Deep Learning Forecasts of Canadian Crop Suitability for 2050 and 2100

A 6.1 MB PDF document authored by Amanjot Bhullar, published on figshare in April 2026, under a CC-BY-4.0 license. It presents forecasts of Canada's cropland suitability for major annual crops in 2050 and 2100 under two climate scenarios (RCP 4.5 and 8.5), generated using a deep learning model trained on historical soil, yield, and climate data.

Use Cases

• Forecast regional crop suitability shifts based on projected climate change scenarios.
• Evaluate the impact of CO2 fertilization assumptions on yield projections.
• Identify areas for crop diversification or heat-resilient variety development based on projected suitability losses.
• Compare suitability trends for different crops (canola, peas, wheat, soy, barley, oats) across Canadian regions.

Strengths

• The forecasts are based on a deep learning model trained on historical soil, yield, and climate data.
• Projections cover two future time points (2050 and 2100) and two climate scenarios (RCP 4.5 and RCP 8.5).
• The dataset is openly available under a CC-BY-4.0 license.

Limitations

• The underlying data structure (columns, rows) is unknown, limiting suitability assessment for direct analysis.
• The data is contained in a PDF report (6.1 MB), which may require extraction to access structured data.
• The model's projections hinge on contentious assumptions about CO2 fertilization effects.

Provenance

Source

figshare

Collection Method

Generated using a deep learning approach trained on historical soil, yield, and climate data.

Time Range

Projections for 2050 and 2100.

Freshness

Last updated 2026-04-24 05:39:48; freshness should be verified.

Geography

Canada, with focus on the Prairies, central British Columbia, north of Southern Ontario, and Southern Quebec.