A dynamic spectrum contains time, frequency and intensity. It also contains
missing channels, terrestrial interference and calibration artefacts. We test
scan paths that preserve two dimensions instead of flattening everything into
one melody.
Frequency-by-time intensity matrix → mark unavailable cells → select a scan
direction → map frequency to pitch and intensity to loudness → keep flagged
noise audible or explicitly muted. Silence is also an editorial decision.
Students annotate which features are observations, which are suspected noise
and which cannot be classified from the data alone.
Observation guide 05
Interference
Use this page before the slides, then return to the checkpoint after the lab.
After this week, you can
Read time, frequency and intensity in a dynamic spectrum.
Keep interference flags distinct from missing and measured values.
Choose a scan path that does not collapse two dimensions without disclosure.
Inspectable teaching model
relative spectral power
0.120.140.170.840.820.190.220.18
A synthetic frequency scan with one narrow persistent spike. Values are a teaching model, not an astronomical observation.
Worked notebook
Retain the suspicious channel
A synthetic frequency scan with one narrow persistent spike.
Plot the channel values before filtering.
Mark the two high bins as suspected interference rather than confirmed solar emission.
Render flagged bins with a separable timbre while preserving their positions.
Compare the full and muted versions against the same identification task.
What the method supportsThe flagged cue preserves where the anomaly occurred without pretending its cause is known. Muting it creates a cleaner result but removes evidence of the editorial decision.
CheckpointWhat is the honest audible treatment when a bright feature may be terrestrial interference but cannot be classified from this dataset alone?Reveal the reasoning
Keep it present, flag it distinctly and label the cause unknown.
Deleting the feature would overstate certainty. Presenting it as solar would do the same. A separate cue lets the listener inspect the data while preserving the unresolved classification.