APSIM-Wheat: 9,100 Virtual Wheat Genotypes Simulated Across 9,000 Environments

9,100 virtual wheat genotypes were simulated using APSIM-Wheat across 9,000 distinct environments. The dataset was created by Tien-Cheng Wang of Humboldt-Universität zu Berlin, varying 90 physiological parameters for each genotype. Environments combine 125 years of historical climate data from four Australian locations with varying CO2 levels, nitrogen fertilization, and sowing dates.

Use Cases

• Predicting wheat yield based on simulated physiological parameters and environmental conditions.
• Analyzing yield stability across diverse climate and management scenarios described in the dataset.
• Studying genotype-by-environment interactions using the 9,100 virtual genotypes and 9,000 simulated environments.
• Training machine learning models to identify key physiological parameters influencing crop performance.

Strengths

• Large scale with 9,100 distinct virtual genotypes and 9,000 simulated environments.
• Environmental conditions are detailed, including 125 years of historical climate data from four locations.
• Genotypes are systematically varied across 90 independent physiological parameters.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data is simulated, not observed, which may limit direct applicability to real-world field conditions.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.

Provenance

Source

Humboldt-Universität zu Berlin (Author: Tien-Cheng Wang), via paperswithcode.

Collection Method

Computational simulation using the APSIM-Wheat model, based on methods from Casadebaig et al., 2016.

Time Range

Climate data spans 1889 to 2013.

Geography

Four locations in Australia: Emerald, Narrabri, Yanco, and Merredin.