Data-driven fertigation model for greenhouse tomato production

Overview

This solution is a data-driven fertigation model for greenhouse tomato production that predicts optimal fertilizer and irrigation regimes under varying growth media and climatic conditions. By combining sensor-based monitoring with cloud-connected analytics, the system delivers real-time fertigation recommendations that improve water and nutrient use efficiency while increasing crop yield. The approach is grounded in prior field experiments demonstrating that fertigation regime and greenhouse environment significantly influence tomato growth, yield, and resource efficiency.

Technical specifications

• Sensor monitoring layer captures weather conditions, water availability, electrical conductivity (EC), pH, and nitrate levels in real time.
• IoT cloud connectivity transmits sensor data for continuous analysis and model updating.
• Modeling environment uses system identification techniques within MATLAB to develop and refine the fertigation prediction model.
• Edge deployment runs the trained model on a Raspberry Pi microcontroller that actuates the water pump for automated fertigation.
• Experimental validation framework employs a 3 × 3 × 2 factorial design with split-split-plot layout across four replications, testing three fertilizer rates (100%, 80%, 60%), three irrigation regimes (100%, 80%, 60%), and two tomato genotypes (Jalila F1 and Yetty F1).
• Quantified impact from prior trials: greenhouse environment variation accounted for approximately 18% of variation in mean fruit weight, 44.4% in mean water use efficiency, and 46.18% in mean nutrient use efficiency.
• Farmer-facing alerts inform growers of real-time fertigation requirements at each time step.

Technology readiness level

The concept is supported by completed greenhouse trials that validated the influence of fertigation regimes on tomato yield, water use efficiency, and nutrient use efficiency. The proposed next phase focuses on building and validating the integrated IoT-to-MATLAB-to-Raspberry Pi system. The model has not yet been deployed in a commercial greenhouse setting, and further field validation is required to confirm performance across diverse substrates and climates before full commercial readiness.