USGS Earthquake (ComCat) Skill

Key parameters:

See references/api_reference.md for the full parameter catalog, GeoJSON response structure, and all network/magnitude type codes.

PNGE-Relevant Region Presets

These are the key regions for petroleum engineering induced seismicity research. Use these coordinates as starting points.

Oklahoma Induced Seismicity Zone

Center: 35.5N, 97.5W | Radius: 200 km

latitude=35.5&longitude=-97.5&maxradiuskm=200

Peak crisis: 2014-2016 (900+ M3+ events/year). Caused by massive Arbuckle Group saltwater disposal from Mississippian Lime and Woodford Shale plays. Oklahoma Corporation Commission Traffic Light Protocol reduced disposal volumes, seismicity declined but remains elevated above historic baseline.

Permian Basin / Delaware Basin (TX/NM)

Center: 31.8N, 103.5W | Radius: 150 km

latitude=31.8&longitude=-103.5&maxradiuskm=150

Currently the most active induced seismicity region in the U.S. Over 1,400 M2+ events in 2024 alone. Driven by record-high saltwater disposal volumes from Permian Basin horizontal wells. Texas Railroad Commission and NM OCD have implemented seismic response plans.

Appalachian Basin (WV/PA/OH)

Bounding box: 37.0-40.5N, 82.5-77.5W

minlatitude=37.0&maxlatitude=40.5&minlongitude=-82.5&maxlongitude=-77.5

Lower seismicity rate but critical for WV PNGE research. Marcellus/Utica disposal wells in Ohio have triggered felt events (e.g., Youngstown 2011-2012, Poland Township). WV/PA have fewer events but increasing monitoring.

Eagle Ford Shale (South TX)

Center: 28.5N, 98.5W | Radius: 150 km

latitude=28.5&longitude=-98.5&maxradiuskm=150

Raton Basin (CO/NM)

Center: 37.0N, 104.8W | Radius: 50 km

latitude=37.0&longitude=-104.8&maxradiuskm=50

Classic case study — coalbed methane wastewater injection caused Trinidad, CO earthquake swarm (2001-2011).

Workflow

Step 1 — Resolve Intent

Map the user's question to a query type:

Step 2 — Get Count First

Always query the /count endpoint before fetching data to gauge result size:

curl -s "https://earthquake.usgs.gov/fdsnws/event/1/count?starttime=2024-01-01&endtime=2024-12-31&minmagnitude=2.0&latitude=35.5&longitude=-97.5&maxradiuskm=200"

Returns a plain integer (e.g., 1750). If the count exceeds 500, consider:

• Narrowing the time window
• Increasing the minimum magnitude
• Reducing the search radius
• Paginating with limit and offset

Step 3 — Fetch Data

Build the query URL with appropriate filters. Use GeoJSON format for structured parsing.

Oklahoma M2.5+ events in 2024 (top 20 by magnitude):

curl -s "https://earthquake.usgs.gov/fdsnws/event/1/query?\
format=geojson&\
starttime=2024-01-01&\
endtime=2024-12-31&\
minmagnitude=2.5&\
latitude=35.5&\
longitude=-97.5&\
maxradiuskm=200&\
limit=20&\
orderby=magnitude" | jq '.features[] | {mag: .properties.mag, place: .properties.place, time: (.properties.time/1000 | todate), depth: .geometry.coordinates[2]}'

Permian Basin M2.0+ events in 2024:

curl -s "https://earthquake.usgs.gov/fdsnws/event/1/query?\
format=geojson&\
starttime=2024-01-01&\
endtime=2024-12-31&\
minmagnitude=2.0&\
latitude=31.8&\
longitude=-103.5&\
maxradiuskm=150&\
limit=20&\
orderby=magnitude" | jq '.features[] | {mag: .properties.mag, place: .properties.place, depth: .geometry.coordinates[2]}'

Appalachian Basin M1.0+ events (2020-2024):

curl -s "https://earthquake.usgs.gov/fdsnws/event/1/query?\
format=geojson&\
starttime=2020-01-01&\
endtime=2024-12-31&\
minmagnitude=1.0&\
minlatitude=37.0&\
maxlatitude=40.5&\
minlongitude=-82.5&\
maxlongitude=-77.5&\
limit=20&\
orderby=magnitude" | jq '.features[] | {mag: .properties.mag, place: .properties.place, depth: .geometry.coordinates[2]}'

