RexySDR daily RF report — 2026-10-07

Rolling 24-hour evidence window; see source limitations

Historical report for 2026-10-07; this page is not a live observation. Values are relative and uncalibrated. Source limitations remain authoritative; complete capture coverage is not established.

Plain-English summary

- The receiver kept good records for the whole day: 289 of 289 scheduled readings across all eight shortwave amateur bands, plus one full 5 kHz–62.5 MHz spectrum "snapshot" taken at 6:02 am on Wed 7 Oct (local time).

- The most visible event: between roughly 4:25 pm and 9:00 pm on 6 Oct (local), five bands (80 m, 40 m, 30 m, 20 m, 17 m) simultaneously measured about 8–11 dB higher power than during that day's morning, and the 10 m band briefly rose around 6 pm. Levels came back down by ~9 pm. This is a measured observation — the evidence does not say what caused it.

- The 160 m band showed a strong daily swing, sitting noticeably higher from mid-afternoon through the following morning than on the previous mid-morning.

- Separate overnight comparison flags ("persistent level shift") exist on five bands; these are deterministic bookkeeping markers for the operator's screening list, not faults, and the built-in attention scores attached to them (highest: MEDIUM) are screening values only — not probabilities, not anomaly scores.

- Weather and space-weather feeds were healthy and mostly quiet; they are background context only and prove nothing about the radio measurements.

- Because the historical baseline at this station is still young (and not yet validated as sufficient), the report cannot yet say whether any of these levels are "normal" or "unusual" for this time of day, and no source, cause, or interference classification is permitted from this evidence.

Observation status

| Source | Status | Evidence |

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

| D1 8-band series (P19) | Complete | 289/289 points at 5-min spacing (derived: a gapless 24 h at this cadence); latest observation id 21312 at 2026-10-06T20:19:00Z, completeness 8/8 bands, 95 s fresh at contract generation. `raw_values_are_authoritative: true`; no display-normalized values included. |

| P20 panorama snapshot | Single, valid | observation 261006_200230, observed 2026-10-06T20:02:30Z (06:02:30 AEST), received 20:04:06Z; 12,500 points, 23 configured centres, scan 13.06 s, completeness 1; all file hashes verified; A and B channel valid-bin masks identical (12,431 bins, 69 DC-hole bins excluded). |

| Historical references (P20) | Available | previous (19:32:30Z), ~1 h earlier (19:02:30Z, within 40-min tolerance), previous-day same-time (261005_200230, within 90-min tolerance) all present. |

| Time-conditioned reference | Available | bucket 06–09 AEST, status DEVELOPING for every band metric. |

| Environmental context | Available | solar computed; BOM space weather OK (cache HIT, 7 s old); Open-Meteo OK (cache HIT). |

| FT8 | Available | 1,345 recorded minutes in the 1,440-minute window (coverage gap of unknown cause: 95 min); 16,485 declared decodes. |

Eight-band RF behaviour

Measured scalars over the 24 h, with window medians derived from the raw series (all dB, relative/uncalibrated):

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

| 160m | −65.6 | −82.1→−64.3 | −79.4 | −64.9 | −65.5 | −64.9 |

| 80m | −79.3 | −83.4→−71.9 | −81.7 | −73.4 | −79.1 | −80.6 |

| 40m | −71.1 | −74.0→−61.6 | −71.8 | −62.5 | −70.9 | −72.3 |

| 30m | −78.3 | −79.9→−66.7 | −78.4 | −67.8 | −78.5 | −79.3 |

| 20m | −78.6 | −80.7→−67.8 | −79.0 | −68.3 | −78.5 | −79.0 |

| 17m | −73.3 | −76.1→−63.2 | −73.1 | −64.5 | −73.9 | −75.7 |

| 15m | −82.2 | −85.8→−79.5 | −84.5 | −81.5 | −82.2 | −82.6 |

| 10m | −91.1 | −92.2→−82.9 | −91.5 | −89.6 | −91.0 | −91.1 |

Descriptive notes (measured, then derived):

