Data Sources & Methodology
GoodSeason is committed to transparency. Every data point shown in the app is backed by cited sources. This page explains our methodology, data sources, and known limitations.
Our Principles
Cite everything. Every numeric value has provenance.
Show uncertainty. We display ranges and confidence scores, not false precision.
Don't overclaim. We show assumptions alongside recommendations.
Guide, not judge. We present data for informed choices, not moral verdicts.
Transparency. Our data pipeline, sources, and limitations are documented here.
Greenhouse Gas (GHG) Emission Factors
Primary: Poore & Nemecek (2018)
Our baseline GHG factors come from the comprehensive meta-analysis by J. Poore and T. Nemecek, “Reducing food's environmental impacts through producers and consumers,” published in Science, Vol. 360, Issue 6392, pp. 987-992 (2018). This study synthesized data from ~38,700 farms and 1,600 processors across 119 countries.
We use the dataset as distributed by Our World in Data (OWID), which provides accessible, tabulated versions of the Poore & Nemecek data.
Regional: AGRIBALYSE 3.x
Where available, we use AGRIBALYSE (maintained by ADEME, France) for region-specific LCA data, particularly for French/EU food products. This provides higher-quality, locally-relevant emission factors.
License: Open data (Etalab 2.0). Factors marked “High Quality Data” when using region-specific LCA.
Seasonality Data
Seasonality information is derived from FAO crop calendar data and supplemented with national/regional growing season information. We map each food to known harvest windows by country and climate zone.
User location is resolved to a country + admin region, then mapped to a climate zone (using a simplified Köppen classification). The “in-season probability” reflects how likely a food is to be available from local production in a given month.
Confidence scores are assigned based on data source quality: High (direct national crop calendar), Medium (regional/climate-zone inference), Low (global generalization).
Water-Stress Risk
Water-stress risk indicators use data from the WRI Aqueduct Water Risk Atlas (World Resources Institute). We compute a risk bucket for likely origin regions of each food.
Important: This is a risk indicator, not a precise water footprint. It reflects regional water stress conditions, not crop-specific water consumption. Use it as one signal among many.
Heated Greenhouse Badge
The “Heated Greenhouse Likely” badge appears when warm-season crops (e.g., tomatoes, peppers, cucumbers) are viewed in cold-climate regions during winter months. Research shows that greenhouse heating can significantly increase emissions:
Theurl et al. (2014) found that heated greenhouse production of tomatoes in Austria resulted in 2-3x higher GHG emissions compared to imports from Spain during winter months. Similarly, Hospido et al. (2009) demonstrated that transport emissions from warmer regions can be lower than the heating energy required for local greenhouse production.
This badge is a heuristic based on crop type, climate zone, and month. It does not apply to unheated greenhouses, tunnels, or mild-climate regions.
Data Quality Badges
Region-specific LCA data (e.g., AGRIBALYSE for EU)
Reputable global average (Poore & Nemecek 2018)
Imputed or estimated from similar foods
Known Limitations
- Global averages mask local variability. Emissions for the same food can vary 10-50x between producers. Our ranges capture some but not all of this.
- Seasonality is approximate. Actual growing seasons depend on specific microclimates, varieties, and farming practices.
- Water-risk is regional, not crop-specific. A region's water stress doesn't tell you the specific water footprint of a crop.
- Transport is simplified. We don't model specific supply chains or distinguish between road, rail, ship, and air freight.
- Data currency. Primary data is from 2018 meta-analysis. Farming practices and emissions may have changed since then.
- System boundaries. LCA data typically covers farm-to-retail. Consumer transport, cooking, and food waste are not included.
- Not a complete picture. Environmental impact includes biodiversity, land use, pollution, and social factors that are not captured by CO₂e alone.
Factor Selection Logic
When displaying emission factors, we follow this priority:
- Region-specific factor if available for user's country/region
- Continent-level factor if available
- Global average (Poore & Nemecek 2018)
System modifiers (e.g., heated greenhouse vs. open field) are only applied when supported by data. Otherwise, the system is shown as “unknown.”