Segy Operations

Skill: SEG-Y Operations

Read, manipulate, and write SEG-Y seismic data using segyio library.

Purpose

Handle seismic data operations across all oil & gas pipeline skills. Provides code patterns for reading SEG-Y files, extracting headers, visualization, and writing output.

Reference Materials

Primary source: Equinor segyio-notebooks

Key notebooks:

• notebooks/basic/01_basic_tutorial.ipynb - Complete workflow
• notebooks/basic/02_segy_quicklook.ipynb - Header analysis
• notebooks/basic/03_basic_segy_editing.ipynb - Trace manipulation
• notebooks/pylops/01_seismic_inversion.ipynb - Inversion workflow

Library: segyio

Dependencies

pip install segyio numpy pandas matplotlib

Capabilities

1. Read SEG-Y Files

import segyio
import numpy as np

# Open file
with segyio.open('seismic.sgy', 'r') as segyfile:
    # Read as 3D cube
    data = segyio.tools.cube(segyfile)
    print(f"Shape: {data.shape}")  # (inlines, crosslines, samples)
    
    # Access inline
    inline_100 = segyfile.iline[100]
    
    # Access crossline
    crossline_200 = segyfile.xline[200]
    
    # Access specific trace
    trace_50 = segyfile.trace[50]

2. Extract Headers to DataFrame

import segyio
import pandas as pd

with segyio.open('seismic.sgy', 'r') as segyfile:
    # Binary header
    binary_header = segyfile.bin
    
    # Trace headers
    headers = []
    for i, trace_header in enumerate(segyfile.header):
        headers.append({
            'trace_number': i,
            'iline': trace_header[segyio.su.iline],
            'xline': trace_header[segyio.su.xline],
            'cdp_x': trace_header[segyio.su.cdp_x] / 10,  # usually scaled
            'cdp_y': trace_header[segyio.su.cdp_y] / 10,
            'samples': trace_header[segyio.su.ns],
        })
    
    df = pd.DataFrame(headers)
    print(df.head())

3. Visualize Seismic Data

import segyio
import numpy as np
import matplotlib.pyplot as plt

with segyio.open('seismic.sgy', 'r') as segyfile:
    data = segyio.tools.cube(segyfile)
    
    # Time slice
    time_slice = data[:, :, 100]  # sample index 100
    
    plt.figure(figsize=(12, 8))
    plt.imshow(time_slice.T, cmap='seismic', aspect='auto')
    plt.colorbar(label='Amplitude')
    plt.title('Time Slice at Sample 100')
    plt.xlabel('Crossline')
    plt.ylabel('Inline')
    plt.savefig('time_slice.png', dpi=150)

4. Manipulate Trace Length

import segyio
from scipy.signal import resample

# Resample or cut traces
with segyio.open('input.sgy', 'r') as src:
    # Cut to specific sample range
    data = segyio.tools.cube(src)
    cut_data = data[:, :, 100:500]  # samples 100-500
    
    # Or resample
    resampled = resample(data, 500, axis=2)  # resample to 500 samples

5. Write SEG-Y Files

import segyio
import numpy as np

# Create new SEG-Y from numpy array
with segyio.open('template.sgy', 'r') as src:
    # Use source as template for headers
    spec = segyio.spec()
    spec.format = src.format
    spec.sorting = src.sorting
    spec.iline = src.header.iline
    spec.xline = src.header.xline
    spec.samples = src.samples
    spec.ilines = src.ilines
    spec.xlines = src.xlines

# Write new file
data = np.random.randn(100, 120, 500)  # example data

with segyio.open('output.sgy', 'w', spec) as dst:
    dst.trace = data.reshape(-1, data.shape[-1])
    
    # Copy headers from source if needed
    with segyio.open('template.sgy', 'r') as src:
        dst.header = src.header

6. Create Fault Volume from Similarity

import segyio

with segyio.open('similarity.sgy', 'r') as segyfile:
    similarity = segyio.tools.cube(segyfile)
    
    # Convert similarity to discontinuity (fault indicator)
    # Low similarity = high discontinuity = likely fault
    discontinuity = 1 - similarity
    
    # Threshold to create fault mask
    fault_mask = (discontinuity > 0.6).astype(float)
    
    # Write fault volume
    # (use write pattern from above)

Workflow Integration

Used by:

• oil-gas-pipelines/exploration - Seismic interpretation
• oil-gas-pipelines/reservoir-production - Horizon picking, attribute analysis

Common Issues

File won't open

# Try strict=False for non-standard files
with segyio.open('file.sgy', 'r', strict=False) as segyfile:
    data = segyio.tools.cube(segyfile)

Wrong dimensions

# Check file structure
with segyio.open('file.sgy', 'r') as segyfile:
    print(f"Inlines: {segyfile.ilines}")
    print(f"Crosslines: {segyfile.xlines}")
    print(f"Samples: {len(segyfile.samples)}")
    print(f"Sorting: {segyfile.sorting}")

Software Tasks

If user requests:

• Seismic visualization web app → invoke petropowers:oil-gas-delegation
• SEG-Y database system → invoke petropowers:oil-gas-delegation
• Automated SEG-Y processing pipeline → invoke petropowers:oil-gas-delegation

Quick Reference

Example Session

User: "What's the amplitude range in this SEG-Y file?"

