Public
Full-Waveform Inversion (FWI) — Public Tier
(5 scenes)Full-access development tier with all data visible.
What you get
Measurements (y), ideal forward operator (H), spec ranges, ground truth (x_true), and true mismatch spec.
How to use
Load HDF5 → compare reconstruction vs x_true → check consistency → iterate.
What to submit
Reconstructed signals (x_hat) and corrected spec as HDF5.
Parameter Specifications
✓
True spec visible — use these exact values for Scenario III oracle reconstruction.
| Parameter | Spec Range | True Value | Unit |
|---|---|---|---|
| starting_velocity_model_error | -3.0 – 6.0 | 1.5 | - |
| source_wavelet_error | -2.0 – 4.0 | 1.0 | - |
| anelastic_attenuation_(q) | -100.0 – 200.0 | 50.0 | - |
| source_location_error | -20.0 – 40.0 | 10.0 | m |
InverseNet Baseline Scores
Method: CPU_baseline — Mismatch parameter: nominal
Scenario I (Ideal)
7.77 dB
SSIM 0.3772
Scenario II (Mismatch)
7.65 dB
SSIM 0.2288
Scenario III (Oracle)
16.34 dB
SSIM 0.4542
Per-scene breakdown (4 scenes)
| Scene | PSNR I | SSIM I | PSNR II | SSIM II | PSNR III | SSIM III |
|---|---|---|---|---|---|---|
| scene_00 | 7.57 | 0.3615 | 7.52 | 0.2249 | 16.28 | 0.4660 |
| scene_01 | 7.70 | 0.3833 | 7.81 | 0.2328 | 16.28 | 0.4522 |
| scene_02 | 7.97 | 0.3846 | 7.55 | 0.2285 | 16.43 | 0.4500 |
| scene_03 | 7.84 | 0.3794 | 7.72 | 0.2289 | 16.35 | 0.4487 |
Public Tier Leaderboard
| # | Method | Score | PSNR | SSIM | Consistency | Trust | Source |
|---|---|---|---|---|---|---|---|
| 1 | VelocityGAN + gradient | 0.739 | 29.39 | 0.907 | 0.91 | ✓ Certified | Zhang & Lin, JGR 2020 |
| 2 | InversionNet + gradient | 0.716 | 28.63 | 0.893 | 0.86 | ✓ Certified | Wu & Lin, JGR 2019 |
| 3 | TV-Reg FWI + gradient | 0.644 | 25.03 | 0.803 | 0.86 | ✓ Certified | Esser et al., Geophysics 2018 |
| 4 | L-BFGS FWI + gradient | 0.542 | 20.71 | 0.632 | 0.91 | ✓ Certified | Virieux & Operto, Geophysics 2009 |
Visible Data Fields
y
H_ideal
spec_ranges
x_true
true_spec
Dataset
Format: HDF5
Scenes: 5
Scoring Formula
0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)
PSNR: 40%
SSIM: 40%
Consistency: 20%