Skip to main content

How Detection Works

An overview of Grillo's earthquake detection system and how it identifies seismic events from your sensor network.

Detection overview

Grillo's detection system uses data from multiple sensors to:

  1. Identify potential earthquake signals
  2. Confirm events across multiple stations
  3. Calculate event parameters
  4. Generate alerts

The detection process

Step 1: Continuous monitoring

Each sensor in your network continuously streams data to Grillo Cloud:

  • Real-time accelerometer data
  • Typically 100 samples per second
  • Data arrives with minimal latency

Step 2: Signal detection

Algorithms analyze incoming data for earthquake signatures:

P-wave detection:

  • First seismic wave to arrive
  • Faster but weaker
  • Enables early warning

S-wave detection:

  • Second wave type
  • Slower but stronger
  • More damaging

P and S wave diagram

Step 3: Multi-station confirmation

A single sensor trigger isn't enough—confirmation requires multiple sensors:

Sensors triggeredAction
1Monitor, no event
2-3Possible event, continue monitoring
4+Likely event, begin characterization

This multi-station approach:

  • Reduces false positives
  • Improves location accuracy
  • Increases confidence

Step 4: Event characterization

Once confirmed, the system calculates:

Location (epicenter):

  • Triangulation from multiple stations
  • Uses arrival time differences
  • Accuracy depends on network geometry

Depth:

  • Estimated from arrival patterns
  • May be constrained for shallow networks

Magnitude:

  • Calculated from amplitude
  • Multiple magnitude types possible
  • Refined as more data arrives

Origin time:

  • When the earthquake occurred
  • Back-calculated from detections

Step 5: Alert generation

Based on event parameters:

  1. Event compared against alert thresholds
  2. Notifications sent to configured channels
  3. Event added to catalog

Detection parameters

Sensitivity

How sensitive the detection is to ground motion:

SettingEffect
HighMore detections, more false positives
MediumBalanced approach
LowFewer detections, fewer false positives

Trigger threshold

The minimum signal level to consider:

  • Lower = more sensitive
  • Higher = fewer triggers
  • Adjusted based on local noise

Confirmation requirements

How many sensors must trigger:

  • More sensors = higher confidence
  • Fewer sensors = faster detection
  • Trade-off between speed and accuracy

Detection latency

Time from earthquake to detection:

StageTypical time
Wave travel to first sensorVaries by distance
First sensor triggerUnder 1 second
Multi-station confirmation2-5 seconds
Event characterization5-15 seconds
Alert deliveryUnder 1 second

Total time: Typically 5-20 seconds from earthquake origin

Early warning time

The warning time before strong shaking:

  • Depends on distance from epicenter
  • P-wave arrives before damaging S-wave
  • Seconds to tens of seconds possible

What affects detection quality

Network factors

FactorImpact
Number of sensorsMore sensors = better detection
Sensor spacingAffects location accuracy
Geographic coverageLarger area = detect more events
Online percentageMore sensors online = better coverage

Data quality factors

FactorImpact
Installation qualityPoor mounting = noisy data
Site conditionsSoft soil can amplify noise
Local vibrationsTraffic, machinery interfere
Signal strengthWeak connectivity = gaps

Event factors

FactorImpact
MagnitudeLarger events detected more easily
DistanceCloser events detected faster
DepthVery deep events harder to locate

Limitations

What we can detect

  • Earthquakes within/near your network
  • Events producing ground motion above noise
  • Events triggering multiple sensors

What we cannot detect

  • Very small earthquakes (below network threshold)
  • Distant earthquakes (insufficient signal)
  • Events during network outages
  • Non-earthquake events (explosions, etc.) may trigger

False positives

Some non-earthquake events may trigger:

  • Large vehicle traffic
  • Industrial activity
  • Weather (wind, thunder)
  • Sensor malfunction

The multi-station confirmation reduces but doesn't eliminate false positives.

Improving detection

Optimize your network

  1. Add more sensors - Improves detection and accuracy
  2. Better distribution - Even spacing helps location
  3. Quality installations - Reduce noise
  4. Maintain sensors - Keep high online percentage

Fine-tune settings

  1. Adjust sensitivity - Based on false positive rate
  2. Set appropriate thresholds - Match your needs
  3. Review detections - Learn what triggers your network