Skip to content

Weather Event Example

Scenario: Cold Front Passing Through

Context

On November 20, 2025, a cold front moved through central Greece, causing temperatures to drop sharply across multiple stations.

Detection Run

python anomaly_detector.py \
  --end "2025-11-20 14:00:00" \
  --window 6 \
  --temporal-method arima \
  --spatial-verify

Console Output

═══════════════════════════════════════════════
 ANOMALY DETECTION REPORT
═══════════════════════════════════════════════
End Time: 2025-11-20 14:00:00
Window: 6 hours
Method: arima
Spatial Verification: Enabled

Total Stations: 14
Anomalous Stations: 4
Normal Stations: 10

Anomaly Breakdown:
  🔴 Device Failures: 0      <-- ✅ No hardware issues
  🌧️ Weather Events: 4       <-- 4 stations affected by cold front
  ⚠️ Suspected: 0

═══════════════════════════════════════════════
 DETAILED REPORTS
═══════════════════════════════════════════════

[ STATION: uth_volos (Volos - University) ]
  ⚠️  Temperature Anomaly:
      Method: arima
      Expected: 15.2°C | Actual: 10.1°C
      • 2025-11-20 14:00:00: 10.10°C -> 🌧️ Extreme Weather / Env Change
        └─ Diag: Trend Consistent (Corr: 0.89, 3 neighbors)

[ STATION: volos (Volos City) ]
  ⚠️  Temperature Anomaly:
      Method: arima
      Expected: 15.5°C | Actual: 10.3°C
      • 2025-11-20 14:00:00: 10.30°C -> 🌧️ Extreme Weather / Env Change
        └─ Diag: Trend Consistent (Corr: 0.92, 4 neighbors)

[ STATION: zagora (Zagora) ]
  ⚠️  Temperature Anomaly:
      Method: arima
      Expected: 12.8°C | Actual: 8.2°C
      • 2025-11-20 14:00:00: 8.20°C -> 🌧️ Extreme Weather / Env Change
        └─ Diag: Trend Consistent (Corr: 0.87, 2 neighbors)

[ STATION: larissa (Larissa) ]
  ⚠️  Temperature Anomaly:
      Method: arima
      Expected: 16.1°C | Actual: 11.4°C
      • 2025-11-20 14:00:00: 11.40°C -> 🌧️ Extreme Weather / Env Change
        └─ Diag: Trend Consistent (Corr: 0.91, 4 neighbors)

═══════════════════════════════════════════════
 NEIGHBOR COMPARISON - Station: uth_volos
═══════════════════════════════════════════════

Time                 | uth_volos | volos | zagora | larissa
---------------------|-----------|-------|--------|--------
2025-11-20 08:00:00  | 15.2      | 15.5  | 12.8   | 16.1
2025-11-20 08:30:00  | 15.1      | 15.4  | 12.7   | 16.0
2025-11-20 09:00:00  | 14.8      | 15.2  | 12.5   | 15.8
2025-11-20 09:30:00  | 14.4      | 14.9  | 12.1   | 15.4
2025-11-20 10:00:00  | 13.9      | 14.4  | 11.6   | 14.9
2025-11-20 10:30:00  | 13.2      | 13.7  | 10.9   | 14.2
2025-11-20 11:00:00  | 12.4      | 12.9  | 10.1   | 13.4
2025-11-20 11:30:00  | 11.6      | 12.1  | 9.4    | 12.7
2025-11-20 12:00:00  | 11.0      | 11.4  | 8.8    | 12.1
2025-11-20 12:30:00  | 10.5      | 10.9  | 8.5    | 11.7
2025-11-20 13:00:00  | 10.3      | 10.6  | 8.3    | 11.5
2025-11-20 13:30:00  | 10.2      | 10.4  | 8.2    | 11.4
2025-11-20 14:00:00  | 10.1 ⚠️   | 10.3  | 8.2    | 11.4

Observation: All stations show synchronized temperature drops
→ Classification: Weather Event (Cold Front)

Analysis

Why Was This Classified as Weather Event?

  1. Multiple Stations Affected: 4 out of 14 stations showed anomalies
  2. High Spatial Correlation: Average correlation = 0.90 (> 0.6 threshold)
  3. Synchronized Timing: All anomalies occurred at the same timestamp
  4. Similar Magnitude: All stations dropped 4-5°C from expected

Spatial Correlation Details

Station Pair         | Correlation | Distance
---------------------|-------------|----------
uth_volos ↔ volos    | 0.98        | 3.2 km
uth_volos ↔ zagora   | 0.87        | 28.5 km
uth_volos ↔ larissa  | 0.82        | 62.4 km
volos ↔ larissa      | 0.94        | 65.1 km

All pairs show high correlation → Indicates regional weather pattern

Time Series Visualization

Temperature Trend (6-hour window)

Temp (°C)
    17 ┤
    16 ┤●───╮
    15 ┤    ╰──╮
    14 ┤       ╰──╮
    13 ┤          ╰──╮
    12 ┤             ╰──╮
    11 ┤                ╰──╮
    10 ┤                   ●────●────● ← Anomaly detected
     9 ┤
       └─────────────────────────────
       08:00              14:00

Pattern: Smooth, consistent decline (typical of cold front passage)

What Would Happen Without Spatial Verification?

# Without --spatial-verify flag
python anomaly_detector.py \
  --end "2025-11-20 14:00:00" \
  --temporal-method arima

Result: All 4 stations would be flagged as potential failures!

Anomaly Breakdown:
  🔴 Device Failures: 4      <-- ❌ FALSE ALARMS!
  🌧️ Weather Events: 0
  ⚠️ Suspected: 0

Impact: Operations team would waste time investigating 4 "failures" that are actually normal weather.

Meteorological Context

Cold Front Characteristics

  • Front Speed: ~30 km/h
  • Temperature Drop: 5°C over 6 hours
  • Affected Radius: ~100 km
  • Duration: 8-12 hours

Station Positions Relative to Front

       N
       ↑
   zagora ●

larissa ● ← ← [COLD FRONT] → → → uth_volos ●
                                            volos ●

Front moved from west (larissa) to east (volos), causing sequential temperature drops.

Actionable Insights

For Operations Teams

✅ No action required - This is normal weather

For Meteorologists

  • Cold front passage confirmed by sensor network
  • Front speed: ~30 km/h
  • Can be used to validate weather models

For Researchers

  • Example of successful dual-verification
  • Spatial correlation accurately distinguished weather from failure
  • System prevented 4 false alarm notifications

Key Takeaways

  1. Spatial verification is critical - Reduced false positives from 4 to 0
  2. High correlation (> 0.6) = Weather event - Multiple stations show similar patterns
  3. Distance doesn't matter much - Stations 60km apart still show 0.82 correlation
  4. ARIMA detected the anomaly - But spatial verification classified it correctly