Classification of earthquakes, explosions and mining-induced earthquakes based on XGBoost algorithm

The classification of low-magnitude tectonic earthquakes, explosions and mining-induced earthquakes is an important task in regional earthquake monitoring. Seismic events occurring at local and regional distances are classified primarily based on the characteristics of their waveform. We established...

Celý popis

Uloženo v:

Podrobná bibliografie
Vydáno v:	Computers & geosciences Ročník 170; s. 105242
Hlavní autoři:	Wang, Tingting, Bian, Yinju, Zhang, Yixiao, Hou, Xiaolin
Médium:	Journal Article
Jazyk:	angličtina
Vydáno:	Elsevier Ltd 01.01.2023
Témata:	data collection earthquakes Feature extraction geophysics Model performance index Seismic events identification spectral analysis support vector machines tectonics XGBoost algorithm Feature extraction XGBoost algorithm Seismic events identification Model performance index
ISSN:	0098-3004, 1873-7803
On-line přístup:	Získat plný text
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Popis
Shrnutí:	The classification of low-magnitude tectonic earthquakes, explosions and mining-induced earthquakes is an important task in regional earthquake monitoring. Seismic events occurring at local and regional distances are classified primarily based on the characteristics of their waveform. We established 36-dimensional and 201-dimensional datasets by using feature extraction and amplitude spectral analysis. The Extreme Gradient Boosting (XGBoost) supervised algorithm is introduced for the discrimination of couples-class and three-class. The accuracies in the earthquakes/explosions discrimination with feature extraction dataset and amplitude spectrum dataset are 97.48% and 95.12%, respectively, which shows that feature extraction can effectively quantify the differences between earthquakes and explosions. For the classification of earthquakes/mining-induced earthquakes and explosions/mining-induced earthquakes, the performance of XGBoost with the amplitude spectrum dataset is greater, with accuracies of 99.24% and 95.33%, respectively. In the classification of the three types of events, the accuracies of XGBoost are 96.41% for earthquakes, 90.38% for explosions, and 94.04% for mining-induced earthquakes. The performance indices of XGBoost for different input parameters are invariably greater than those of the support vector machine (SVM), with stable classification ability, suggesting that the XGBoost model has good prospects for application in seismic event classification. •Apply XGBoost Algorithm to study the classification of seismic events occurring at local and regional distances.•The input dataset is based on feature extraction and amplitude spectrum.•Compare the performance of XGBoost and SVM using confusion matrix.•XGBoost outperforms SVM in different performance indices (AUC, Accuracy, TPR, TNR).
Bibliografie:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23
ISSN:	0098-3004 1873-7803
DOI:	10.1016/j.cageo.2022.105242