RexySDR weekly RF report — 2026-W40

1/7 daily reports available; ISO week Monday–Sunday

Historical report for 2026-W40; this page is not a live observation. Values are relative and uncalibrated. Source limitations remain authoritative; complete capture coverage is not established.

Missing daily reports: 2026-09-28, 2026-09-29, 2026-09-30, 2026-10-01, 2026-10-02, 2026-10-03

Coverage and evidence quality

Within the one available day:

- P19 eight-band series [measured]: 215 points; 214 of 288 5-minute grid slots present (≈74% coverage); latest observation 06:01Z (id 17824, completeness 8/8, freshness 265 s at generation). Two gaps: 07:05–09:40Z (32 slots) and 23:15Z–02:40Z (42 slots). `raw_values_are_authoritative: true`; no display-normalized values were included. [derived] 288−214 = 74 missing slots; gap durations follow deterministically from the timestamps.

- P20 wideband [measured]: exactly **one** panorama snapshot (261004_060231, 06:02:31Z; 12,500 points, 5 kHz–62.5 MHz at 5 kHz spacing, 13.19 s scan, 12,431 valid bins after excluding 69 DC-hole bins; all file hashes verified; Power A and Power B valid-bin masks identical).

- FT8 [measured]: 870 recorded minutes, 750 active minutes, 6,372 declared decodes.

- Environmental feeds [measured]: both status OK; BOM data is Australian-regional context (not a local geomagnetic measurement); Open-Meteo values are model grid-cell context (cell ≈440 m from site, not co-located).

[interpretation] Evidence quality *within* the day is good: raw values authoritative, hashes verified, internal consistency explicit, and the daily report carries its own caveats. Evidence quality *as a week* is minimal: one anchor day, ~26% of the P19 grid unobserved, a single P20 snapshot, and no second day for any comparison other than the two calendar days contained inside the one rolling window.

Eight-band RF behaviour

Whole-window per-band statistics over the 215 supplied points [derived from measured points; dB rel]:

| Band | Min | Median | Max | Range |

|---|---|---|---|---|

| 160m | −80.02 (Oct4 03:40Z) | −55.86 | −54.15 | 25.9 dB |

| 80m | −80.66 (Oct4 03:55Z) | −63.18 | −61.95 | 18.7 dB |

| 40m | −73.10 (Oct4 03:00Z) | −63.05 | −56.70 | 16.4 dB |

| 30m | −79.24 (Oct4 05:20Z) | −75.04 | −65.21 | 14.0 dB |

| 20m | −80.10 (Oct3 23:05Z) | −77.87 | −67.54 | 12.6 dB |

| 17m | −76.57 (Oct4 04:25Z) | −59.55 | −49.99 | 26.6 dB |

| 15m | −84.25 (Oct4 05:25Z) | −63.47 | −57.83 | 26.4 dB |

| 10m | −91.53 (Oct4 02:55Z) | −87.51 | −79.67 | 11.9 dB |

Phase medians within the window [derived] show the day's arc: an Oct3 evening step-up (09:45–11:05Z, local 19:45–21:05), a plateau (11:30–19:15Z), a modest step-down from 19:20Z, then — after the 3h35m gap — Oct4 local-midday readings at or below every band's pre-gap phase median (e.g. 160m phase median −79.6 vs −55.0 on the plateau), with partial recovery on 160m/10m in the final samples.

[interpretation] The dominant organisation across bands in this window separates three groupings: low bands (160m/80m), mid bands (40m/30m/20m), and 17m/15m, which show the largest swings (26+ dB range). 10m shows the least absolute variation with isolated excursions. This is a description of one window only — `rf_classification_permitted: false` forbids labelling these states, and with no second day their repeatability is unknown.

Material uncertainty: two of the day's sharpest transitions (the 09:45Z step and the Oct4 02:45Z reading level) sit at or across the 07:05–09:40Z and 23:15Z–02:40Z gaps, so unobserved intervals are folded into those comparisons.

