15. EMTF

pycsamt.emtf is the format-neutral scientific core for electromagnetic transfer functions: matrix-oriented TransferFunction and StatisticalEstimate objects held inside one EMTF document, together with the adapters that read and write it as historical SEG EDI, EDI SPECTRA, or modern EMTF XML. The central design rule, following Kelbert (2020)’s EMTF XML paper: EDI and EMTF XML are formats; the transfer function is the scientific object. Impedance, tipper, variance, and full covariance are represented once, independently of which file produced or will store them, and a conversion between formats never silently discards or fabricates content – a missing element stays missing rather than becoming a zero or an invented default.

This section assumes the metadata objects covered in Metadata – SiteMeta, ProvenanceMeta, SiteLayout, OrientationMeta, ProcessingMeta, TransferFunctionQuality – since EMTF holds exactly those objects as its document metadata rather than reinventing a format-specific metadata model. Use this section when the question is about the transfer function itself: its matrix shape and channel axes, how it survives a round trip between EDI and XML, how EDI SPECTRA’s cross-power matrices are recovered into full covariance, or how rotation is applied without disturbing the physical site layout that produced it. For the complete callable reference, see pycsamt.emtf.

• 15.1. The EMTF Document
  • 15.1.1. Loading a Real Document
  • 15.1.2. Retrieving a Transfer Function by Name or Alias
  • 15.1.3. Building a Document From Scratch
  • 15.1.4. Convenience Accessors and Units
  • 15.1.5. A Real Sounding Curve
  • 15.1.6. Bridging to TFBundle
  • 15.1.7. Legacy Field Conflict Detection
  • 15.1.8. Choosing the Right Accessor
  • 15.1.9. Next Steps
• 15.2. Transfer Functions and Estimates
  • 15.2.1. Matrix Shape and Validation
  • 15.2.2. Name Normalization and Units
  • 15.2.3. The Missing-Value Policy
  • 15.2.4. Statistical Estimates
  • 15.2.5. The Datatype Registry
  • 15.2.6. Registering a Custom Datatype
  • 15.2.7. Choosing the Right Object
  • 15.2.8. Next Steps
• 15.3. EDI Interoperability
  • 15.3.1. What EDI Extraction Preserves
  • 15.3.2. Variance and the Legacy z_err Convention
  • 15.3.3. Converting Back to EDI
  • 15.3.4. When EDI Writing Refuses Outright
  • 15.3.5. Round-Trip Verification
  • 15.3.6. The TFBundle Bridge
  • 15.3.7. Choosing the Right Function
  • 15.3.8. Next Steps
• 15.4. EDI SPECTRA Recovery
  • 15.4.1. A Real Remote-Reference File
  • 15.4.2. Recovering Transfer Functions and Full Covariance
  • 15.4.3. spectra_to_emtf and EMTF.from_edi’s Default Preference
  • 15.4.4. Why the Preferred Path Matters: a Real Sign Difference
  • 15.4.5. Choosing the Right Function
  • 15.4.6. Next Steps
• 15.5. Reading and Writing EMTF XML
  • 15.5.1. Reading a Real Document Directly
  • 15.5.2. Strict Rejects, Permissive Recovers
  • 15.5.3. Missing Stays Missing on Round Trip
  • 15.5.4. Deterministic Element Order and Precision
  • 15.5.5. When Writing Refuses
  • 15.5.6. Full Round-Trip Verification
  • 15.5.7. Choosing the Right Function
  • 15.5.8. Next Steps
• 15.6. Rotation and Covariance
  • 15.6.1. Rotation Matrices
  • 15.6.2. Orientation Must Be Unambiguous First
  • 15.6.3. Approximate Variance vs. Exact Covariance
  • 15.6.4. The True Determinant Is Rotation-Invariant
  • 15.6.5. Identity, Round Trip, and the Untouched Site Layout
  • 15.6.6. Legacy EDI Rotation and Rotating Back to Site Layout
  • 15.6.7. Choosing the Right Option
  • 15.6.8. Next Steps