Local Level Lab

Airborne measurements. Open methods.

Isolating the remaining curvature improves every fixed probe

Keeping the known calibration terms and correcting only the remaining curvature improves all twelve generated probes.

Processing case Shared-source RMS mismatch After residual correction
Zero removal 9.07 m 0.64 m
Profiled removal 45.35 m 3.28 m

The largest remaining coordinate mismatches are about 4.0 m and 17.9 m. These numbers compare the same six saved epochs. No coefficients or calibration points were fitted to obtain the correction.

This identifies a more useful approximation than averaging the whole derivative. It does not yet provide a guaranteed prediction range. The derivatives at known generated probes are diagnostic information; the actual aircraft calibration remains unknown.

The next mathematical step is a uniform error bound for this residual correction. It must handle packet-clock boundaries rather than assume that derivatives are smooth everywhere. All parameter regions and unsupported cases must remain accounted for before new aircraft prediction fits.

Verified calculation · Next mathematical step · Current work