Geosoft-GDB-Conversion

Work looking at converting Geosoft GDBs to useable open format for analysis, machine learning and AI

📋 Important: Feasibility Analysis Available

See GDB_REVERSE_ENGINEERING_FEASIBILITY_ANALYSIS.md for a comprehensive analysis of GDB reverse engineering feasibility.

Key Finding: While reverse engineering exploration is educational, using the official geosoft.gxpy Python API is strongly recommended for all practical applications. See src/example_conversion.py for a working example.

---

Geosoft GDB info

• "Geosoft database files are made up of straight binary data"
• https://help.seequent.com/Oasismontaj/2023.2/Content/ss/prepare_om/work_with_databases/c/oasis_databases.htm
  • To quote "proprietary 3-dimensional-file format"
    • Lines, channels, elements

Data storage

VV

• Vector Vectors - like (18,2) - 18 records, 0 has the actual measurements, 1 has the fiducial.

VA

• Vectory Array - like (18,13) - 18 records, e.g. IP data so it can show up as plotted lines in the interface

• Given it is a 'spreadsheet' - you can get (0,) data as well for empty channels (or fields?)

Example data

• Plenty of GDBs to be found
  • e.g. https://geoscience.data.qld.gov.au/ [search for geophysics, choose filetype on the left]
    • https://geoscience.data.qld.gov.au/data/magnetic/mg001116

Binary Headers

• Need to compare the first part of a reasonable sample of files
  • QLD has 300, perhaps pull those

Lines

• Name - TYPE number.version:flight

Channels

• Names have no spaces

Elements

• byte
• unsigned 2-byte integer
• short 2 int
• long 4 int
• float 4
• double 8
• string

Channels

Fiducial

• Incrementing field

Conversion format

• What to store as?

• Zarr / netcdf

  • Use xarray [not looked at tree format]
  • Depending if archiving or wanting to use for analysis
  • Although if you archive as netcdf you can kerchunk it later

• HDF5

  • Nested
  • Geoh5
    • would need to implicitly make dummy coordinates

Compression

• Looks like gzip signature found as reported by Loop/Fatiando
  • but 1 and in the middle?

binwalk

 binwalk -e DB_1116.gdb

DECIMAL       HEXADECIMAL     DESCRIPTION
--------------------------------------------------------------------------------
4996991       0x4C3F7F        Certificate in DER format (x509 v3), header length: 4, sequence length: 18436
11004320      0xA7E9A0        MySQL MISAM index file Version 8
11666672      0xB204F0        MySQL MISAM compressed data file Version 6

Note: binwalk results show false positives. See feasibility analysis for details.

---

Recommended Approach: Official geosoft.gxpy API

For practical GDB file access, use the official Python API:

pip install geosoft

Example usage (see src/example_conversion.py for complete working code):

import geosoft.gxpy.gdb as gxdb
import geosoft.gxpy.gx as gxp
import numpy as np

with gxp.GXpy() as gxp:
    gdb = gxdb.Geosoft_gdb.open('your_file.gdb')
    
    # List available data
    lines = gdb.list_lines()
    channels = gdb.list_channels()
    
    # Read data
    for line in lines:
        for channel in channels:
            data = gdb.read_channel_vv(line, channel)
            np_array = np.asarray(data)
            # Process your data here

Resources:

• Official API: https://github.com/GeosoftInc/gxpy
• Documentation: https://geosoftinc.github.io/gxpy/
• Full analysis: GDB_REVERSE_ENGINEERING_FEASIBILITY_ANALYSIS.md