68-16 Comparison of EPO and DS in Removing the Moisture Effect from Soil Visnir Reflectance Spectra.

See more from this Division: SSSA Division: Pedology
See more from this Session: Pedology: I (includes student competition)

Monday, November 16, 2015: 3:30 PM
Minneapolis Convention Center, L100 E

Nuwan K Wijewardane, Biological Systems Engineering, University of Nebraska Lincoln, Lincoln, NE, Yufeng Ge, Biological and Agricultural Engineering, University of Nebraska-Linclon, Lincoln, NE, Cristine L. S. Morgan, MS 2474 TAMU, Texas A&M University, College Station, TX and Michael Pearson, NSSC KSSL, USDA NRCS, Lincoln, NE
Abstract:
Visible and near infrared (VisNIR) reflectance spectroscopy is a useful tool in quantifying properties of dry ground soil such as clay
and organic carbon under laboratory conditions. One of the main reasons that hinder its potential application under the field condition
is the presence of soil moisture. The prediction performance of models developed from dry ground soil is greatly reduced when they are
applied to field moist samples. Researchers have suggested External Parameter Orthogonalization (EPO) as a promising technique to
remove the effect of moisture from soil VisNIR spectra to enhance the predictive power of models developed from dry ground soil
spectra. A second feasible method for moisture effect removal, known as Direct Standardization (DS), has also been suggested recently.
The objective of this research was to investigate and compare EPO and DS in removing the moisture effect from soil VisNIR spectra and
then to predict seven different soil properties (clay, sand, total carbon, organic carbon, inorganic, phosphorous, and pH). One hundred
and sixty seven soil samples were selected from the USDA-NRCS soil archive. They were wetted and scanned at eight different moisture
levels (33, 29, 25, 21, 17, 13, 10 and 8%). In addition, another 185 soil samples were selected from the same archive and scanned to
develop dry ground VisNIR models. Both EPO and DS were implemented for moisture effect removal, and transformed spectra were modeled
with partial least squares regression for model calibration and validation. The results showed that both EPO and DS can be used
satisfactorily to remove moisture effect and improve the predict power of dry ground models applied to moist samples.

See more from this Division: SSSA Division: Pedology
See more from this Session: Pedology: I (includes student competition)