Soil macronutrient detection based on visible and near-infrared absorption spectroscopy

Abstract

Precision agriculture using cost-effective soil fertility measurement is important to obtain adequate quality and quantity of crops. Modern agriculture uses soil spectroscopy, which is a fast, cost-effective, environmentally friendly and reusable method. Soil fertility is used in modern agriculture to sustain plant growth and optimize crop yield. However, most existing light sources and computerized photodetection modules in soil spectroscopy are bulky in size, consume high power and expensive such as tungsten-halogen lamps, deuterium lamps and commercial spectrometer. This thesis proposes an improved experimental module based on absorbance spectroscopy to determine the nitrogen (N), phosphorus (P) and potassium (K) in various soil samples which are extracted using colour-developing reagent. The experimental module consisting light-emitting diode (LED) in visible and near-infrared range, and integrated passively quenched silicon photodiode. The optical absorption of various soil samples, including agricultural and non-agricultural soils are experimentally investigated in absorbance mode using an optimal wavelength range of 467 nm until 741 nm. Beer-Lambert Law (BLL) is applied to identify the relationship between the nutrient concentration and the amount of absorbed light. At a wavelength (λ) of 467 nm, N gives a coefficient of determination (R2) between 0.49 and 0.63 for agriculture soil samples. Meanwhile, R2 of agricultural soils for K gives a value from 0.54 to 0.73. At λ = 741nm, P produces R2 in the range of 0.47 to 0.82. Furthermore, research findings using LED and photodiode follow the BLL. BLL states that high concentration has many chemical absorbing species which will lower the transmitted light intensity and give low output voltage. This study has shown that absorbance spectroscopy with proposed LED and photodiode modules are able to distinguish the nutrient concentration in the soil.