- **Evening multi-band excursion (6 Oct):** from 16:25 to 20:55 local (06:25–10:55Z), 80 m/40 m/30 m/20 m/17 m each held ≥5 dB above their own local-midday medians (derived: 55 elevated samples per band, i.e. the same 5-min samples across bands). Peak hourly medians in that window were +8 to +11 dB above the morning baseline (30 m −66.9 at 18 h vs −78.4 baseline; 20 m −68.0; 40 m −62.2). Levels returned to baseline by ~21:00 local. 10 m showed a short ~+5 dB rise at 17:55–18:25 local; 15 m a mild ~+3 dB evening lift; the excursion did not touch all bands equally.

- **160 m daily swing:** lowest measured window was local 08:00–12:00 (≈ −81), rising from ~13:00 and staying near −65 through the evening, overnight and to the period end (24 h range 17.7 dB — the largest of any band).

- **20 m, 30 m, 15 m, 10 m** sat near their baselines overnight and into the 07 Oct sunrise end of period (end values within ~1 dB of the morning baseline on 20 m/30 m/15 m/10 m; 17 m ended ~1.5–2.5 dB lower).

- This report offers no cause for the evening excursion. The contract forbids anomaly/RFI/source classification, and no corroborating per-source spectral evidence aligned in time to the excursion is included in this package (the P20 snapshot was taken ~9 h after it ended).

P20 wideband evidence

Snapshot 261006_200230 (single 13.06 s full-span scan, 5 kHz–62.5 MHz, 5 kHz spacing, 12,431 valid bins). Power A and Power B are two parallel measurement channels — their physical meaning is not defined by this contract, and no ordering or "true channel" claim is made. All values relative/uncalibrated; deterministic features only.

Global (measured, via the features contract):

| | A | B |

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

| median | −70.75 | −71.58 |

| MAD | 3.49 | 2.59 |

| p90 / p99 above median | +15.9 / +27.8 | +9.7 / +21.8 |

| peak | −19.55 @ 9.480 MHz (+51.2 over median) | −26.35 @ 15.720 MHz (+45.2 over median) |

| active fraction (>median+6 dB) | 35.4 % (4,397 bins) | 18.4 % (2,289 bins) |

- Strongest A peaks (measured positions): 9.480 MHz; then a low-frequency cluster 0.705 / 0.970 / 1.170 / 1.205 / 1.235 / 1.265 / 1.310 / 1.420 MHz; 2.255 MHz; 6.100 / 6.175 MHz.

- Strongest B peaks: 15.720 MHz; 0.050 / 0.085 / 0.175 MHz; 14.820 / 14.905 / 14.970 / 15.040 MHz; 35.900 / 37.200 MHz.

- A−B (derived): median +0.49 dB, p10 −5.13, p90 +16.90, MAD 3.37 — i.e. A most often reports higher relative power, with the largest A-over-B gaps at some frequencies, but the distribution is wide.

- FT8-neighbourhood analysis windows (±100 kHz around centres — analysis windows, not band allocations): A-window medians range −47.8 (160 m, peak 1.850 MHz) to −71.0 (10 m); B-window medians −59.0 (17 m) to −74.3 (40 m). 160 m A and 80 m A each contain bins >6 dB over their own window medians (2 and 1 bins); the other six windows contain none. Within-window A-vs-B separation varies by band (derived): ~21.9 dB (160 m) and ~23.3 dB (40 m), but only ~2.3 dB (20 m) and ~3.2 dB (17 m).

- **Single-observation caution:** these are snapshot features. No statement about how any P20 spectral feature changed over time is possible from this one observation; time-change statements appear only in the next section, via deterministic comparisons to separate P20 observations.

Historical P20 evidence

`historical_p20` available. Basis: 101 archived P20 observations, maturity label BASIC_HISTORY_AVAILABLE. Cadence to the "previous" reference is 30.0 min with `valid_for_sequence: true` for every band, so sequence/persistence description is permitted by the contract. Classification permissions: anomaly = false, RFI = false, source identification = false; historical description and deterministic comparison = true.

