Hyperspectral Spectroscopic Study of Soil Properties- A Review

Abstract

Soil analysis is required for efficient use of inputs viz. seeds, fertilizers, irrigation water and other agricultural planning. However, there are several disadvantages of soil analysis such as they are time consuming, expensive and labour intensive. Many approaches are developed to overcome these difficulties. Hyperspectral spectroscopy is emerging as a promising tool for studying soil, water and vegetation. Therefore, an attempt has been made to review the scope of using hyperspectral reflectance spectroscopy for estimation of soil properties as an alternative to traditional laboratory soil analysis methods.
 Spectral signature of soil can be used for fast and non destructive estimation of soil properties. Diffuse reflectance in 350-2500 nm range of electromagnetic spectra forms the basis of hyperspectral spectroscopy. An object is characterized by the characteristic absorptions and peaks in the electromagnetic spectra. A number of calibration techniques are applied for establishing relationship between reflectance spectra and soil properties. Multiple Linear Regression (MLR), Principal Component Regression (PCR) and Partial Least Square Regression (PLSR) are most commonly used techniques.
 MLR, PCR and PLSR are also used for prediction of several soil properties such as pH, soil organic carbon content, nitrogen, phosphorus, potassium, calcium, magnesium, sodium, iron, manganese, zinc, copper, boron, molybdenum, sand silt, clay and soil moisture. Some commonly used spectral indices are also applied for prediction of soil properties. Some of the soil physical properties viz. sand, silt and clay as well as chemical properties viz. pH and organic carbon could be estimated with good to very good prediction using pure spectra of soil. However, contrasting results of prediction of soil properties using multivariate analysis techniques have also been reported. The content of this review article will be helpful for researchers who are working on alternate methods of estimation of soil properties.