Skip to content

Python tool for aircraft Doppler audio geolocation using historical ADS-B flight data (Source-available)

Align coherent harmonic overlap with the existing event contract

Cinna Daviscff7d1646932
diff --git a/docs/COHERENT_DISCOVERY.md b/docs/COHERENT_DISCOVERY.md
index d9d1405..6d366f9 100644
--- a/docs/COHERENT_DISCOVERY.md
+++ b/docs/COHERENT_DISCOVERY.md
@@ -21,9 +21,18 @@ Alternatively, add the reference checkout's `src` to `PYTHONPATH`. The script
processes the entire original recording with the reference template bank. It
captures actual measured ridge samples, skips unused straight-pass profile-range
calculations, and exports the existing detected-event schema. It reuses the
-reference's 40-point and 1% relative-fit-residual requirements without its hard
-aircraft/close type gate. Slow and nearby proposals remain eligible. Duplicate
-and harmonic ridges are grouped; their count is not an independent aircraft count.
+reference's 40-point and 1% relative-fit-residual requirements. It removes the
+later aircraft/close classification cutoffs at 30 m/s and 150 m. Duplicate and
+harmonic ridges are grouped; their count is not an independent aircraft count.
+
+The reference template bank and refinement limits remain unchanged. In
+[`_refine_all`](https://github.com/digitalis-p/doppler-locate/blob/ed84e0f/src/dopploc/detect.py),
+accepted fits require `6 < speed < 290` m/s and `35 < slant distance < 30000` m.
+The fitted center must stay within `max(10 seconds, half-window / 2)` of the
+template center, and the fit residual must not exceed `1.5 × harmonic order` Hz.
+Tracing and fitting can also fail. Broad downstream matching priors cannot
+recover a ridge rejected by these upstream limits. This experimental script
+does not remove every speed or distance restriction requested in issue #341.
These are acoustic proposals. They can include ground traffic, and short curves
can fit several unrelated archived flights. Neither a low residual nor ADS-B
@@ -44,6 +53,9 @@ and first two background-labeled clips per publisher location for development,
then the next two per class/location as a held-out set. No labels, flight IDs or
locations enter the detector. The complete case outcomes are in the
[24-row table](../benchmarks/coherent-proposals-341-20261104.csv).
+The table was committed at revision `69b11cf4`, before the supported-overlap correction.
+That correction can change harmonic grouping, but not accepted traces or
+whether a clip has proposals. The recorded rows remain unchanged.
| Partition and publisher label | Clips | Ordinary discovery | Coherent adapter | Union |
| --- | ---: | ---: | ---: | ---: |
diff --git a/src/dopplergeist/coherent_discovery.py b/src/dopplergeist/coherent_discovery.py
index 9862bbd..f3a408f 100644
--- a/src/dopplergeist/coherent_discovery.py
+++ b/src/dopplergeist/coherent_discovery.py
@@ -1,7 +1,8 @@
-"""Adapt measured coherent traces without aircraft-type or geometry rejection.
+"""Adapt already-refined coherent traces without an aircraft/close kind filter.
Experimental audio proposals only. Straight-pass fit quality does not identify
an aircraft; the historical-track matcher and decision stage remain necessary.
+The upstream detector's template and refinement limits still apply.
"""
import numpy as np
@@ -31,8 +32,8 @@ def events_from_coherent_traces(traces):
raise ValueError("Coherent traces require ordered finite measured ridge samples")
paths.append(points)
- # Reference traces have a 0.25 s hop. Use elapsed overlap rather than a
- # frame count tied to the ordinary discovery backend's 0.5 s hop.
+ # _overlap uses a 0.5 s comparison grid even for 0.25 s source traces.
+ # Match prepare_events: count supported samples, not gaps between endpoints.
groups = []
for path in sorted(_merge_duplicates(paths), key=lambda p: np.median(p[:, 1])):
for group in groups:
@@ -40,7 +41,7 @@ def events_from_coherent_traces(traces):
if shared is None:
continue
t, base, frequency = shared
- if np.ptp(t) < .6 * min(np.ptp(group[0][:, 0]), np.ptp(path[:, 0])):
+ if len(t) * .5 < .6 * min(np.ptp(group[0][:, 0]), np.ptp(path[:, 0])):
continue
ratio = frequency / base
harmonic = min(_HARMONIC_RATIOS, key=lambda h: abs(h - np.median(ratio)))