Hyperspectral imaging solution for precision fertigation management in greenhouse tomato production

Overview

This solution addresses the challenge of optimizing fertigation in greenhouse tomato production by using hyperspectral imaging of plant leaves to determine nutritional status and water content. By correlating spectral reflectance data with chemical leaf analysis across varying fertigation regimes, the approach produces regression models that predict nutrient and water needs. The ultimate deliverable is a low-cost mobile multispectral measurement device, termed an "Agroscanner," designed for in-field leaf scanning. The technology aims to reduce fertilizer waste, prevent over- or under-irrigation, and support more sustainable and productive greenhouse operations.

Technical specifications

Core approach:

• Hyperspectral imaging of tomato leaves at multiple plant heights (top, middle, bottom) across different greenhouse rows
• Controlled variation of fertigation levels to generate a wide range of leaf chemical compositions
• Periodic laboratory chemical analysis of leaves for macronutrients (N, P, K, Mg, Ca, S), micronutrients (Na, Fe, Mn, B, Cu, Zn), and water content
• Regression models correlating spectral data with chemical measurements to predict plant nutritional status
• Algorithm development to convert spectral readings into actionable fertigation recommendations

Expected performance benchmarks (based on prior published studies on maize and soybean):

• Leaf water content prediction: high accuracy (R² ≈ 0.93, RPD ≈ 3.8)
• Macronutrient quantification: satisfactory accuracy (R² from 0.69 to 0.92, RPD from 1.62 to 3.62)
• Micronutrient quantification: lower accuracy (R² from 0.19 to 0.86, RPD from 1.09 to 2.69)

Planned deliverable:

• A low-cost, mobile "Agroscanner" device for routine multispectral leaf measurements in greenhouse settings

Technology readiness level

The project builds on prior proof-of-concept work using conventional digital imaging to correlate leaf area and vegetation indices with weekly tomato production. The hyperspectral extension is at an early-to-mid research stage, with validation planned through controlled greenhouse trials across multiple fertigation regimes. Model accuracy and the Agroscanner prototype will be refined through iterative testing. The technology is currently at a concept-to-prototype stage and is not yet commercially deployed; successful validation would position it for pilot deployment in commercial greenhouse operations.