A Views & Comments paper published in Engineering systematically reviews mainstream on-site soil nutrient sensing technologies, outlining their respective operational logics, applicable scenarios and persistent technical limitations, while sorting out prevailing development trends for precision agricultural monitoring.

Authored by researchers from the Beijing Academy of Agriculture and Forestry Sciences and Oklahoma State University, the work opens by noting that soil carries fundamental agricultural value, yet varied crops bear distinct nutrient demands: leafy vegetables require abundant nitrogen for foliage development, fruit-bearing varieties depend on phosphorus and potassium for reproductive organ formation, cereal crops need nitrogen and silicon, and fruit trees show sensitivity to calcium and magnesium.

Widespread chemical fertilizer application has lifted crop yields, but improper use creates environmental risks to underground water bodies; meanwhile, conventional soil nutrient testing relies on field sampling followed by laboratory analysis, a lengthy workflow that restricts real-time variable-rate fertilization and hinders mature precision agriculture implementation.

Two core barriers complicate direct field detection: soil nutrients exist in diverse chemical forms, with only nitrate nitrogen (NO₃⁻​−N) and ammonium nitrogen (NH₄⁺​−N) readily absorbable by crops, and the heterogeneous soil matrix creates inconsistent physical and chemical conditions even within a single farm plot.

The paper categorizes established field sensors into four technical branches. Onsite colorimetric and electrochemical sensors originate from simplified lab analytical workflows, with portable multi-band colorimetric devices and paper-based microfluidic chips already commercialized for semi-quantitative field testing, though their performance fluctuates with reagent quality and human operation errors.

Ion-selective electrodes (ISEs) lower manual interference, yet cross-sensitivity among soil extract components, electrode baseline drift and surface poisoning limit long-term continuous testing, and both sensor types still require soil pre-extraction steps.

Near-infrared (NIR) spectroscopy leverages molecular spectral fingerprint features and permits minimally processed or direct soil measurement, supported by multiple commercial handheld and machinery-mounted devices, yet the overlapping, indistinct absorption signals of NIR spectra demand chemometric and machine learning interpretation, and calibrated prediction models often lose validity when transferred across different fields or growing seasons.

Laser-induced breakdown spectroscopy (LIBS), adapted from planetary rock detection technology used by Mars rovers, delivers rapid multi-element quantification without complex pretreatment, yet atmospheric nitrogen interference obstructs accurate soil nitrogen measurement, and the technique cannot distinguish bioavailable nutrient forms from total elemental content, limiting agricultural applicability.

γ-ray detection passively captures natural soil isotope radiation to map nutrient distributions via vehicle-mounted detectors, yet stable isotope proportional relationships are disrupted by rainfall and fertilization activities, requiring repeated pre-calibration through laboratory testing to maintain reliable outputs.

The paper concludes that no single sensing method balances field operability, detection accuracy and cost performance perfectly, and integrating artificial intelligence algorithms into sensor systems stands as a clear research direction.

Researchers have built machine learning frameworks to derive multiple soil nutrient indicators from single-sensor readings, while the long-term development target for soil nutrient sensing focuses on pretreatment-free, buryable in-situ sensors. From soil science perspectives, resolving on-site quantification of plant-available nutrients such as nitrate nitrogen (NO₃⁻​−N) remains the top priority for subsequent technical iteration.

The paper “The War for Fertile Soil: Advancements in Soil Nutrient Field Sensors,” is authored by Daming Dong, Ning Wang, Hongwu Tian, Shixiang Ma, Chunjiang Zhao. Full text of the open access paper: https://doi.org/10.1016/j.eng.2025.04.029. For more information about Engineering, visit the website at https://www.sciencedirect.com/journal/engineering.