Data Foundation

Scientific datasets powering Eden

Eden prioritizes transparent, validated, and institutionally recognized data sources to ensure forecast credibility.

CHIRPS v2.0

Very High credibility

Climate Hazards Center, UC Santa Barbara

Satellite-gauge blended precipitation observations used for rainfall trend and anomaly analysis.

CADENCEDaily and dekadal updates

COVERAGE50N-50S, 1981-present

FORMATNetCDF / GeoTIFF

  • - Peer-reviewed methodology and validation
  • - Widely adopted by humanitarian and research institutions
  • - Long operational archive suitable for climate baselines

NASA POWER

High credibility

NASA Langley Research Center

Global meteorological and solar radiation variables used for evapotranspiration and stress modeling.

CADENCENear-real-time daily

COVERAGEGlobal coverage

FORMATAPI / JSON

  • - Maintained by NASA scientific infrastructure
  • - Standardized access through stable API endpoints
  • - Used in climate and energy modeling workflows

FAOSTAT and FAO Agro-informatics

Very High credibility

Food and Agriculture Organization

Crop production baselines, agroeconomic variables, and national statistics for model calibration and benchmarking.

CADENCEAnnual and periodic releases

COVERAGECountry and regional

FORMATCSV / API

  • - Institutional standard for agricultural statistics
  • - Longitudinal historical data for trend evaluation
  • - Compatible with country-level policy reporting

Sentinel-2 Vegetation Index Layer

High credibility

Copernicus Open Access

Optical satellite imagery transformed into vegetation health indicators to improve crop stress detection.

CADENCE5-day revisit frequency

COVERAGEContinental, parcel to district scale

FORMATSAFE / Cloud-optimized GeoTIFF

  • - European Space Agency mission with open access
  • - High-resolution spatial detail for agricultural monitoring
  • - Consistent spectral bands for vegetation analytics

Credibility Controls

How data quality is monitored before model execution

Each cycle includes integrity checks to prevent low-quality signals from propagating into recommendations.

Source Provenance Tracking

Every record maintains source, timestamp, and transformation lineage for reproducible analytical audits.

Automated Validation Rules

Outlier detection, missingness thresholds, and regional consistency checks are applied before features enter model pipelines.

Versioned Data Snapshots

Each run uses immutable snapshots, enabling historical comparisons and clear attribution of model performance changes.