GeoVeil API Documentation

Returns interference events for a station within a date range.

Parameter	Type		Description
station_code	string	req	Station code (e.g. SULI)
start_date	date	req	Start date (YYYY-MM-DD)
end_date	date	req	End date (YYYY-MM-DD)

[
  {
    "id": 12345,
    "stationid": "SULI",
    "datetime": "2026-05-26T14:23:00Z",
    "rfpath": "L1",
    "centfreq_mhz": 1575.42,
    "bandwidth_mhz": 10.0,
    "powerloss_dbm": -42.5,
    "hestpowspecdens_dbmhz": -95.3
  },
  ...
]

Returns the raw events from the last hour. Without parameters it returns all stations (potentially large — 100k+ events). With station_code it filters per station; limit caps the row count (recommended when opening a station modal). For aggregate views, prefer /lh/summary.

Parameter	Type		Description
station_code	string	opt	Filter per station (e.g. SULI)
limit	int	opt	Maximum rows returned (ORDER BY datetime DESC). Recommended: 3000 when station_code is set.

[
  {
    "stationid": "SULI",
    "datetime": "2026-05-26T12:14:00Z",
    "rfpath": "L1",
    "centfreq_mhz": 1575.42,
    "bandwidth_mhz": 8.0,
    "powerloss_dbm": -38.2,
    "hestpowspecdens_dbmhz": -92.1
  },
  ...
]

Returns one row per active station with pre-aggregated stats: event count, dominant band, minimum power loss, 6 ten-minute time buckets and the best spoofing analysis. Ideal for the Threats page — returns ~32 rows instead of 100k+ raw events.

None. Returns all stations active during the last hour.

[
  {
    "station_code": "SULI",
    "stationid": 64,
    "count": 4823,
    "dom_rfpath": "L1",
    "min_powerloss": -61.4,
    "buckets": [820, 910, 750, 890, 780, 673],
    "threat_type": "spoofing_candidate",
    "confidence": 0.87,
    "severity": "HIGH"
  },
  ...
]

Field	Description
station_code	Station code (e.g. SULI)
stationid	Internal numeric ID
count	Total events in the last hour
dom_rfpath	Dominant RF band (e.g. L1)
min_powerloss	Largest power loss (dBm, most negative)
buckets	Array of 6 values: event count per 10-minute window (chronological)
threat_type	Spoofing threat type (null if no analysis today)
confidence	Spoofing analysis confidence score (0–1)
severity	HIGH / MEDIUM / LOW / null

Returns the last-N-min interference snapshot for one station, with per-event PSD threat level + bandwidth shape, summary aggregates, sweep/chirp pattern detection over a wider context, and best-effort V1 enrichment of overlapping events.

Parameter	Type		Description
station_code	string	req	V2 station code (e.g. COST)
window	int	opt	Snapshot window in seconds, 1–900 (default 120)
context	int	opt	Wider lookback in seconds for sweep detection (window..3600, default 600)
include_v1	0/1	opt	Enrich with V1 events that overlap the window (default 1)

Level	PSD
noise_floor	≤ -130
info	-130 .. -120
noticeable	-120 .. -110
significant	-110 .. -103
high	-103 .. -98
critical	> -98

narrowband ≤ 5 MHz · chirp 5..20 MHz · wideband > 20 MHz

{
  "station": {"station_code": "COST", "station_name": "Constanta", "lat": 44.16, "lon": 28.66},
  "window":  {"from": "...", "to": "...", "seconds": 120, "context_seconds": 600},
  "summary": {
    "total_events": 47,
    "max_threat_level": "high",
    "by_threat_level": {"noise_floor": 5, "info": 10, "noticeable": 12, "significant": 14, "high": 6, "critical": 0},
    "by_shape": {"narrowband": 41, "chirp": 5, "wideband": 1},
    "by_band":  {"L1": 8, "L2": 28, "L5": 11},
    "peak_psd_dbmhz": -101.3,
    "peak_powerloss_dbm": -42.5
  },
  "pattern": {
    "sweep_detected": true,
    "events_in_window": 7,
    "distinct_minutes": 3,
    "cf_span_mhz": 18.24,
    "time_span_s": 120.0,
    "drift_mhz_per_s": 0.152
  },
  "events":     [ /* per-event with threat_level + shape */ ],
  "v1_active":  [ /* V1 events overlapping the window */ ],
  "v1_included": true
}

Detects groups of events with the same frequency and bandwidth across multiple stations within the specified interval (cross-station correlation).

Parameter	Type		Description
start_date	date	req	Start date
end_date	date	req	End date
min_stations	int	opt	Minimum stations (default: 2)

[
  {
    "centfreq_mhz": 1575.42,
    "bandwidth_mhz": 10.0,
    "station_count": 5,
    "stations": ["SULI", "BBUC", "DEVA", "TIMI", "ARAD"],
    "first_seen": "2026-05-26T12:00:00Z",
    "last_seen": "2026-05-26T14:30:00Z"
  },
  ...
]

Returns the spoofing analysis for a station and interval. The engine scores 7 indicators: temporal, spatial, multi-band, spectral power, cross-station correlation, frequency pattern and temporal drift.

