Benchmark id: inv2a

Scalar source-model metrics for the different solutions

SN Solution Mo (Nm) Mw Max. Slip (m) dz,dx Dimensions (km) Eff. Dimensions (km) Slip Centroid (km) Waveform misfits x 100
Width Length Width Length Xc Yc Zc 1-Norm 2-Norm RMSE %VR
  1   SIVdata 3.548e+19   7.00   3.63 0.49, 0.49 20.00   40.00   15.01   29.46   0.91   -7.34   7.75   0.0   0.0   0.0   100.0  
  2   asano 4.041e+19   7.04   3.87 2.00, 2.00 20.00   40.00   14.14   28.97   0.74   -6.75   8.76   140.4   4.2   0.8   95.3  
  3   cedrict 3.347e+19   6.98   5.62 0.80, 0.80 20.00   40.00   13.49   21.80   -0.39   -6.25   6.81   521.7   48.2   2.7   51.4  
  4   cedrict2 3.347e+19   6.98   8.09 0.80, 0.80 20.00   40.00   9.29   18.45   -1.19   -5.85   6.41   565.9   59.6   3.0   36.5  
  5   cedrict3 2.991e+19   6.95   5.19 0.80, 0.80 20.00   40.00   14.57   23.08   0.68   -7.54   8.10   497.4   46.4   2.7   49.6  
  6   gallovic001 4.145e+19   7.05   4.30 1.00, 1.33 20.00   40.00   14.58   29.19   1.27   -6.98   7.50   117.6   15.3   3.0   53.4  
  7   gallovic01 4.398e+19   7.06   3.37 1.00, 1.33 20.00   40.00   14.73   29.84   1.38   -6.98   7.50   116.7   14.9   3.0   52.9  
  8   gallovic1 4.277e+19   7.05   3.68 2.00, 2.00 20.00   40.00   14.20   29.29   1.35   -7.00   8.16   63.8   3.0   1.5   83.2  
  9   hoby 4.054e+19   7.04   4.13 4.00, 4.00 24.00   44.00   14.76   34.62   0.41   6.16   9.57   311.6   14.7   1.7   73.0  
  10   ids 4.064e+19   7.04   8.24 1.00, 1.00 24.00   61.00   16.40   34.07   0.72   -6.57   7.99   227.2   16.3   1.5   86.1  
The table lists standard parameters of the source. Mo refers to seismic moment. Moment magnitude Mw is computed as Mw = 2/3*(log10(Mo) - 9.05). The effective source dimensions (or Eff. Dimensions) is computed from autocorrelation width of the slip distribution following Mai and Beroza (2000). The slip centroid (Xc, Yc, Zc) is estimated as slip-weighted averaged coordinates (x, y, and z)

The misfit metrics given below are computed for each waveform data and averaged for all the components.
1-Norm (sum of average absoluture errors) = Σ|y i - f(xi)|
2-Norm (sum of squares of the errors) = Σ(y i - f(xi))2
RMSE (root mean square errors) = √ Σ(y i - f(xi))2 /n
Variance Reduction (VR) (scaled sum of the squares of the errors) = 1 - [ Σ(y i - f(xi))2 ] / [ Σ y 2]

Note : The waveform misfits are scaled up by multiplying with 100 to highlight differences in the small values. Additionally, the entries indicated with *, if the waveform data has not been provided.



source inversion validation database | version 1.0.0