P20 wideband patterns

All P20 content comes from the **single** snapshot 261004_060231 [measured; `features_are_deterministic: true`]. A single observation supports no statement about change over time; "patterns" here means within-snapshot frequency-domain structure only. Power A and Power B are two parallel measurement channels with **undefined physical meaning**; their difference is not treated as diagnostic.

Global statistics over 12,431 valid bins (dB rel) [measured]:

| Statistic | Power A | Power B |

|---|---|---|

| Median | −65.90 | −69.26 |

| P90 / P99 | −51.07 / −40.87 | −60.78 / −48.27 |

| Peak | −21.79 @ 1.335 MHz | −32.09 @ 0.085 MHz |

| MAD | 7.34 | 3.64 |

| Bins > median+6 dB | 4,283 (34.5%) | 1,945 (15.6%) |

A−B difference channel [derived]: median +1.46 dB, P10 −5.31 dB, P90 +17.36 dB, MAD 4.25 dB.

Peak structure [measured]: in A, the top 8 of 12 listed peaks sit between 0.51 and 1.6 MHz, plus one bin at 17.615 MHz (−28.14 dB rel, +37.8 above the global A median). In B, top peaks sit at 85 kHz and across 0.17–0.565 MHz, plus four bins in the 15.32–15.71 MHz region.

FT8-neighbourhood windows (±100 kHz of configured centres; analysis windows, not band allocations) [measured/derived]: the 160m window is the only window with a bin exceeding median+6 dB (1 of 41, at 1765 kHz; A-window median −46.60 vs B −68.17, Δ +21.6 dB). A−B median Δ across windows ranges +3.7 dB (10m) to +21.7 dB (40m).

[interpretation, flagged as such] These are frequency locations only. The package contains no source identification and `no_rfi_classification: true` applies, so no peak is attributed to any source, service, or class (the daily report notes external world knowledge about 17.615 MHz standard-time services as unverified — retained here only as a reason to keep the observation non-attributed). The relationship between the P20 snapshot and the P19 series is one adjacent sample and cannot be generalised.

FT8 observational trends

No trend can be established — trends require at least two days and one is available. What follows is the single window's cross-band distribution.

[measured, contract summary] 870 recorded minutes, 750 active minutes, 6,372 declared decodes:

| Band | Decodes | Active min | Unique calls | Mean SNR dB | Min / Max SNR dB |

|---|---|---|---|---|---|

| 20m | 3,742 | 630 | 289 | −8.08 | −15 / +19 |

| 15m | 1,068 | 207 | 141 | −9.28 | −16 / +14 |

| 30m | 644 | 294 | 53 | −10.46 | −16 / +5 |

| 40m | 477 | 260 | 49 | −11.03 | −15 / +3 |

| 17m | 212 | 118 | 47 | −9.04 | −16 / +5 |

| 10m | 204 | 117 | 22 | −9.04 | −14 / +9 |

| 80m | 13 | 6 | 2 | +6.54 | −11 / +22 |

[derived] Band rows sum to 6,360 decodes against the declared 6,372 — a 12-decode discrepancy not resolvable from this package. Per-band active minutes sum to 1,632 against the 750 overall figure because a minute active on multiple bands is counted per band; the totals are not directly comparable. No 160m row appears in band activity.

Recent-decode descriptions (last 20, 05:59:45–06:01Z Oct 4) [measured metadata]: 20m includes JE1LES/JE1LET (PM95), YJ8RN (no grid), VK4BPK (no grid); 15m includes JR3REX/JA3KVT/JH3FHQ (PM74), BH3NZN (ON90), JF3SFK (PM75); 17m includes JF1RDH (PM95), JA8RHU (QN04); DT values 0.07–0.70 s.

Per contract rules these are decode metadata only: SNR values support **no inference about transmitter distance, path quality, proximity, or propagation mode**, and nothing in this section is a propagation claim.

Historical-reference maturity

[measured] The package reports 16 candidate reference cycles near the 06:01Z centre; 12 QUALIFIED (Sep 21–24, Sep 26–Oct 3), 4 REJECTED (two empty: Jul 23 2025 and Sep 19/25; Sep 20 failed centre/left-bin tests). However the package itself reports `maturity_status: UNQUALIFIED`, `classification_permitted: false`, `rf_state: INSUFFICIENT`, `anomaly_status: INSUFFICIENT`.

