Soil Spectroscopy Evolution: A Review of Homemade Sensors, Benchtop Systems, and Mobile Instruments Coupled with Machine Learning Algorithms in Soil Diagnosis for Precision Agriculture

Abstract

In precision agriculture, soil spectroscopy has become an invaluable tool for rapid, low-cost, and nondestructive diagnostic approaches. Various instrument configurations are utilized to obtain spectral data over a range of wavelengths, such as homemade sensors, benchtop systems, and mobile instruments. These data are then modeled using a variety of calibration algorithms, including Partial Least Squares Regression (PLSR), Principal Component Regression (PCR), and Support Vector Machines (SVM), these datasets are further improved and optimized. Given the increasing demand for cost-effective and portable solutions, homemade sensors and mobile instruments have gained popularity in recent years. This review paper assesses the current state of soil spectroscopy by comparing the performance, accuracy, precision, and applicability of homemade sensors, mobile spectrometers, and traditional benchtop instruments. The discussion encompasses the technological advancements in homemade sensors, exploring innovative approaches taken by researchers and farmers, as well as developing affordable and efficient soil spectroscopy tools. Mobile and benchtop spectrometers, equipped with cutting-edge technology, have enabled easy soil diagnosis, transforming the landscape of soil analysis.

Keywords:

• Soil spectroscopy
• machine learning
• nondestructive testing
• precision agriculture