Abstract
Models used to evaluate leaching of contaminants to groundwater are very sensitive to sorption coefficients (Kd). These models need reliable Kd data at the field scale, but the number of samples required makes the classic batch sorption experiments inappropriate for this purpose. Since visible–near infrared (vis–NIR) spectroscopy is an inexpensive and fast method, it has been used for predicting soil properties related to soil sorption capacity. In this study, we aimed to predict the spatial variation of Kd from vis–NIR spectra for two contaminants: phenanthrene (sorbed on organic fractions) and glyphosate (sorbed on mineral fractions). Forty-five bulk soil samples were collected from an agricultural field in Estrup, Denmark, in a 15 m × 15 m grid. Samples were air-dried, sieved to 2 mm and analysed for selected soil properties. Sorption coefficients were obtained from a batch equilibration experiment. Soil samples were measured with a bench-top spectrometer covering the vis–NIR range between 400 nm and 2500 nm. Partial least squares regression with full cross-validation was used to correlate the soil spectra with Kd values and soil properties. The sorption coefficients ranged from 345 L kg−1 to 886 L kg−1 and from 162 L kg−1 to 536 L kg−1 for phenanthrene and glyphosate, respectively. The regression coefficients showed that phenanthrene sorption was correlated with total organic carbon, aluminium oxides and cation exchange capacity, and glyphosate sorption with clay minerals and iron oxides. By means of the vis–NIR spectra we were able to predict phenanthrene (R2 = 0.95, RMSECV = 31 L kg−1) and glyphosate (R2 = 0.79, RMSECV = 45 L kg−1) sorption capacities. A model using vis–NIR spectra plus pH values improved the prediction of glyphosate sorption capacity (R2 = 0.88, RMSECV = 34 L kg−1). The models obtained from vis–NIR spectra successfully predicted Kd within the investigated field, indicating the potential of vis–NIR spectroscopy as a fast method for determining Kd for input to leaching risk assessment models. However, further studies of different soil types and geographical scales are needed to confirm our findings.
© 2016 The Author(s)
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription
Add to BibSonomy