Measurement of physiochemical parameters in soil using remote laser induced breakdown spectroscopy technique

Abstract

In recent years, the global agricultural production has undergone a newlinesignificant transition by utilizing the state-of-the-art technologies. This has led newlineto the increase in food production. In addition to it, the soil analysis (physical newlineand chemical) will make a way for increase in food cultivation with high newlinequality. But, the complete soil analysis is expensive, since it requires a high newlineend instrument and numerous chemicals to complete the testing. Soil newlinespectroscopy, also known as dry chemistry, is a rapidly growing technique newlinethat uses a non-destructive, reproducible, and repeatable analytical method to newlinemanage large-scale measurements of soil characteristics. In this research newlinework, a remote Laser Induced Breakdown Spectroscopy (LIBS) system newlinecombined with machine learning algorithm is used to determine the physical newline(Texture, Moisture) and chemical (Nutrients and pH) attributes in the soil. newlineThe existing soil nutrient prediction using LIBS data combined with newlinemachine learning algorithm faces lot of uncertainties as quoted in the newlineliteratures. In this research, it is proposed to use the soil surface reflection newlinecharacteristics to minimize the uncertainty in LIBS data. A black slit newlinearrangement is used to eliminate the plasma reflection from the soil surface. newlineThe correlation between the soil nutrient concentration obtained from newlinestandard test lab and LIBS elemental peak get enhanced in data obtained with newlineslit arrangement. The Pearson correlation coefficient obtained for Nitrogen is newliner=0.4 (without slit) and r=0.65 (with slit). Similarly, the machine learning newlinealgorithm s predictive ability is also get enhanced after slit arrangement. The newlineCoefficient of Determination (COD) and Root Mean Square Error are the newlineevaluation metrics. newline