1. Choose a workflow

The user guide explains how to move electromagnetic observations from field files to reviewed scientific products. It assumes that pyCSAMT is already installed and importable. If that is not yet true, follow the guided Install pyCSAMT page or consult the complete Installation reference.

Most workflows share the same backbone:

identify the input format
    -> load and inspect the survey
    -> check data quality
    -> correct or transform the responses
    -> model or invert
    -> interpret and report

The exact route depends on the instrument output and the scientific question. Choose the nearest starting point below rather than reading every section in order.

1.1. Starting from EDI files

Begin with Loading electromagnetic data. It explains the public read_edis survey view, the ensure_sites processing boundary, station ordering, duplicate handling, and frequency-grid checks. Continue to EM Tools Guide for quality control, noise removal, frequency editing, dimensionality analysis, and static-shift correction.

Map the inspected stations with Map Tools. If elevation must influence a section or model, prepare it with Topography before interpreting the result. A syntactically valid EDI file is not necessarily scientifically usable, so do not skip the inventory and quality-control stages.

1.2. Starting from instrument or intermediate formats

Use Transformers when the input is AVG, J-format, spectra, time-series, or transient electromagnetic data that must be converted or normalized before the standard EDI workflow.

Geometrics/EMI Stratagem surveys have additional hardware-specific steps. Stratagem Surveys covers raw import, coordinate injection, WinGLink interchange, and validation before the data enter the general processing tools. For connected acquisition systems, IoT-Enabled Field Acquisition covers device configuration, telemetry, timing, edge quality control, and provenance.

1.3. Preparing an inversion

After reviewing the observations, use Inversion to choose an inversion path. That guide separates three related layers:

• Classical model integrations configures and runs classical external solvers;

• AI inversion covers learned and physics-informed inversion;

• Forward Modelling generates responses from known earth models for sensitivity studies, validation, and synthetic training data.

An inversion result is not the end of the workflow. Continue to Interpretation to evaluate uncertainty, geological context, and the claims that the data can support. Use the mapping and topography guides to prepare spatially defensible figures rather than treating model colours as a standalone interpretation.

1.4. Working with stations directly

Site Tools documents station-level selection, coordinate and metadata editing, profile construction, normalization, and export. Use it when the scientific task concerns which stations enter the workflow or how station metadata are represented. Use EM Tools Guide when the task concerns the electromagnetic responses themselves.

1.5. Automating a reviewed workflow

First establish the workflow with explicit API calls and inspect its intermediate products. Then use Pipeline System to record the sequence as a named, versioned recipe. Agents adds higher-level automation and orchestration; it does not remove the need to validate data, configuration, and scientific assumptions.

Warning

Automation can reproduce an invalid workflow as reliably as a valid one. Record station identities, input paths, coordinate reference information, processing parameters, software version, and random seeds where applicable.

1.6. Finding the right kind of documentation

Use the user guide for concepts, decisions, and reusable workflows. The other documentation sections serve different purposes:

• Getting Started installs pyCSAMT and completes a first run;

• Tutorials provides longer end-to-end worked examples;

• Scientific Background explains the physical and mathematical background;

• API reference gives signatures and object-level API details.

When you know the scientific goal but not the function name, stay in the user guide. When you already know the object and need its exact parameters or return type, switch to the API reference.