Parameter	Type		Description
station_code	string	req	Station code
start_date	date	req	Start date
end_date	date	req	End date
limit	int	opt	Maximum rows returned (default 500, max 5000)

[
  {
    "stationid": "SULI",
    "datetime": "2026-05-26T14:23:00Z",
    "centfreq_mhz": 1575.42,
    "threat_type": "spoofing_candidate",
    "confidence": 0.87,
    "severity": "HIGH",
    "temporal_score": 0.9,
    "spatial_score": 0.75,
    "multiband_score": 0.8,
    "psd_score": 0.85,
    "correlation_score": 0.7,
    "freq_pattern_score": 0.9,
    "drift_score": 0.65,
    "analysis_notes": "Strong temporal clustering with spatial spread..."
  },
  ...
]

Value	Description
spoofing_candidate	Signal with spoofing characteristics (7 indicators active)
matched_spectrum	Spectrum matching a known spoofing source
meaconing	Delayed re-broadcast (meaconing)
clean	No threat detected

Aggregated statistics: top stations per constellation, frequency distribution, hour×day heatmap, BW/PL scatter, daily trend, weekday vs weekend.

Parameter	Type		Description
days	int	opt	Lookback days (default: 30)

Daily event counts + station×station co-occurrence matrix (stations active on the same day).

Generates a 24h interference forecast for a station and a specific hour. Returns probability, hourly forecast, per-RF-band predictions and model info.

Field	Type		Description
station_code	string	req	Station code (e.g. "SULI")
datetime	string	req	ISO 8601 (e.g. "2026-05-26T14:00")

POST /api/predict
Content-Type: application/json

{
  "station_code": "SULI",
  "datetime": "2026-05-26T14:00"
}

{
  "station_code": "SULI",
  "datetime": "2026-05-26T14:00:00",
  "probability": 73.0,
  // 0–100 scale (not 0–1)
  "expected_events": 45,
  "forecast_24h": [
    { "hour": 14, "probability": 73.0, "expected_events": 45 },
    { "hour": 15, "probability": 68.0, "expected_events": 38 },
    ...
  ],
  "band_predictions": [
    { "rfpath": "L1", "centfreq_mhz": 1575.42, "avg_bw": 8.2, "avg_pl": -42.1,
      "avg_psd": -94.3, "occurrences": 2841, "pct": 58.9 },
    { "rfpath": "L2", "centfreq_mhz": 1227.60, "avg_bw": 6.1, "avg_pl": -51.0,
      "avg_psd": -97.2, "occurrences": 1204, "pct": 24.9 },
    ...
  ],
  "model_info": {
    "model": "v1",
    "classifier_accuracy": 0.823,
    "classifier_auc": 0.909,
    "classifier_f1": 0.828,
    "training_samples": 549182,
    "trained_at": "2026-05-26T19:24:00Z"
  }
}

Current state of the LightGBM model: trained or training, accuracy, feature importance.

{
  "trained": true,
  "is_training": false,
  "accuracy": {
    "classifier_accuracy": 0.823,
    "classifier_auc": 0.909,
    "classifier_f1": 0.828,
    "regressor_mae": 127,
    "regressor_r2": 0.81
  },
  "training_samples": 549182,
  "stations_with_baseline": 83,
  "training_time": 480.3,
  // seconds
  "feature_importance": {
    "hour": 0.312,
    "station_avg_psd": 0.198,
    "dow": 0.154,
    "baseline_eph": 0.121,
    "hour_sin": 0.089,
    ...
  },
  "trained_at": "2026-05-26T19:24:00Z"
}

Side-by-side metrics for LightGBM and LSTM, with the winner per metric. If the LSTM is training, the is_training field is true and training_progress exposes the current step (e.g. "Epoch 32/60").

{
  "lightgbm": {
    "trained": true,
    "is_training": false,
    "accuracy": {
      "classifier_accuracy": 0.823,
      "classifier_auc": 0.909,
      "classifier_f1": 0.828,
      "regressor_mae": 127,
      "regressor_r2": 0.81
    }
  },
  "lstm": {
    "trained": true,
    "is_training": false,
    "training_progress": null,
    "accuracy": {
      "classifier_accuracy": 0.919,
      "classifier_auc": 0.969,
      "classifier_f1": 0.811,
      "regressor_mae": 18.69
    }
  },
  "winner": {
    "accuracy": "lightgbm",
    "auc": "lightgbm",
    "f1": "lightgbm",
    "mae": "lstm"
  }
}

Forces a retrain of the LightGBM model. The LSTM retraining is chained automatically after completion. Returns 409 if a training run is already in progress.

No body required.

{ "status": "retraining started" }

{ "error": "Training already in progress" }

Classic endpoint — returns events for a station with legacy field names. Used by the V1 polar chart.

Parameter	Type		Description
statie	string	req	Station name (e.g. Sulina)
start_date	date	req	Start date (YYYY-MM-DD)
end_date	date	req	End date (YYYY-MM-DD)

[
  {
    "DataOra": "2026-04-05T14:23:00",
    "DataOra_fin": "2026-04-05T14:45:00",
    "BandaRF": "L1",
    "Frecv_cent_MHz": 1575.42,
    "LunBanda_MHz": 10.0,
    "Pie_put_est_dBm": -42.5
  },
  ...
]

Aggregated per-RF-band statistics: occurrences, CF min/max, BW max, PL max, time range.

Parameter	Type		Description
statie	string	req	Station name
start_date	date	req	Start date
end_date	date	req	End date

Events with the same frequency detected simultaneously across multiple stations (cross-station correlation). Different implementation from V2 /ce (kept for historical reference).

Parameter	Type		Description
start_date	date	req	Start date
end_date	date	req	End date

VADASE (Variometric Approach for Displacement Analysis Stand-alone Engine) displacement data for detecting GNSS station positional changes. Available for 74 ROMPOS stations with active VADASE licences. Useful for correlating interferences with positional displacements.

Parameter	Type		Description
statie	string	req	Station name
start_date	date	req	Start date
end_date	date	req	End date

[
  {
    "DataOra": "2026-04-05T14:00:00",
    "East_mm": 2.3,
    "North_mm": -1.1,
    "Up_mm": 4.7
  },
  ...
]