Jilin Daxue Xuebao (Gongxueban)/Journal of Jilin University (Engineering and Technology Edition) ISSN: 1671-5497 E-Publication: Online Open Access Vol: 42 Issue: 10-2023 DOI: 10.5281/zenodo.8429553 Oct 2023 | 33 EVALUATING REFLECTIVE SPECTROSCOPY FOR PREDICTING SOIL PROPERTIES IN GAJAPATI DISTRICT, ODISHA RAHUL ADHIKARY* Assistant Professor, Department of Soil Science, Centurion University of Technology and Management, Odisha, 761211, India. *Corresponding Author Email: rahul.adhikary@cutm.ac.in, ORCID ID: 0000-0003-0020-4161 SATYA NARAYAN SATAPATHY Assistant Professor, Department of Entomology, Faculty of Agricultural Sciences, Siksha ‘O’ Anusandhan Deemed to be University, Bhubaneswar, Odisha, India. BISWAJIT LENKA Assistant Professor, Department of Genetics and Plant Breeding, Faculty of Agricultural Sciences, Siksha ‘O’ Anusandhan Deemed to be University, Bhubaneswar, Odisha, India. ASIT PRASAD DASH Associate Professor, Department of Genetics and Plant Breeding, Faculty of Agricultural Sciences, Siksha ‘O’ Anusandhan Deemed to be University, Bhubaneswar, Odisha, India. RAJESH KUMAR KAR Assistant Professor, Department of Genetics and Plant Breeding, Faculty of Agricultural Sciences, Siksha ‘O’ Anusandhan Deemed to be University, Bhubaneswar, Odisha, India. SUBHAM ACHARYA Assistant Professor, Department of Agronomy, Faculty of Agricultural Sciences, Siksha ‘O’ Anusandhan Deemed to be University, Bhubaneswar, Odisha, India. Abstract Visible near-infrared spectroscopy, renowned for its non-destructive nature, rapidity, cost-efficiency, and minimal sample preparation requirements, holds promise as a substitute for in vitro techniques. This ongoing study aims to evaluate the viability of reflective spectroscopy for predicting soil properties in ion farming plains across Gajapati district Odisha. A meticulous collection of 110 soil samples from these regions formed the basis, with their core attributes established using conventional in vitro methods. Employing a land spectroscopic device, the soil samples underwent spectral analysis within the wavelength band of 240 to 400 nm. Following spectrum recording, diverse pre-processing approaches were assessed, paving the way for the application of PCA (Principal Component Analysis) and PLSR (Partial Least Squares Regression) models to decipher pivotal soil properties. The superior model choice was subsequently employed to formulate regressive functions, facilitating the prediction of targeted parameters through linear regression. Findings spotlight the precision of both PCA and PLSR models in elucidating soil properties, with the latter displaying heightened accuracy. Evaluated using the RPD (Ratio of Performance to Deviation) metric, the most accurate estimations were achieved for minerals (RPD=9.34), pH (RPD=4.45), and nitrogen (RPD>2), all classified within category A. In contrast, accuracy proved lower for variables like clay, silt, gravel, phosphorus, potassium, calcium, magnesium, and gypsum, where RPD values ranged between 0.01 and 0.28. These values collectively affirm the satisfactory precision of spectral regressive functions in forecasting the targeted foundational properties. In summary, outcomes of this study underscore the commendable precision of both PCA and PLSR models in determining crucial soil parameters. Moreover, soil spectral data emerges as an effective indirect means to estimate the physical and chemical attributes of soil. Compared to conventional laboratory methods, this technique emerges as