Table 1 Final MEQ spatial extent

From: Deep learning forecasts the spatiotemporal evolution of fluid-induced microearthquakes

   

1 s model

15 s model

30 s model

Dataset

Pct

Obs [m]

Pred [m]

Diff [m]

σ [m]

Pred [m]

Diff [m]

σ [m]

Pred [m]

Diff [m]

σ [m]

Train

P50

5.92

6.29

0.38

0.67

6.59

0.68

0.20

2.56

3.35

0.79

 

P95

10.23

10.74

0.51

1.46

11.36

1.13

0.51

5.32

4.91

1.43

Val

P50

5.71

6.16

0.46

0.65

6.50

0.79

0.20

2.45

3.25

0.77

 

P95

14.08

15.01

0.93

2.04

16.02

1.94

0.71

7.61

6.47

1.99

Test

P50

9.92

10.25

0.34

1.11

10.79

0.87

0.34

4.29

5.62

1.32

 

P95

15.23

15.95

0.72

2.18

16.82

1.58

0.76

7.60

7.63

2.16

  1. For each split and percentile, the observed radius is followed by the neural-network (NN) prediction (Pred), absolute difference ∣Obs − Pred∣ (Diff), and model uncertainty (σ) for the 1-, 15-, and 30-s forecasting ranges.
Back to article page