Deterministic changes at the 06:02 AEST snapshot vs its three references (A/B window medians, dB):

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

| 160m | +0.00 / −0.32 | +0.42 / −0.10 | +0.38 / +0.14 |

| 80m | +0.30 / +0.30 | +0.61 / +0.12 | +0.41 / +0.49 |

| 40m | −0.43 / −0.32 | −0.48 / +0.02 | +0.44 / −0.35 |

| 30m | −1.43 / +0.18 | −1.35 / +0.04 | +0.22 / +0.13 |

| 20m | +0.22 / +0.29 | −0.70 / −0.46 | +0.79 / −0.85 |

| 17m | −1.83 / +1.49 | −0.98 / +2.11 | −1.93 / −2.16 |

| 15m | −0.77 / −0.82 | +0.14 / −0.06 | −1.68 / −1.83 |

| 10m | +0.03 / +0.25 | −0.49 / +0.51 | +0.72 / +0.78 |

Larger deterministic deltas worth noting: 160 m p99-above-median fell −7.96 dB (A) / −7.01 dB (B) vs the previous day's same-time observation; 80 m vs ~1 h earlier rose +1.72 dB (A) / +2.38 dB (B) — described as numbers only.

Active persistence sequences (contract flags, cadence valid — described as the deterministic bookkeeping it is):

- 40 m, 30 m: ≥6 dB step flagged from observation 261006_110230 (21:02 AEST 6 Oct), ongoing 19 steps / 540 min through the snapshot, trigger channel A.

- 20 m, 17 m: same trigger window, channels A+B.

- 10 m: trigger 261006_093230 (19:32 AEST), 22 steps / 630 min, channel B.

- 160 m: an earlier flagged step (trigger 261005_220230, channel A) is marked `step_no_longer_persistent`; 80 m and 15 m show no ≥6 dB step in the latest sequence.

- **Uncertainty:** the compact contract does not fully specify the baseline against which the ≥6 dB "step" is measured, so the exact magnitude of each persistent step cannot be quoted here.

Shadow screening values (operator screening only — explicitly **not** probabilities, anomaly scores, or RF classifications): highest band attention for the package is MEDIUM (source field), with per-band scores 30 m 21.19 MEDIUM, 17 m 28.38 MEDIUM, 20 m 19.79 LOW, 40 m 18.75 LOW, 10 m 17.10 LOW, 160 m 5.29 LOW, 15 m 5.97 LOW, 80 m 2.31 LOW. The MEDIUM bands are flagged `PERSISTENT_LEVEL_SHIFT`. Screening flags do not convert to anomaly claims under the interpretation rules.

Time-conditioned reference status

`time_conditioned_p20` available, and `same_observation_as_daily_p20: true` — the reference comparison was generated from observation 261006_200230, the same snapshot this report describes, so bucket comparisons below are like-for-like. Bucket: 06:00–09:00 local (Australia/Brisbane) — near-sunrise; station-level model, no seasonal or day-of-week conditioning.

**Status: DEVELOPING** for every band/channel metric (12–15 samples, 2 distinct local days across buckets; the current bucket has 12 samples over 2 local days). Per the maturity rules, DEVELOPING permits descriptive comparison but is immature; it does **not** enable anomaly or RFI classification, and robust-z or p10–p90 membership are deterministic comparisons only.

