Pretreatment method standard normal variate (SNV) and baseline shift correction (BSC) on the NIRS-based soil spectrum for rapid prediction of soil nitrogen content

Abstract

Enhancing agricultural production quality is a crucial role of nitrogen fertilization. Generally, the utilization of nitrogen fertilizer can promote plant growth, impart a more intense green hue to leaves and stems, and significantly augment biomass production. However, excessive application of nitrogen fertilizers can result in substantial cultivation cost increase and serious environmental degradation. Hence, it’s important to regulate the nitrogen content within the soil. The primary goal of this research is to employ the NIRS technique to rapidly and eco-friendly predict soil nitrogen content. Soil samples were obtained from farmlands situated in the Aceh Besar district, with spectral data gathered within a wavelength range of 1000 – 2500 nm at 0.2 nm intervals.. NIRS and Partial Least Square (PLS) spectrum data were used to make a prediction model of the nitrogen harvesting element. The results showed that the PLS method with SNV (R=0,72) and BSC (R=0,73) pretreatment showed predictive model capabilities that were almost the same as the raw data (R=0,75), although there was a slight improvement, this suggested that nitrogen elements in Aceh Besar region have not or can use the model with the suggestion to increase the amount of data and improve the treatment of the soil harvester method so that the nitrous content elements present in the ground do not easily evaporate so as to provide better model prediction results.