics-T_10.csv, NN_layers9Landers_OUT_event_metrics-T_10.csv, NN_layers9Hector_Mine_OUT_event_metrics-T_10.csv

NN_layers9North_Palm_Springs_OUT_event_metrics-T_5.csv, NN_layers9Northridge_OUT_event_metrics-T_5.csv, NN_layers9Landers_OUT_event_metrics-T_5.csv, NN_layers9Hector_Mine_OUT_event_metrics-T_5.csv

NN_layers9North_Palm_Springs_OUT_event_metrics-T_3.csv, NN_layers9Northridge_OUT_event_metrics-T_3.csv, NN_layers9Landers_OUT_event_metrics-T_3.csv, NN_layers9Hector_Mine_OUT_event_metrics-T_3.csv

NN_layers9North_Palm_Springs_OUT_event_metrics-T_2.csv, NN_layers9Northridge_OUT_event_metrics-T_2.csv, NN_layers9Landers_OUT_event_metrics-T_2.csv, NN_layers9Hector_Mine_OUT_event_metrics-T_2.csv

Supplementary Fig 7

ASK-14 GMPE's

df_InputDataPred_EQreal_Whittier_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Landers_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Northridge_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_North_Palm_Springs_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Hector_Mine_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

RF and DNN

Prediction_NN_BS_256_9capas_Northridge_3s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Landers_3s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Whittier_3s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_North_Palm_Springs_3s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Hector_Mine_3s_log10_Review_ALL.csv

Suplementary Fig 8

ASK-14 GMPE's

df_InputDataPred_EQreal_Whittier_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Landers_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Northridge_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_North_Palm_Springs_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Hector_Mine_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

RF and DNN

Prediction_NN_BS_256_9capas_Northridge_5s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Landers_5s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Whittier_5s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_North_Palm_Springs_5s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Hector_Mine_5s_log10_Review_ALL.csv

Suplementary Fig 9

ASK-14 GMPE's

df_InputDataPred_EQreal_Whittier_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Landers_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Northridge_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_North_Palm_Springs_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

df_InputDataPred_EQreal_Hector_Mine_plus_CS_sites_Scott_REVIEW_ASK_2014.csv

RF and DNN

Prediction_NN_BS_256_9capas_Northridge_10s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Landers_10s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Whittier_10s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_North_Palm_Springs_10s_log10_Review_ALL.csv

Prediction_NN_BS_256_9capas_Hector_Mine_10s_log10_Review_ALL.csv

Suplementary Fig 10

HyperParameters_T2s_log10.csv