Shallow events only (depth < 10 km — likely injection-induced):

curl -s "https://earthquake.usgs.gov/fdsnws/event/1/query?\
format=geojson&\
starttime=2024-01-01&\
endtime=2024-12-31&\
minmagnitude=3.0&\
latitude=35.5&\
longitude=-97.5&\
maxradiuskm=200&\
maxdepth=10&\
limit=20&\
orderby=magnitude"

CSV format (useful for spreadsheet analysis):

curl -s "https://earthquake.usgs.gov/fdsnws/event/1/query?\
format=csv&\
starttime=2024-01-01&\
endtime=2024-12-31&\
minmagnitude=2.5&\
latitude=35.5&\
longitude=-97.5&\
maxradiuskm=200&\
orderby=time"

Step 4 — Parse Response

GeoJSON response structure:

{
  "type": "FeatureCollection",
  "metadata": {
    "generated": 1772732955000,
    "title": "USGS Earthquakes",
    "status": 200,
    "api": "1.14.1",
    "limit": 20,
    "offset": 1,
    "count": 20
  },
  "features": [
    {
      "type": "Feature",
      "properties": {
        "mag": 5.06,
        "place": "8 km NW of Prague, Oklahoma",
        "time": 1706937868269,
        "felt": 26480,
        "cdi": 6.9,
        "mmi": 7.167,
        "alert": "green",
        "status": "reviewed",
        "sig": 1084,
        "net": "ok",
        "magType": "mww",
        "type": "earthquake",
        "title": "M 5.1 - 8 km NW of Prague, Oklahoma"
      },
      "geometry": {
        "type": "Point",
        "coordinates": [-96.764, 35.534, 3.0]
      },
      "id": "ok2024cish"
    }
  ]
}

Key fields for PNGE analysis:

Converting time: The time field is Unix milliseconds. In jq:

jq '.properties.time / 1000 | todate'

In Python:

from datetime import datetime, timezone
dt = datetime.fromtimestamp(time_ms / 1000, tz=timezone.utc)

Step 5 — Produce Output

Format: Data Table + Narrative Summary + PNGE Context

Present a markdown table of results (cap at ~20 rows), then a narrative summary with induced seismicity context.

Example output structure:

## Oklahoma Induced Seismicity — M2.5+ Events (2024)

| Date (UTC) | Mag | Depth (km) | Location | ID |
|------------|-----|------------|----------|----|
| 2024-02-03 | 5.1 | 3.0 | 8 km NW of Prague, OK | ok2024cish |
| 2024-01-13 | 4.3 | 6.7 | 6 km W of Arcadia, OK | ok2024awfn |
| 2024-01-13 | 4.1 | 6.6 | 6 km ENE of Edmond, OK | ok2024awva |
| ... | ... | ... | ... | ... |

**Summary:** 65 earthquakes of M2.5+ occurred within 200 km of Oklahoma City
in 2024, including a M5.1 event near Prague on Feb 3 — the largest Oklahoma
earthquake since the 2016 Pawnee M5.8. The Prague event had a shallow depth
of 3.0 km and 26,480 felt reports, consistent with injection-induced
seismicity in the Arbuckle Group disposal zone.

93% of events were shallower than 10 km, a key indicator of induced origin.
Magnitudes ranged from M2.5 to M5.1 with a mean of M3.0.

**Induced seismicity context:** Oklahoma seismicity peaked in 2015 with 900+
M3+ events/year, driven by Arbuckle Group saltwater disposal from
Mississippian Lime horizontal wells. Disposal volume reductions mandated by
the Oklahoma Corporation Commission have reduced event rates, but the region
remains seismically active above pre-2009 baseline levels.

*Cross-reference with `pnge:epa-enviro` (UIC Class II well locations) and
`pnge:eia-data` (disposal volumes) for injection-seismicity correlation.*

**Data quality:** Events are analyst-reviewed. USGS ComCat detection
threshold in Oklahoma is ~M1.0 due to dense OGS monitoring network. Data
is from the USGS Earthquake Hazards Program ComCat catalog.

