Visible-near infrared reflectance spectroscopy for assessment of soil properties in a semi-arid area of Turkey

摘要: Abstract Reflectance spectroscopy can be used to nondestructively characterize materials for a wide range of applications. In this study, visible-near infrared reflectance (VNIR) was evaluated prediction diverse soil properties related four different series the Entisol group within single field in northern Turkey. Soil samples were collected from 512 locations 25 × 25 m sampling grid over 32 ha (800 × 400 m) area. Air-dried scanned at 1 nm resolution 350 2500 nm, and calibrations between physical chemical spectra developed using cross-validation under partial least squares regression (PLSR) multivariate adaptive splines (MARS). Raw first derivative data separately combined all set. Data additionally divided into two random subsets 70 30% full data, which each calibration validation. Overall, MARS provided better predictions when cross-validation. However, PLSR results comparable terms accuracy separate sets No improvement obtained by combining raw data. Strongest correlations with exchangeable Ca Mg, cation exchange capacity, organic matter, clay, sand, CaCO 3 contents. When classified groups, VNIR estimated class memberships well, especially texture. conclusion, variably successful estimating scale, showed potential substituting laboratory analyses or providing inexpensive co-variable