Window-median position vs the same-bucket historical distribution (derived from the reference; current − reference median, robust-z, inside/outside p10–p90):

| Band | A: Δ / z / range | B: Δ / z / range |

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

| 160m | +2.64 / 0.93 / outside | +1.32 / 0.68 / outside |

| 80m | +1.33 / 1.73 / outside | +0.50 / 4.12 / outside |

| 40m | +1.37 / 2.95 / outside | −0.42 / −0.96 / outside |

| 30m | −0.47 / −0.85 / inside | +0.11 / 0.88 / inside |

| 20m | +0.43 / 0.43 / inside | +0.15 / 0.09 / inside |

| 17m | −1.31 / −1.05 / outside | +4.43 / 1.69 / outside |

| 15m | −1.16 / −1.00 / inside | −0.93 / −1.18 / outside |

| 10m | +0.22 / 1.12 / inside | +0.83 / 3.45 / outside |

Also deterministic: 80 m A "p99 above median" is +2.32 dB over its bucket reference (robust-z 4.16, outside p10–p90), and its active-fraction (0.025) exceeds a bucket reference median of 0.0 (robust-z undefined — reference MAD is 0). 17 m A, 15 m B and 10 m A p99-above-median also sit outside their bucket ranges. None of this makes any value "unusual", "anomalous", or "RFI" — the two-day, 12-sample reference spans too little history for that, and the contract forbids it. With only two distinct local days, the distribution itself is provisional.

Environmental context

`environmental_context_p20` available and generated from observation 261006_200230 — the **same** observation as the daily P20 snapshot, so this context aligns directly with the 06:02:30 AEST panorama. (The contract's top-level D1 environment block is evaluated at the D1 series end, 20:19Z / 06:19 AEST — 17 min later — and is quoted only where noted.) All items are context; `causal_inference_permitted: false`, and association description must not be read as RF causation.

- **Solar geometry** (locally computed, NOAA-style analytical approximation): at 20:02:30Z the Sun was at 0.84° elevation, azimuth 94.9° (east) — essentially just above the horizon at sunrise for this site; declination −4.93°, equation of time +12.16 min; phase classified DAY by the deterministic criterion. At the D1 series end (20:19Z) elevation was 4.86°. Solar elevation/phase here describes time-of-day geometry only — it is not proof of any propagation mode or path behaviour.

- **Space weather (BOM, Australian-region context — not measured at the receiving antenna):** status OK; planetary A-index 14 (valid 2026-10-05 00:00, a daily regional aggregate), K-index 1.0 (valid 2026-10-06 18:00, analysed 19:59Z), Dst −25 nT (valid 19:45Z); zero magnetic alerts and warnings in force. These indices characterise the Australian region broadly; they are not site measurements.

- **Local meteorology (Open-Meteo model/grid cell — not a co-located physical weather station):** valid 20:00Z (150 s before the P20 observation): 15.2 °C, relative humidity 96 %, dew point 14.6 °C, apparent 16.1 °C, partly cloudy (51 % cloud), no precipitation, MSL pressure 1,018.6 hPa (961.5 hPa at site elevation), wind 4.4 km/h from 161° with 13 km/h gusts. Near-saturation, cool, calm conditions at dawn at the grid cell.

FT8 observational evidence

Observational only — FT8 decodes are not propagation and not causal evidence; no propagation claim, path quality, or distance inference is made from them (SNR alone supports none of these).

- Coverage: 1,345 recorded minutes within the 1,440-min period (95 min unrecorded; cause not stated by the contract); 1,274 minutes contained at least one decode; 16,485 declared decodes (first minute 2026-10-05T20:19Z, last 2026-10-06T20:19Z).

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

| 20m | 6,639 | 918 | 511 | −7.3 | −16→+17 |

| 15m | 4,160 | 811 | 433 | −9.2 | −16→+9 |

| 10m | 2,865 | 647 | 311 | −6.8 | −16→+20 |

| 40m | 1,495 | 503 | 192 | −9.7 | −15→+2 |

| 30m | 584 | 317 | 69 | −9.8 | −16→+4 |

| 17m | 492 | 271 | 90 | −9.9 | −15→0 |

| 80m | 248 | 134 | 26 | −8.8 | −15→+1 |

- 160 m has no FT8 activity row — reported as "none in this evidence set", not as "none possible".