Agent:

import segyio
import numpy as np

with segyio.open('seismic.sgy', 'r') as segyfile:
    data = segyio.tools.cube(segyfile)
    print(f"Amplitude range: {data.min():.2f} to {data.max():.2f}")
    print(f"Mean: {data.mean():.2f}, Std: {data.std():.2f}")

Output:

Amplitude range: -32768.00 to 32767.00
Mean: 12.34, Std: 4567.89

OSDU Integration

SEG-Y to VDS/ZGY Conversion

OSDU (Open Subsurface Data Universe) uses cloud-native formats for seismic data:

• VDS (Voxel Data Service) - Cloud-optimized seismic format for random access
• ZGY - Compressed seismic format for visualization

Conversion Workflow:

1. Upload SEG-Y file to OSDU (via Dataset Service)
2. Trigger SegYToVDS DAG conversion workflow
3. VDS file stored in cloud storage
4. Access via Seismic DMS (Data Management Service)

OSDU Schemas for Seismic

OSDU Dataset Service API

import requests

# Get storage instructions
storage_resp = requests.post(
    "https://api.osdu.com/api/dataset/v1/storageInstructions",
    headers={
        "Authorization": "Bearer <token>",
        "data-partition-id": "mypartition"
    },
    json={
        "kind": "osdu:wks:dataset--File.Seismic:1.0.0",
        "dataContentType": "application/x-segy"
    }
)

# Upload file to provided URL
upload_url = storage_resp.json()["storageLocation"]["signedUrl"]
with open("seismic.sgy", "rb") as f:
    requests.put(upload_url, data=f.read())

# Register dataset
register_resp = requests.post(
    "https://api.osdu.com/api/dataset/v1/register",
    headers={
        "Authorization": "Bearer <token>",
        "data-partition-id": "mypartition"
    },
    json={
        "id": "dataset--File.Seismic:1654321",
        "kind": "osdu:wks:dataset--File.Seismic:1.0.0",
        "ancestry": {
            "parents": ["work-product--SeismicTraceData:12345"]
        }
    }
)

Create OSDU Manifest for Seismic

import json
from datetime import datetime

def create_seismic_manifest(survey_id, segy_file, legal_tags):
    """Create OSDU manifest for 3D seismic survey"""
    
    manifest = {
        "kind": "osdu:wks:Manifest:1.0.0",
        "ReferenceData": [],
        "MasterData": [],
        "Data": {
            "WorkProduct": {
                "kind": "osdu:wks:work-product--SeismicTraceData:1.0.0",
                "id": f"work-product--SeismicTraceData:{survey_id}",
                "legal": legal_tags,
                "data": {
                    "Name": f"Seismic Survey {survey_id}",
                    "Description": "3D seismic data",
                    "CreationDateTime": datetime.utcnow().isoformat(),
                    "DatasetIDs": [f"dataset--File.Seismic:{segy_file}"],
                    "ExtensionProperties": {}
                }
            },
            "Datasets": [{
                "kind": "osdu:wks:dataset--File.Seismic:1.0.0",
                "id": f"dataset--File.Seismic:{segy_file}",
                "data": {
                    "Name": segy_file,
                    "DataContentType": "application/x-segy"
                }
            }]
        }
    }
    return json.dumps(manifest, indent=2)

# Example usage
manifest = create_seismic_manifest(
    survey_id="survey-2024-001",
    segy_file="seismic_001.sgy",
    legal_tags={"legaltags": ["my-legal-tag"], "otherRelevantDataCountries": ["US"]}
)

From SEG-Y to OSDU Workflow

1. Quality check SEG-Y file using segyio (see Capabilities section above)
2. Create OSDU manifest with SeismicBinGrid and SeismicLineGeometry3D schemas
3. Upload SEG-Y to cloud storage via Dataset Service
4. Trigger SegYToVDS DAG for conversion to VDS format
5. Access VDS via Seismic DMS API

OSDU Reference Documentation

• OSDU SEGY to ZGY Conversion: https://gitlab.opengroup.org/osdu/r3-program-activities/docs/-/raw/master/R3%20Document%20Snapshot/osdu-segy-to-zgy-conversion.pdf
• OSDU Dataset Service: https://community.opengroup.org/osdu/platform/system/dataset
• Seismic Schemas: https://community.opengroup.org/osdu/data/data-definitions

Software Tasks

If user requests:

• OSDU data ingestion pipeline → invoke petropowers:oil-gas-delegation
• Seismic visualization web app → invoke petropowers:oil-gas-delegation
• OSDU schema validation → stay in domain skill (validate against OSDU schemas)