Comparison of Ultraviolet, Visible, and Near Infrared Sensing for Soil Phosphorus

Identifying the most suitable electromagnetic region is crucial in designing a spectral-based sensor for determining phosphorus (P) concentration. The reflectance of soils in ultraviolet, visible, and near infrared regions was measured. For each spectral region, calibration and validation data sets were obtained using simple randomised sampling. The reflectance and P concentration of soils were used in linear partial least-squares regression (PLS) analysis to predict P concentration of soils. Soils were analysed for water-soluble P, Mehlich-1 P, and total P. Partial least-squares regression results showed that strong relationships were found between reflectance and P concentration with coefficients of determination of 0·93, 0·95 and 0·76 for total, Mehlich-1 and water-soluble P, respectively, in the near infrared region. Good relationships were observed between reflectance and P concentration with coefficients of determination of 0·83, 0·67 and 0·61 for total, Mehlich-1 and water-soluble P, respectively, in the visible region. However, weaker relationships between reflectance and P concentration were observed in the ultraviolet region.