The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

Dataset Card for XCO2Learn

Dataset Description

XCO2Learn is a unified, reproducible benchmark for satellite-based column-averaged carbon dioxide (XCO₂) reconstruction and forecasting. It standardizes data preprocessing, spatial–temporal splits, evaluation metrics, and model baselines to address the lack of consistent benchmarks in sparse satellite XCO₂ research.

Curated by: XCO2Learn Team
License: CC-BY-4.0
Data Format: NetCDF
Spatial Coverage: Global 0.25° grid
Temporal Coverage: 2010–2019 (10 years)
Includes: Daily files + Monthly aggregates

Dataset Structure

Daily data: Daily NetCDF files with consistent variables
Monthly data: Monthly averaged products
Unified spatial grid (0.25°)
Standardized train / validation / test splits
Identical data structure across all years

Uses

Direct Use

Sparse satellite XCO₂ reconstruction (daily & monthly)
Temporal forecasting of carbon concentrations
Evaluation and comparison of ML / physics-based models
Climate science and carbon cycle research

Out-of-Scope Use

Not intended for real-time operational carbon monitoring

Dataset Creation

Curation Rationale

Built to resolve the lack of standardized benchmarks in XCO₂ reconstruction research, enabling fair and reproducible model comparison.

Source Data

GOSAT satellite XCO₂ retrievals
CAMS reanalysis
ERA5 meteorology
ODIAC anthropogenic emissions
MODIS vegetation indices
Static geographic covariates

All data are quality-controlled, spatially aligned, and formatted into a unified global grid.