This project aims to develop site-specific fertilizer recommendations for cereal crops in Ethiopia’s Amhara region. By compiling 15+ years of on-farm fertilizer trial data and applying Machine Learning (ML) and AI models, the project seeks to move away from one-size-fits-all recommendations toward data-driven, location-specific guidance that improves yields, optimizes fertilizer use, and reduces environmental and economic waste.

Objectives

– Collate and standardize historical fertilizer response data from multiple locations and years.

– Analyze yield responses across different fertilizer rates, soils, and climatic conditions.

– Identify drivers of variability in fertilizer response at field and regional scales.

– Benchmark the Amhara data against continental datasets for broader applicability.

– Develop and validate ML/AI models to generate accurate, site-specific fertilizer recommendations.

Research Questions

– What are the yield response patterns to fertilizer across different agro-ecological zones?

– Which environmental and management factors explain variability in fertilizer response?

– How accurately can ML/AI models predict optimal fertilizer rates at farm and field scales?

– Can ML/AI-based recommendations be translated into practical decision-support tools for farmers?

More details about the project : 

– Main collaborators: This project is implemented in collaboration with Kansas State University and the Amhara Region Agricultural Research Institute.

– Target cropping systems: Cereal cropping systems

– Target regions: Amhara Region, Ethiopia