- The 20 most recent decodes (period end, ~06:14–06:19 AEST) are dominated by 40 m, with callsign prefixes JA/DS (Japan/Korea registration blocks) and VK2/VK3/VK4 (Australian blocks) appearing; grids when present are east-Asian (PM) and Australian (QG/QF) Maidenhead fields. Decode counts are a receiver-and-schedule artefact; band ranking here reflects decoded traffic volume only.

- Mean SNRs sit between −6.8 and −9.9 dB across active bands; strongest single decode +20 dB (10 m), weakest −16 dB. Per contract rules, these SNRs support no distance, path, proximity, or propagation-mode inference.

Historical-reference status

Three reference layers exist, each with distinct maturity — and none currently permits classification:

1. **D1 eight-band same-time-of-day reference:** 18 historical cycles found, 12 qualified (Q=12), but overall maturity `UNQUALIFIED` — the contract rule states "Historical maturity Q>=4 has not yet been qualified as sufficient". `rf_state` and `anomaly_status` are both **INSUFFICIENT**. Six cycles were rejected for thin local sampling (reasons: minute_bins_lt_45 ×5, centre_missing ×5, left_bins_lt_20 ×5, right_bins_lt_20 ×6; qualified cycles span 2026-09-20 → 2026-10-05 at the ~20:19Z anchor). Additionally, history is keyed by station_id + time, not by test-site coordinates — it is **not location-stratified**. Conclusion: for the 8-band series, reliable same-time historical description cannot yet be provided.

2. **P20 archive history:** 101 observations, BASIC_HISTORY_AVAILABLE; previous / 1-hour / previous-day references all present. Deterministic description and comparison permitted; anomaly, RFI and source-identification classification all denied.

3. **Time-conditioned P20 bucket (06–09 local):** **DEVELOPING** — descriptive comparison permitted, but only 2 distinct local days back it; every numeric comparison above should be read accordingly. DESCRIPTIVE_REFERENCE/ESTABLISHED_REFERENCE have not been reached anywhere.

Notable measured changes

Separating: **[M]** measured, **[D]** deterministic calculation, **[I]** interpretation offered as description only.

1. [M] 6 Oct local 16:25–20:55: 80/40/30/20/17 m series each ≥5 dB above their own midday medians in the same 5-min samples (peaks up to ~+11 dB at hourly-median level); 10 m ~+5 dB for ~30 min near 18:00; baseline restored by ~21:00. Cause undetermined; no classification permitted.

2. [M] 160 m varied ~17.7 dB over 24 h (−82.1→−64.3), lowest ~08–12 local, elevated from mid-afternoon to period end — the largest band-level swing in the series.

3. [D] P20 persistence flags: ≥6 dB steps active across 540–630 min on 40/30/20/17 m (from ~21:02 AEST 6 Oct) and 10 m B (from ~19:32 AEST), trigger sequence cadence valid; 160 m's older flagged step is marked no-longer-persistent. Step baseline definition not fully specified in the compact contract (stated uncertainty).

4. [D] P20 vs previous-day same-time: generally small median moves (≤±0.9 dB) except 17 m (−1.9/−2.2 dB) and 15 m (−1.7/−1.8 dB); 160 m p99-above-median −8.0/−7.0 dB lower than a day earlier.

5. [D] Time-bucket comparisons (immature DEVELOPING reference): 17 m B window median +4.43 dB above its bucket reference median; 160 m A +2.64 dB; 80 m A p99-above-median +2.32 dB (robust-z 4.16); several other outside-p10–p90 cases tabled above. Deterministic positions, not anomaly verdicts.

6. [M] P20 snapshot peaks: A strongest at 9.480 MHz (+51.2 dB over A's global median, i.e. −19.5 dB rel); a dense low-frequency peak cluster (0.7–1.4 MHz, 2.255, 6.100/6.175 MHz). B strongest at 15.720 MHz with a 14.8–15.1 MHz cluster and 35.9/37.2 MHz peaks. These positions are single-snapshot facts; no identity or persistence claim made.

7. [D] A−B window medians differ strongly by band (~22–23 dB on 160/80/40 m vs ~2–3 dB on 20/17 m), while globally A−B median is only +0.49 dB.

