Soil Properties Predicted from MIRS Analysis for Lime Trials in Kakamega Kenya, 2021

61 topsoil and 61 subsoil samples were collected in 2021 before establishing on-farm lime trials in Kakamega, Kenya. Selected soil properties were predicted from mid-infrared spectral (MIRS) analysis, with a subset calibrated using wet chemistry results from the International Centre for Research in Agroforestry (ICRAF). The dataset can be linked to future yield data via a unique farm identifier.

Use Cases

• Predict soil nutrient levels based on spectral analysis data.
• Calibrate machine learning models for soil property estimation using wet chemistry reference data.
• Analyze the relationship between predicted soil properties and agricultural lime application rates.
• Link soil data to crop yield outcomes using the unique farm identifier mentioned in the description.

Strengths

• Includes 122 total samples (61 topsoil and 61 subsoil) providing paired depth data.
• Uses a machine learning algorithm calibrated with wet chemistry analysis from a recognized research center (ICRAF).
• Designed for linkage with yield data via a unique farm identifier (fid).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Coordinates were truncated for privacy, limiting precise geospatial analysis.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

CIMMYT Harvested Dataverse, author Lavinia Mmbaga MADAGA.

Collection Method

Soil properties predicted from mid-infrared spectral (MIRS) analysis of collected samples, with calibration from wet chemistry.

Time Range

2021

Freshness

Last updated 2026-04-20 14:54:01; freshness should be verified.

Geography

Kakamega, Kenya