Local Level Lab

Airborne measurements. Open methods.

Training-only forecast controls

3/24 fits passed the complete software checks. The procedure did not qualify.

3/24 saved fits passed the numerical stopping checks. The run charged 2,928 native calculations.

No aircraft observations were used. These are short generated tracks with known answers, not Earth-model detections or calibrated confidence intervals.

Small forecast errors do not qualify a failed stopping check.

Model Removal profiled Clock Start Stopping check Largest later axis RMS (mm)
Globe + field False header_elapsed 0 Unfinished 0.000646697
Globe + field False header_elapsed 1 Unfinished 0.00156194
Globe + field False nominal_200Hz 0 Unfinished 0.00359475
Globe + field False nominal_200Hz 1 Unfinished 0.000282559
Globe + field True header_elapsed 0 Unfinished 5.57074e-05
Globe + field True header_elapsed 1 Unfinished 0.000792647
Globe + field True nominal_200Hz 0 Unfinished 0.000237222
Globe + field True nominal_200Hz 1 Unfinished 0.000373246
Globe, zero field False header_elapsed 0 Unfinished 0.000120341
Globe, zero field False header_elapsed 1 Unfinished 3.84911e-05
Globe, zero field False nominal_200Hz 0 Unfinished 0.000138332
Globe, zero field False nominal_200Hz 1 Unfinished 0.000201485
Globe, zero field True header_elapsed 0 Unfinished 1.70241e-05
Globe, zero field True header_elapsed 1 Unfinished 7.84501e-05
Globe, zero field True nominal_200Hz 0 Unfinished 2.56132e-05
Globe, zero field True nominal_200Hz 1 Unfinished 7.13027e-05
Stationary disc False header_elapsed 0 Unfinished 1.62878e-05
Stationary disc False header_elapsed 1 Unfinished 3.65207e-06
Stationary disc False nominal_200Hz 0 Passed 9.44334e-06
Stationary disc False nominal_200Hz 1 Unfinished 3.90644e-06
Stationary disc True header_elapsed 0 Passed 1.50108e-06
Stationary disc True header_elapsed 1 Passed 2.48797e-06
Stationary disc True nominal_200Hz 0 Unfinished 0.000213731
Stationary disc True nominal_200Hz 1 Unfinished 0.000226493

The saved audit also compares both starts on each identical track. Forecast differences describe numerical sensitivity; they are not uncertainty bounds. Failures remain unchanged and this completed allowance cannot be restarted.

Frozen question and limits · Pilot inventory · Prediction protocol