Skip to content
Mostly Right

mostlyright.weather.catalog.iem

mostlyright.weather.catalog.iem

Section titled “mostlyright.weather.catalog.iem”

IEM catalog adapter (CATALOG-01).

Wraps mostlyright.weather._iem.iem_to_observation() (and parse_iem_file) into a class that satisfies the WeatherAdapter Protocol and emits a canonical schema.observation.v1 DataFrame with overlay columns + correct SI units (m/s, metres, mm — see _obs_projection.PROJECTION_SPEC).

Source IDs handled:

  • "iem.archive" — historical CSV pulls (the v0.14.1 parity baseline)
  • "iem.live" — same parser, current-month staging cache (no semantic

difference at the parser level; the source ID drives Validator dispatch)

MOS forecast leg deferred to Phase 3 per the Phase 2 PLAN Open Q1 resolution. fetch_forecasts() raises NotImplementedError.

Pitfall 8 (codex cross-ref): IEM’s missing-data sentinel M is converted to None by the underlying parser. The projection in mostlyright.weather.catalog._obs_projection preserves None through unit conversion.

Module Attributes

Section titled “Module Attributes”
IEM_METAR_LAGKnowledge-time lag for METAR/SPECI reports.

Classes

Section titled “Classes”
IEMAdapter()IEM ASOS observation adapter.

class mostlyright.weather.catalog.iem.IEMAdapter

Section titled “class mostlyright.weather.catalog.iem.IEMAdapter”

Bases: object

IEM ASOS observation adapter.

SUPPORTED_SOURCES : ClassVar[list[str]] = [‘iem.archive’, ‘iem.live’]

Section titled “SUPPORTED_SOURCES : ClassVar[list[str]] = [‘iem.archive’, ‘iem.live’]”

fetch_forecasts(source, station, from_date, to_date)

Section titled “fetch_forecasts(source, station, from_date, to_date)”

Fetch IEM MOS forecasts via fetch_iem_mos() (Phase 17 PLAN-08).

Replaces the Phase-2 NotImplementedError stub. The default model is nbe (preferred for US stations); other supported models (gfs / lav / met / ecm) are reachable via direct fetch_iem_mos() calls.

  • Parameters:
    • source (str) – "iem.archive" (historical) or "iem.live" (v0.2 — currently raises NotImplementedError).
    • station (str) – ICAO code.
    • from_date (str) – ISO YYYY-MM-DD lower bound (inclusive).
    • to_date (str) – ISO YYYY-MM-DD upper bound (inclusive).
  • Return type: DataFrame
  • Returns: DataFrame matching schema.forecast.iem_mos.v1.
  • Raises:

fetch_observations(source, station, from_date, to_date)

Section titled “fetch_observations(source, station, from_date, to_date)”

Fetch observations from IEM ASOS.

v0.1.0: NOT YET WIRED. research() and snapshot continue to call _fetchers/iem_asos.py directly. The adapter contract is in place for Phase 3 Mode-2 dispatch.

  • Return type: DataFrame
  • Parameters:

static from_rows(rows, , source=‘iem.archive’, retrieved_at=None)

Section titled “static from_rows(rows, , source=‘iem.archive’, retrieved_at=None)”

Project parser-output rows to a canonical observation DataFrame.

Applies the shared canonical-units projection (mostlyright.weather.catalog._obs_projection): knots → m/s, miles → metres, feet → metres, inches → mm. None (IEM M sentinel) passes through unchanged.

mostlyright.weather.catalog.iem.iem_to_observation(row, observation_type_override=None)

Section titled “mostlyright.weather.catalog.iem.iem_to_observation(row, observation_type_override=None)”

Convert a single IEM CSV row to an observation schema dict.

Takes a dict from csv.DictReader (comma-delimited). If observation_type_override is set, uses that instead of detecting from raw METAR text. IEM strips the SPECI keyword from raw METAR, so callers that request report_type=3 or report_type=4 separately should pass the override. Returns observation dict with exactly 30 fields, or None if row should be skipped (invalid station, bad timestamp, or all key vars missing).