Skip to content

dts.rst

Latest commit

 

History

History
105 lines (74 loc) · 2.49 KB

dts.rst

File metadata and controls

105 lines (74 loc) · 2.49 KB

DTS Interface

.. currentmodule:: supy.dts

The DTS (Derived Type Structure) interface provides a direct Pydantic-to-Fortran execution path for SUEWS. It bypasses the intermediate df_state conversion layer used by the traditional backend.

Note

This page is an advanced API reference for developers and maintainers. Standard user workflows remain unchanged and require no DTS-specific steps.

Build Requirements

DTS features are available only in a full DTS build:

make clean && make dev-dts

Fast builds (make dev) do not include DTS type wrappers.

For most users, DTS is an internal execution detail. The public SUEWSSimulation workflow remains the recommended interface.

Main DTS Runners

.. autosummary::
    :toctree: _autosummary

    run_dts
    run_dts_multi

Factory Functions

.. autosummary::
    :toctree: _autosummary

    create_suews_config
    create_suews_site
    create_suews_state
    create_suews_forcing
    create_suews_timer
    create_output_line

Population Functions

.. autosummary::
    :toctree: _autosummary

    populate_config_from_pydantic
    populate_site_from_pydantic
    populate_state_from_pydantic
    populate_forcing_from_row
    populate_timer_from_datetime
    populate_atmstate
    populate_roughnessstate
    populate_ohmstate_defaults
    populate_storedrainprm

Extraction Functions

.. autosummary::
    :toctree: _autosummary

    extract_output_line_to_dict
    build_output_dataframe_from_block
    build_full_output_dataframe
    extract_state_from_dts

Performance Considerations and Best Practice

  • Use DTS when your workflow is YAML/Pydantic-centred.
  • For multi-grid workloads, set n_jobs via sim.run(backend="dts", n_jobs=...).
  • Use chunk_day for long runs to bound memory usage.
  • Keep forcing validation and output QA checks identical across backends.

FAQ

Why does DTS fail in a fast build?
DTS requires type wrappers that are not present in make dev builds.
Can I save DTS runs with ``sim.save()``?
Not yet. sim.save() currently requires DataFrame final state from the traditional backend.
How do I continue a DTS run?
Use sim.initial_states_final (Pydantic InitialStates) as the state source for the next run.

Related Documentation