Visible and near infrared reflectance spectroscopy to determine chemical properties of paddy soils

Abstract

The ability of obtaining soil properties estimations from time and cost efficient remotely sensed techniques has been identified as a valuable technique with great demand for larger amounts of good quality soil data to be used in environmental monitoring, modelling and precision agriculture. Visible (Vis) and Near Infrared (NIR) spectroscopy provides a good alternative that may be used to enhance or replace conventional methods of soil analysis. This study site comprised of 118 plots (142 ha) of paddy fields in the Tanjung Karang Rice Irrigation Scheme, Malaysia. The aim of this paper is to evaluate the abilities of Vis (350-700 nm) and NIR (700-2500 nm) regions for prediction of selected soil chemical properties in Malaysian paddy soils. Savitzky-Golay algorithm was implemented as spectral pre-processing and applied Stepwise Multiple Linear Regression (SMLR) to construct calibration models. The soil properties examined in this study were soil pH, electrical conductivity (EC), organic carbon (OC), cation exchange capacity (CEC), total nitrogen (N), available phosphorus (P) and exchangeable potassium (K). All the soil samples tested in this study were shown to have similar reflectance spectra and greater numbers of reflectance peaks in the NIR region especially around λ= 1150, λ= 1650 and 2200 nm). The study also revealed the accuracy of SMLR prediction in each of the Vis and NIR spectral regions. The NIR produced more accurate predictions for most of the measured soil properties; however, higher significant correlation was obtained using the Vis for EC and available P. This work demonstrated that Vis and NIR spectroscopy can be considered as a good tool to assess soil chemical properties in Malaysian paddy fields.