8. [M] FT8: 20 m carried 40 % of the period's 16,485 decodes; no 160 m activity recorded; 95 min of the day lacks FT8 coverage records.

Cross-source observations

- **Same-evening overlap (adjacent but not aligned):** the D1 5-min multi-band rise ran 16:25–20:55 AEST; the P20 persistence sequences were triggered at 19:32 AEST (10 m B) and 21:02 AEST (A channels) and remained active through the 06:02 snapshot. These are different instruments, metrics and cadences (5-min band scalar vs 30-min panorama-window medians); temporal adjacency in the same evening is a deterministic coincidence of coverage, not established identity of an event, and the contract permits no causal or classification link.

- **Morning snapshot still elevated relative to bucket norms:** at the post-excursion snapshot, 40 m A, 80 m A and 17 m B window medians sit above their immature 06–09 bucket reference medians (with active persistence flags on 40/17 m) — while D1 series values for the same bands at the period end had returned to within ~1 dB of their own baselines. The two series measure different quantities on different scales, so this is a documented divergence, not a contradiction, and not an anomaly statement.

- **Wideband vs narrowband views:** P20's strongest A peak (9.480 MHz) lies outside all eight FT8 analysis windows, and the densest peak clusters (0.7–1.4 MHz; 14.8–15.1 MHz) lie outside the analysed amateur windows too — so the eight-band scalars are structurally blind to those features. Nothing in this package lets the evening D1 excursion be spectrally matched to any P20 peak (no P20 observation exists during the excursion in this evidence set).

- **FT8 vs band-level scalars:** FT8 decode volume ranked 20 m > 15 m > 10 m while D1 band medians ranked those same bands among the lower-measured and the evening excursion touched 40 m more strongly. Decode counts and relative power scalars measure different things (schedule/duty-cycle-sensitive vs energy in a band region); no reconciliation or inference is attempted.

- **Environment:** the excursion window spanned afternoon into ~2 h after local sunset (Sun at −16.98° site; sunset context not itself computed in this contract), and the snapshot sits at sunrise — recorded as time-of-day context only; solar, geomagnetic and model-weather values here demonstrate no effect on the RF observations.

What cannot yet be concluded

- **No anomaly or RFI classification of anything** — both permissions are false in every layer of this contract; the evening excursion, persistence flags and outside-range bucket positions are measured/calculated facts without verdicts.

- **No cause for the 6 Oct evening excursion** (propagation, atmospheric, man-made or instrumental are all untested against this evidence); no transmitter, source identity, distance, proximity, path or propagation-mode inference — explicitly barred from levels and from SNR alone.

- **No statement that any current level is normal or unusual for the time of day:** D1 historical reference is UNQUALIFIED/INSUFFICIENT; the time-conditioned bucket is DEVELOPING on 2 distinct local days; P20 history is only "BASIC_HISTORY_AVAILABLE".

- **No evolution claims for P20 spectral features:** the panorama evidence is a single snapshot (features valid at 06:02 AEST only); changes between P20 observations are deterministic 30-min-cadence comparisons, not continuous monitoring.

- **Power A and Power B remain undefined physical channels** — usable only as parallel measurements; their per-band separation pattern is arithmetic, not physical diagnosis.

- **Shadow attention scores (max MEDIUM) and the ≥6 dB "step" flags cannot be read as probabilities, severities, or anomaly scores**, and the step baseline definition is unstated.

- **Environment is context only:** BOM indices are regional, Open-Meteo is a model grid cell (not co-located), solar geometry is computed; none can establish, support or exclude an RF cause, and association here does not license association-learning claims.

- **95 minutes of FT8 coverage are missing** without stated reason; 160 m FT8 silence is absence of records, not absence of signals.

- **Location stratification:** RF history is keyed to station_id/time, not the test-site coordinates — cross-site comparability of history is unresolved.

- Finally: this report recommends no changes to RexySDR configuration.