GribStream

GribStream Blog

NOAA proposes replacing GEFS aerosol products with GCAFS v1.0

|
gefschem gefsatmos

NOAA/NCEP is soliciting comments on replacing GEFS v12 aerosol products with the Global Chemistry and Aerosol Forecast System (GCAFS) v1.0, including new cadence, new analyses (GCDAS), and updated aerosol modeling and assimilation.

On January 28, 2026, NOAA's Environmental Modeling Center (EMC) released Public Information Statement 26-07, soliciting comments through February 28, 2026 on a proposal to replace GEFS v12 aerosol products with the Global Chemistry and Aerosol Forecast System (GCAFS) v1.0.

The stated goal is to improve predictions of dust, sea salt, wildfire smoke, and other aerosols that affect visibility and human health, while separating aerosol prediction from the operational GEFS framework.

What would change (high level)

  • GEFS aerosol products would be retired, and GCAFS v1.0 would become the primary publicly distributed aerosol forecast product.
  • GCAFS introduces a companion analysis system, GCDAS (Global Constituent Data Assimilation System), producing analyses used to initialize the forecasts.
  • The proposal shifts the forecast cadence to two forecast cycles per day (00Z and 12Z) for the main aerosol forecast run.

Proposed cadence, horizons, and product structure

Based on the GCAFS v1.0 product supplement, the proposed public products include:

  • Forecasts: a 5-day GCAFS forecast at 00Z and 12Z, with 3-hourly output.
  • Analyses: GCDAS analyses at 00Z/06Z/12Z/18Z, with products valid at initialization time only.
  • 2D "column + surface" aerosol products (described as a 0.25 deg pgrb2 output), including:
    • Aerosol optical depth (AOD) and related optical parameters across multiple spectral bands.
    • Surface particulate matter fields (PM2.5 / PM10 style products).
    • Species-resolved column mass (dust/sea salt/sulfate/organic matter/black carbon).
  • 3D aerosol mixing ratio products across a set of isobaric pressure levels, intended to capture vertical structure (surface vs lofted plumes).

If you currently ingest "GEFS aerosols" as a single family of fields, treat this proposal as a product-family migration: output cadence, file naming, and field lists are likely to differ even if NOAA aims for continuity.

Notable technical changes EMC highlighted

EMC's proposal calls out several concrete system updates, including:

  • Vertical resolution increase from 64 to 127 layers (to match the GFS vertical resolution).
  • Updates to dust emission parameterization (fengsha), wet scavenging, and settling velocity.
  • A chemical mechanism update to GOCART2G aerosol modeling.
  • Emissions dataset updates (including a CEDS 2022 base version mentioned for anthropogenic emissions).
  • Assimilation of VIIRS aerosol optical depth (AOD) retrievals from NOAA/NESDIS using a 3D-Var approach in JEDI.

Why it matters

  • Breaking changes for pipelines: Even when products are intended to be "continuous", new files/fields and cadence differences can break download and parsing logic.
  • Backtesting discontinuities: A change in vertical resolution, emissions, and assimilation can shift biases and distributions (especially for AOD/PM products), so long-running baselines should be split across the transition.
  • Operational workflow changes: Moving from four cycles/day to two cycles/day for the forecast run can change update expectations for smoke/dust monitoring workflows.

What this means for GribStream users

GribStream already supports:

This PNS is also a follow-up in the broader NOAA discussion about evolving GEFS aerosol products (see our earlier writeup on the 06Z/18Z aerosol-cycle proposal). If GCAFS v1.0 and its public outputs become available via the AWS Open Data program (or another stable primary distribution channel), GribStream will add support and document:

  • Which products map cleanly to the legacy GEFS aerosol fields (and which do not).
  • The cadence/horizon differences (00/12 only, 5-day horizon, and analysis-only cycles).
  • Recommended migration and verification checks for downstream users.

Links

Sources

Related models

gefschem gefsatmos
Try GribStream