Pagination

The API uses 1-based offset pagination with a maximum limit of 20000.

# First page
/query?format=geojson&limit=1000&offset=1&...

# Second page
/query?format=geojson&limit=1000&offset=1001&...

Always use the /count endpoint first to determine total results:

COUNT=$(curl -s "https://earthquake.usgs.gov/fdsnws/event/1/count?starttime=2024-01-01&endtime=2024-12-31&minmagnitude=2.0&latitude=35.5&longitude=-97.5&maxradiuskm=200")
echo "Total events: $COUNT"

If count exceeds 5000, warn the user and ask if they want a filtered subset or full paginated download.

Error Handling

Common mistakes:

• Combining circle (lat/lon/radius) and rectangle (minlat/maxlat/minlon/maxlon) parameters — use one or the other, not both
• Requesting more than 20000 events without pagination
• Using starttime after endtime
• Omitting format=geojson (default is quakeml XML)

Induced Seismicity Reference

Key Indicators of Injection-Induced Earthquakes

1. Shallow depth — Induced events are typically < 10 km deep (often 2-8 km), corresponding to injection depth in disposal formations (Arbuckle, Ellenburger, basal clastics)
2. Proximity to injection wells — Events cluster within 5-15 km of high-volume disposal wells
3. Temporal correlation — Seismicity rate increases following increased injection volume with a lag of weeks to months
4. Magnitude distribution — Induced sequences often show b-values near 1.0 and can produce M4-5+ events
5. Absence of pre-existing seismicity — Many induced zones had no historical earthquake record before injection began

Major U.S. Induced Seismicity Cases

Cross-References to Other Skills

• pnge:epa-enviro — UIC Class II injection well locations and permits. Query UIC_WELL table for disposal wells near earthquake clusters.
• pnge:eia-data — Disposal volume data. Route petroleum/move/imp/data/ and state-level production data can provide injection volume context.
• pnge:fracfocus — Hydraulic fracturing disclosures. Correlate completion dates with seismicity onset.
• pnge:wvges-wells — WV well locations. Cross-reference with Appalachian earthquake locations.
• pnge:boem-offshore — Offshore production data for Gulf Coast disposal volume context.

Caveats and Data Limitations

1. Induced vs. tectonic classification: The API does not reliably distinguish induced from tectonic events. The eventtype value "induced or triggered event" is rarely applied. Spatial proximity to injection wells and shallow depth are the primary indicators.

2. Detection completeness varies by region:

   • Oklahoma: complete to ~M1.0 (OGS dense network)
   • Permian Basin: complete to ~M1.5-2.0 (TexNet expanding)
   • Appalachian Basin: complete to ~M2.0-2.5
   • Rural/offshore: complete to ~M2.5-3.0

3. Magnitude revisions: Initial automatic magnitudes can shift 0.2-0.5 units after analyst review. Use reviewstatus=reviewed for research data.

4. Location accuracy: Horizontal error is typically 1-5 km for well-monitored regions (OK, TX), 5-20 km elsewhere. Depth is the least constrained parameter — fixed-depth solutions (e.g., 5.0 km) indicate the depth was assumed, not computed.

5. Temporal bias: Monitoring networks have expanded significantly since 2010. Pre-2010 catalogs undercount small events, making long-term trend comparisons unreliable below M3.0.

6. Reporting lag: Automatic locations appear within minutes; reviewed solutions may take hours to days. Very recent events may have preliminary magnitudes and locations.

Implementation Notes

• Prefer bash_tool with curl + jq for API calls. Use python3 (stdlib only: urllib, json) for complex analysis.
• Python client — see references/python_example.py for a full stdlib-only implementation with region presets, pagination, and formatted output.
• GeoJSON coordinates are [longitude, latitude, depth] — note the lon-before-lat order (GeoJSON standard, opposite of many mapping tools).
• Time field is Unix milliseconds, not seconds. Divide by 1000 for standard epoch conversion.
• Count before query — always hit /count first to gauge result size and avoid requesting unnecessarily large datasets.
• USGS updates: events appear within minutes (automatic), reviewed within hours to days. No scheduled bulk update cycle.