[measured, contract rules] The maturity rule states Q≥4 has not yet been qualified as sufficient, and D1 RF history is keyed by station_id and time, not by test-site coordinates — historical maturity is not location-stratified.

Consequence for this weekly report [derived from the rules]: **no baseline-relative statement is admissible anywhere above.** Every comparison is within-window, and the single day-to-day remark below is descriptive only, not a deviation-from-reference finding. The 12 qualified cycles exist as candidates but cannot be used as a baseline; whether they would cohere across a week is unknown from one bundle day.

Repeated or notable observations

**Nothing here is demonstrated to recur across days** — one daily report exists. Each item is a single occurrence within the 2026-10-03/04 window; recurrence is an open question, not an established pattern. All magnitudes [derived deterministically from measured points]; no classification labels applied (`anomaly_classification_permitted: false`).

1. **Oct3 09:45Z step up** (gap-spanning; previous sample 07:00Z): 30m +10.9, 17m +11.4, 15m +10.6, 40m +7.6, 20m +7.6, 80m +6.4, 10m +5.7 dB rel; 160m essentially unchanged (+0.4). Sustained ~80 min. The largest single-window multi-band event.

2. **Oct3 11:10Z partial reversal** in one adjacent step: 20m −9.1, 17m −8.8, 30m −7.6, 10m −7.5, 15m −6.9, 40m −4.9 dB; 80m/160m held higher levels; series then settled to the 11:30–19:15Z plateau.

3. **10m isolated one-sample lifts**: +6.5 dB (11:25Z), +4.5 dB (19:05Z, recovered next sample); +6.3 dB at 20:35Z sustained ~15 min. 10m's signature in this window is short excursions, not sustained movement.

4. **Oct3 19:20Z modest step down**: 15m −3.1, 17m −2.4, 20m −1.2 dB vs plateau; persisted to the 23:10Z gap boundary.

5. **160m single-sample dip** at 21:05Z (−6.7 dB, recovered).

6. **Oct4 02:45Z, first samples after the 3h35m gap**: every band at or below its pre-gap phase median — 160m −23.6, 80m −17.3, 15m −16.6, 17m −14.3, 40m −9.0, 10m −2.2, 30m −3.5, 20m ≈0 dB. Material uncertainty: everything between 23:10Z and 02:45Z is unobserved and cannot be excluded from this step.

7. **160m Oct4**: +11.0 dB two-sample spike at 04:00–04:05Z that decayed, then a sustained +11.1 dB step at 05:05Z holding ~−67.5 dB rel through 05:50Z; 17m lifted ~+2.6 dB from 05:25Z.

8. **Final three samples** (05:55Z–06:01Z): 10m +9.3 dB, 80m +4.9, 20m +4.2, 15m +1.9; 17m/40m essentially unchanged. Window too short to characterise — the freshest open item, and the first sample of the missing "next day" is exactly where observation stopped.

9. **Single-anchor day-to-day comparison within the window** [derived]: comparable local-afternoon hours (Oct3 16:05–17:00 local vs Oct4 15:05–15:50 local medians) — all eight bands lower on Oct4: 15m −15.3, 17m −11.8, 160m −11.1, 80m −10.6, 40m −7.2, 10m −4.0, 30m −2.2, 20m −1.9 dB. One-hour local offset, no qualified reference, and 160m partially recovered by 05:05Z; stated as measurement description only.

Cross-source observations

- **Temporal adjacency** [measured]: FT8's first recorded minute (09:44Z Oct 3) sits one 5-minute grid step before the eight-band step-up (09:45Z). This is co-occurrence in time only; FT8's contract role is observational evidence, and no causal or propagation statement is made.

- **Window close** [measured]: the 05:55–06:01Z lifts on 10m/20m/80m coincide with live FT8 decodes on 20m and 15m in the same minutes. Same caveat — coexistence, no causal claim.

