Natural Resources, School of


First Advisor

Paul Hanson

Date of this Version



A THESIS Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Master of Science, Major: Natural Resource Sciences, Under the Supervision of Professor Paul Hanson. Lincoln, Nebraska : August, 2018

Copyright (c) 2018 Joshua R. Gates


Soil plays an important role in our daily lives, namely producing food, cleaning water and storing carbon. The ability to rapidly and cost-effectively quantify the various components of soils can help us understand and better manage this important resource. This study aims to compare the ability of visible near-infrared (VNIR) spectroscopy and mid-infrared (MIR) spectroscopy to quickly and accurately predict various important soil properties (electrical conductivity, soil pH, cation exchange capacity, exchangeable cations, phosphorus, carbon, beta-glucosidase enzyme activity and nitrogen). Prediction models were developed using partial least squares regression (PLSR) techniques. Three different calibration sampling methods were tested along with various spectral preprocessing techniques to find the best predictive ability of VNIR and MIR. Soil components related to carbon, nitrogen, and cation exchange capacity had good predictive ability (R2 > 0.8) by both VNIR and MIR, but MIR was more accurate. Electrical conductivity, sodium cations, and phosphorus were poorly predicted by both (<0.71). VNIR models were not as robust as MIR models but could be potentially useful for qualitative analyses when rapid analyses are preferred over methods are more accurate. MIR predictions overall yielded more accurate predictions than VNIR and could potentially be used as a surrogate method for timely laboratory techniques for spectrally active soil components.

Advisor: Paul Hanson