Effect of wavelength range on the measurement accuracy of some selected soil constituents using visual-near infrared spectroscopy

The wavelength range is one of the main components affecting the measurement accuracy with visible (vis) and near infrared (NIR) spectroscopy. The performance of two commercially-available spectrophotometers with different wavelength ranges for measurement of selected soil attributes was evaluated. The two spectrophotometers considered were a diode array with a short wavelength range (SWR) of 300-1700nm and a combination of diode array and scanning monochromator with a full wavelength range (FWR) of 350-2500nm. Scanning was performed on wet-fresh (in situ conditions) and dry-processed (laboratory conditions) soil samples, to estimate the effect of moisture content on the performance of the two spectrophotometers. Partial least squares (PLS) regression with the leave-one-out cross validation technique was used to develop calibration models relating soil spectra with chemical attributes. Results showed that under wet field soil conditions pH, available phosphorus (P), cation exchange capacity (CEC), potassium (K) and calcium (Ca) were more accurately predicted with the SWR, whereas total nitrogen (N), total carbon (C), sodium (Na) and magnesium (Mg) were better predicted with the FWR. The significant effect of the water absorption at a wavelength of 1950 nm which created considerable changes in the shape of spectra and subsequently loss of important information available at wavelengths greater than 2000 nm, reduced the efficiency of the FWR instrument. The dry soil conditions led to a superior performance of the FWR instrument over the SWR instrument for measurement of all soil attributes. This suggests that a greater wavelength range than 1700 nm is not useful for improving the measurement accuracy of soil attributes (except N and C) with vis-NIR spectroscopy when measurement is to be carried out in the field under wet conditions.