2018 Volume 80 Issue 1 Pages 43-56
A potable spectrophotometer was used to predict 24 soil parameters based on fresh soil sample spectra collected in a sloped citrus field. One hundred soil samples were collected randomly from the field for a training set (75 samples) and a test set (25 samples). The partial least squares regression (PLSR) analysis was used to develop regression models. Evaluation was conducted based on the values of the coefficient of determination (R2) and the residue prediction deviation (RPD). The results showed that 13 soil parameters were predicted with R2>0.5 and RPD>1.4. Soil parameter maps and yield maps were developed with height change in the sloped field based on the location information collected by a real-time kinematic GPS (RTK-GPS) system.
