---

name: obspy description: | Seismology data processing with ObsPy. Helps with reading seismic waveforms, filtering/processing time series, fetching data from FDSN services, and earthquake analysis. Use when the agent needs to: (1) Read seismic data formats (MiniSEED, SAC, GSE2, SEGY), (2) Filter or process waveforms, (3) Fetch data from IRIS/USGS/FDSN services, (4) Search for earthquakes by magnitude/location, (5) Plot seismograms or spectrograms, (6) Remove instrument response, (7) Analyze station metadata. license: MIT metadata: version: 1.0.1 author: Geoscience Skills tags: '["Seismology", "Waveforms", "FDSN", "Earthquake", "Time Series", "ObsPy", "MiniSEED", "Signal Processing"]' dependencies: '["obspy>=1.4.0"]' complements: '["segyio", "disba"]' workflow_role: processing skill_type: domain

ObsPy - Seismology Data Processing

Quick Reference

from obspy import read, UTCDateTime
from obspy.clients.fdsn import Client

# Read local file (MiniSEED, SAC, etc.)
st = read("data.mseed")
tr = st[0]                          # First trace
print(tr.stats)                     # Metadata

# Fetch from FDSN
client = Client("IRIS")
t = UTCDateTime("2023-02-06T01:17:00")
st = client.get_waveforms("IU", "ANMO", "00", "LHZ", t, t + 3600)
st.plot()

Key Classes

Class	Purpose
Stream	Container for multiple Trace objects
Trace	Single waveform with data + metadata
UTCDateTime	Precise time handling
Inventory	Station/channel metadata
Catalog	Earthquake event information

Essential Operations

Read and Inspect

st = read("data.mseed")              # Auto-detect format
tr = st[0]
print(tr.stats.station, tr.stats.channel, tr.stats.sampling_rate)

Filter and Process

st.detrend("demean")                 # Remove mean
st.detrend("linear")                 # Remove trend
st.taper(max_percentage=0.05)        # Taper edges
st.filter("bandpass", freqmin=0.1, freqmax=10.0)

Fetch Waveforms from FDSN

client = Client("IRIS")
t1 = UTCDateTime("2023-02-06T01:17:00")
st = client.get_waveforms("IU", "ANMO", "00", "LHZ", t1, t1 + 3600)
st.write("output.mseed", format="MSEED")

Search Earthquakes

client = Client("USGS")
cat = client.get_events(
    starttime=UTCDateTime("2023-01-01"),
    endtime=UTCDateTime("2023-12-31"),
    minmagnitude=7.0
)
event = cat[0]
print(f"M{event.magnitudes[0].mag} at {event.origins[0].latitude}")

Get Station Metadata

inv = client.get_stations(
    network="IU", station="ANMO",
    level="response"                 # Required for response removal
)

Remove Instrument Response

st = client.get_waveforms("IU", "ANMO", "00", "LHZ", t, t + 3600)
inv = client.get_stations(network="IU", station="ANMO", level="response")
st.remove_response(inventory=inv, output="VEL")  # VEL, DISP, or ACC

Trim and Select

st.trim(UTCDateTime("2023-01-01"), UTCDateTime("2023-01-01T01:00:00"))
st_z = st.select(channel="*Z")       # Vertical only
st_bh = st.select(channel="BH*")     # BH channels

Merge and Handle Gaps

st.print_gaps()                      # Check for gaps
st.merge(method=1, fill_value="interpolate")

Writing Data

st.write("output.mseed", format="MSEED")
st.write("output.sac", format="SAC")

Error Handling

from obspy.clients.fdsn.header import FDSNNoDataException

try:
    st = client.get_waveforms("IU", "ANMO", "00", "LHZ", t1, t2)
except FDSNNoDataException:
    print("No data available")

Common Tips

1. Always detrend and taper before filtering to avoid artifacts
2. Use level="response" when fetching stations for instrument correction
3. Check for gaps before processing with st.print_gaps()
4. Wildcards work in queries: station="A*", channel="BH?"
5. UTCDateTime accepts ISO strings, timestamps, datetime objects

When to Use vs Alternatives

Tool	Best For
obspy	Full seismology workflows, FDSN data access, waveform processing
segyio	Fast, low-level SEG-Y file I/O for reflection seismic data
scipy.signal	Generic signal processing without seismology-specific features

Use obspy when you need seismological context: FDSN clients, instrument response removal, earthquake catalogs, or standard seismic formats.

Use segyio instead when working exclusively with SEG-Y files for reflection seismic. segyio is faster for large 3D volumes with inline/crossline access.

Use scipy.signal instead when you only need generic filtering or spectral analysis and don't need seismology-specific metadata or data access.

Common Workflows

Fetch and process earthquake waveforms

- [ ] Initialize FDSN client: `Client("IRIS")`
- [ ] Search events with `client.get_events()` for target magnitude/region
- [ ] Fetch waveforms with `client.get_waveforms()` for desired stations
- [ ] Get station metadata with `level="response"` for instrument correction
- [ ] Preprocess: detrend, taper, remove instrument response
- [ ] Filter to frequency band of interest
- [ ] Trim to analysis window and export or plot

References

• FDSN Data Centers - Available data centers and capabilities
• Channel Codes - Channel naming convention details
• Troubleshooting - Common issues and solutions

Scripts

• scripts/fetch_earthquake.py - Fetch waveforms for an earthquake
• scripts/batch_process.py - Batch process seismic files