- **P20 vs P19** [derived from measured]: at adjacent samples (06:01Z P19 160m −65.7 dB rel; 06:02:31Z P20 160m-window A median −46.6 dB rel), the two instruments differ by ~14 dB. Different bandwidths, aggregation, and uncalibrated references mean no reconciliation factor is derivable. Treated as an unquantified cross-instrument inconsistency with material uncertainty.

- **Space-weather context** [measured, contextual only]: BOM regional feeds — A=5 (valid Oct3 00:00), K=3 (valid Oct4 03:00), Dst reported as "20" with the sign convention undefined in the package (ambiguity flagged), no active alerts, one magnetic warning Oct3 01:47Z (coronal-hole cause, G1 forecast for Oct4). Oct4's lower readings fall inside the warning window; the contract makes this context, not proof, and no causal link is asserted.

- **Solar geometry** [derived time-mapping]: episode 1 occurred at local 19:45–21:05 (night); Oct4's low readings at local 12:45–15:50 (day). The gap-straddling Oct4 step (item 6) crosses a day/night boundary, so a diurnal explanation cannot be separated from other possibilities with a single anchor day.

- **Local meteorology** [measured, model grid-cell context]: 21.5 °C, 67% cloud, wind 104°/11.7 km/h with 36.4 km/h gusts. No association with any RF feature is evidenced.

Unresolved questions

- Which days are missing from the weekly cycle and why — collection outage, report-generation gap, or bundle assembly gap? Nothing in the supplied material says.

- Whether any single-day feature repeats: the evening step-up/reversal pair (items 1–2), the 17m/15m large-swing regime, 10m's isolated-lift behaviour, and the all-band Oct4 reduction are each n=1 events.

- Cause of the Oct3 09:45–11:10Z rise-and-partial-reversal: propagation change, local emitter activity, and receive-path behaviour are indistinguishable — no direction, mode, or per-bin observations exist for those minutes, and the cause of the adjacent gaps is also unexplained.

- Whether Oct4's all-band reduction is persistent, diurnal, or a gap/sampling artefact: the transition is inside the 23:15Z–02:40Z gap, observation resumes at local midday, and no admissible baseline exists (see Historical-reference maturity).

- P20: with one snapshot, the 0.5–1.6 MHz A-channel cluster, the 17.615 MHz A peak, the B peaks (85 kHz, 15.3–15.7 MHz), the 1765 kHz 160m-window bin, and the A−B difference structure are all unattributed and untracked. Whether A/B differ systematically or the difference is diagnostic is undefined by the contract.

- The 12-decode FT8 sum discrepancy (6,360 vs 6,372) is unresolvable from this package.

- The Dst sign convention ("20" — +20 or −20?) is undefined in the package.

- What happened in the final minutes after 06:01Z: the 10m +9.3 dB lift and the 160m sustained step were the freshest open items when observation ends.

Weekly summary

This is a **PARTIAL WEEK report built from 1 of 7 daily reports**, and no weekly conclusion — recurrence, trend, or periodicity — is derivable from a single anchor day. The one available day (2026-10-03/04, ≈74% P19 grid coverage, gaps of unknown cause) contained: an evening multi-band step-up and partial reversal; a plateau with 17m/15m showing the largest swings (26+ dB) and 10m showing isolated lifts; and an all-band reduction on Oct4 whose transition is hidden inside an overnight gap and straddles a day/night boundary. FT8 concentrated activity on 20m (3,742 decodes) and 15m (1,068) with no 160m activity, as decode metadata only. P20 contributes one static wideband snapshot, its strongest Power A structure at 0.5–1.6 MHz and one bin 6 dB above window median in the 160m FT8-neighbourhood window; Power A/B remain parallel channels of undefined physical meaning. Historical references (12 of 16 candidate cycles qualified) fall short of the contract's maturity rule, so no baseline-relative statement is admissible, and classification of any observation as anomaly, RFI, or RF state is forbidden and none was applied. Every notable item above is explicitly a single occurrence; evidence of recurrence, if any exists, would come from the six missing days. No configuration recommendations are made.