Table 2.

P-phase picking.

Model	μ	σ	Pr	Re	F1	MAE	MAPE	Training data	Training size	Ref.
EQTransformer	0.00	0.03	0.99	0.99	0.99	0.01	0.00	Global	1.2M	This study
PhaseNet	−0.02	0.08	0.96	0.96	0.96	0.07	0.01	North California	780K	8
GPD	0.03	0.10	0.81	0.80	0.81	0.08	0.01	South California	4.5M	10
PickNet	0.00	0.09	0.81	0.49	0.61	0.07	0.02	Japan	740K	2
PpkNet	−0.01	0.15	0.90	0.90	0.90	0.10	1.90	Japan	30K	5
Yews	0.07	0.13	0.54	0.72	0.61	0.09	0.02	Taiwan	1.4M	4
Kurtosis	−0.03	0.09	0.94	0.79	0.86	0.08	0.01	—	—	17
FilterPicker	−0.01	0.08	0.95	0.82	0.88	0.14	0.02	—	—	18
AIC	−0.04	0.09	0.92	0.83	0.87	0.09	0.01	—	—	19

μ and σ are mean and standard deviation of errors (ground truth—prediction) in seconds respectively. Pr, Re, and F1 are precision, recall, and F1-score respectively. MAEand MAPE are mean absolute error and mean absolute percent error respectively. Note Yews and PpkNet models used here are trained based on different datasets mentioned in the related work section.

Bold values represent the best performance.