TITLE:
Capability of Visible-Near Infrared Spectroscopy in Estimating Soils Carbon, Potassium and Phosphorus
AUTHORS:
Bakamba dite Djeneba Sacko, Souleymane Sanogo, Hamidou Konare, Abdramane Ba, Tiéman Diakite
KEYWORDS:
VIS-NIR Spectroscopy, Soils Chemical Properties, Principal Component Regression (PCR), Partial Least Square Regression (PLSR)
JOURNAL NAME:
Optics and Photonics Journal,
Vol.8 No.5,
May
31,
2018
ABSTRACT: The spectroscopy technique has many advantages over conventional analytical methods since it is fast and easy to implement and with no use of chemical extractants. The objective of this study is to quantify soil total Carbon (C), available Phosphorus (P) and exchangeable potassium (K) using VIS-NIR reflectance spectroscopy. A total of 877 soils samples were collected in various agricultural fields in Mali. Multivariate analysis was applied to the recorded soils spectra to estimate the soil chemical properties. Results reveal the over performance of the Principal Component Regression (PCR) compared to the Partial Least Square Regression (PLSR). For coefficient of determination (R2), PLSR accounts for 0.29, 0.42 and 0.57; while the PCR gave 0.17, 0.34 and 0.50, respectively for C, P and K. Nevertheless, this study demonstrates the potential of the VIS-NIR reflectance spectroscopy in analyzing the soils chemical properties.