PSK Live – Normalised Frequency Residuals (Experimental)

What this page measures
This page analyses the reported receive frequencies of decodes from the PSKReporter “Live” feed for M7SQI. Before the time-window statistics are calculated, the reported frequencies are normalised to reduce offsets caused by different receivers and by separate transmit-frequency populations.
Frequency normalisation
Reports are first separated by receiver, band and mode. Within each of those groups, distinct frequency populations are identified from gaps in the observed frequencies. Each report is then expressed relative to the median of its own frequency population.

This means that, for example, a receiver which normally reports one population several hundred hertz away from another receiver does not create an apparent several-hundred-hertz frequency shift simply because the reporting population changes. Separate FT8/FT4 transmit-frequency populations are treated in the same way rather than being mistaken for propagation-induced movement.

The current population separator uses gaps greater than 15 Hz. Populations are retained and re-referenced independently; they are not discarded merely because their raw frequencies are far apart.
Window statistics
The normalised reports are grouped into short time windows (default 15 s). For each window:
  • IQR (Hz) — interquartile range of the normalised frequency residuals.
  • Δ median (Hz) — median normalised residual for the window, centred on the selected period.
  • n/window — number of decodes contributing to that window.
How to read it
Low IQR indicates that the normalised reports agree closely within that window. A larger IQR indicates greater spread within the active receiver/frequency populations. Movement in Δ median shows a change in the centre of the normalised reporting distribution.

These residuals may contain propagation-related frequency behaviour, but they are not direct Doppler measurements. Receiver calibration, reporting resolution, transmit-frequency choice and changing propagation paths can all contribute to the remaining structure.
WSPR versus FT8/FT4
Fixed-frequency WSPR operation provides the cleanest case because the transmitted frequency is effectively held constant. FT8 and FT4 are a more movable feast: the transmitted audio position can change between transmissions. The population normalisation is therefore particularly important for FT8/FT4, and small residual excursions in those modes should be interpreted more cautiously.
Omnidirectional viewpoint
This is not a single fixed raypath. M7SQI illuminates many directions simultaneously, and the aether decides who hears it. The plotted statistics therefore represent a population-weighted view of the active paths at that time, rather than one chosen great-circle path.
Space-weather context
Kp, GOES X-ray flux and solar-wind magnetic-field data are shown as environmental context. Correlations and lag values are exploratory comparisons only. Agreement in timing may be interesting, but does not by itself demonstrate that the frequency residuals were caused by solar or geomagnetic activity.

File: psk_m7sqi_history.ndjson
Detected: JSON array of spot objects.
Kp overlay: 1264 shaded intervals
GOES overlay: 1268 long + 1268 short points (sat 18)
Wind overlay: Bt Bz By (points: 1262/1262/1262)
Residual reference: calendar-month receiver/source/band/mode baseline  |  Corr: n/a  |  points shown: 5091  |  RMS(median residual): n/a
TX timeline correction / diagnostic — active for FT8, FT4 and FT2

FT8 / FT4 / FT2 plot calibration: fixed calendar-month receiver/source/band median, minimum n=3 TX files loaded: 2026-08.ndjson, tx_live_rig1.ndjson
Relevant TX records: 12 ( Rig 1=12 )
PSK reports considered: 200
Matched: 196  |  Unmatched: 4  |  Ambiguous: 0
Plot correction status: corrected 149  |  awaiting TX 4  |  awaiting baseline 47  |  ambiguous 0
Most common PSK timestamp − TX timestamp offsets: 0 s × 114 | +15 s × 28 | +14 s × 26 | +12 s × 14 | +13 s × 8 | +11 s × 4 | +18 s × 1 | +16 s × 1
Median |reported frequency − known TX frequency|: 15.5 Hz

Correction pipeline activity

Hollow markers are reports not yet included in the corrected residuals: Awaiting TX means no matching logged TX cycle was available; Awaiting baseline means the TX was matched but that receiver/source/band/mode has not yet reached the minimum 3 reports needed for monthly calibration; Ambiguous means more than one TX cycle could plausibly match.

Normalised residual spread (IQR Hz) per window

Median normalised residual (Hz) relative to monthly calibration baseline

Privacy notice