noaa-gml-file-reader

Read NOAA GML atmospheric trace-gas ASCII data files as pandas.DataFrame objects — one function, read_data(path).

Problems this solves

Reach for this if you are trying to:

• Read NOAA GML (ESRL/GMD) CO₂ or CH₄ data files in Python without writing a bespoke header parser.
• Load Mauna Loa (or any station's) CO₂ data into a pandas DataFrame — surface-flask (event, monthly) or in-situ hourly ccgg files.
• Parse NOAA greenhouse-gas ASCII (.txt) files whose #-commented headers and -999.99 missing-value sentinels trip up pandas.read_csv.
• Handle both the pre-2020 (GMD) and current (GML) header formats without knowing which dialect a given file uses — the reader auto-detects.

What this reads

NOAA's Global Monitoring Laboratory (GML) publishes long-term atmospheric trace-gas measurements (CO₂, CH₄, and others) as whitespace-delimited ASCII files with a #-commented header — surface-flask event and monthly series, in-situ hourly/daily series, and more, at gml.noaa.gov/aftp/data/trace_gases. This package turns one of those files into a tidy DataFrame, with the header's column names and the documented missing-value sentinels handled for you.

Lineage (GMD → GML). NOAA renamed ESRL's Global Monitoring Division (GMD) to the Global Monitoring Laboratory (GML) in 2020, and the file header format changed with it. This package was originally noaa_esrl_gmd_file_reader (2020); v2 renames it to noaa-gml-file-reader (import noaa_gml_file_reader) and reads the current header format. See the breaking-change note below.

Install

pip install noaa-gml-file-reader

Requires Python ≥ 3.10 and pandas ≥ 2.0.

Usage

from noaa_gml_file_reader import read_data

df = read_data("co2_mlo_surface-flask_1_ccgg_month.txt")
print(df.head())

Real output (from the committed test fixture of the above Mauna Loa monthly file):

  site  year  month   value
0  MLO  1969      8  322.50
1  MLO  1969      9  321.36
2  MLO  1969     10  320.74
3  MLO  1969     11  321.98
4  MLO  1969     12  323.78

Columns come straight from the file's header; numeric columns are numeric (value is float64, year/month are int64), and NOAA's missing-value sentinels (-999.99, -999.999, nan) are parsed as NaN.

Supported dialects

read_data auto-detects the header dialect — you don't need to know NOAA's format history to read a file:

The verified product set is CO₂/CH₄ surface-flask event + monthly and CO₂ in-situ hourly data at MLO. The header grammar is site- and gas-agnostic, so other stations/species in the same product families parse the same way.

Errors

An unparseable file raises UnrecognizedFormatError (carrying the path and the file's first line) rather than returning data:

from noaa_gml_file_reader import read_data, UnrecognizedFormatError

try:
    df = read_data("not-a-noaa-file.txt")
except UnrecognizedFormatError as err:
    print(err)

v2 breaking changes

v2 is a breaking release (hence the major bump):

• Renamed noaa_esrl_gmd_file_reader → noaa_gml_file_reader (distribution noaa-gml-file-reader).
• Raises UnrecognizedFormatError on unrecognized/malformed input. v1 returned an empty DataFrame on any parse failure — silently indistinguishable from "the file had no data" — which meant v1 failed silently on every current (post-2020) NOAA file. If you relied on the empty-DataFrame behavior, catch the exception instead.
• Relicensed AGPL-3.0 → MIT.

Data & citation

The data are NOAA GML's, not this package's. Please cite the measurements per NOAA GML's guidance — each product directory under gml.noaa.gov/aftp/data/trace_gases ships a species-specific README with the citation text and data providers (the file headers also carry contact and reciprocity information). This library only parses the files; it does not download them.

License

MIT — see LICENSE. Built by Erick Shepherd.