CircuitNotion/agri-ghg-yield-synthetic
Synthetic Crop Yield & GHG Emission Dataset File: agri_ghg_yield_synthetic_dataset.csv — 6,000 rows, 20 columns Why synthetic No single public dataset combines all 18 requested predictors with all 4 targets in a clean, ready-to-train form. The closest real match is USDA GRACEnet / AgCROS, but it's spread across dozens of Excel workbooks per site. This CSV is generated from real agronomic equations (not random noise) so you can build and test a full ML pipeline… See the full description on the dataset page: https://huggingface.co/datasets/CircuitNotion/agri-ghg-yield-synthetic.
Synthetic Crop Yield & GHG Emission Dataset
File: agri_ghg_yield_synthetic_dataset.csv — 6,000 rows, 20 columns
Why synthetic
No single public dataset combines all 18 requested predictors with all 4 targets in a clean, ready-to-train form. The closest real match is USDA GRACEnet / AgCROS, but it's spread across dozens of Excel workbooks per site.
This CSV is generated from real agronomic equations (not random noise) so you can build and test a full ML pipeline today, then swap in or fine-tune against real field data later.
Columns
How targets were derived
- Crop_Yield — Liebig's Law of the Minimum, modulated by weather and stress factors
- N2O_Emission — IPCC Tier-1 style (~1.25% of N input), amplified by moisture/temp/pH
- CO2_Emission — Heterotrophic soil respiration (OM, Q10=2.2, moisture curve)
- CH4_Emission — Upland soils as CH4 sink; switches toward emission above ~45% moisture
Gaussian/lognormal noise mimics measurement variability.
Load with datasets
from datasets import load_dataset
ds = load_dataset("CircuitNotion/agri-ghg-yield-synthetic")