Prediction of soil carbon fractions using a field spectroradiometer equipped with an illuminating contact probe

Authors: Fortuna, Ann Marie; Starks, Patrick; Nelson, Amanda; STEINER, JEAN - Retired ARS Employee

Submitted to: Soil Systems
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/24/2019
Publication Date: 10/28/2019
Citation: Fortuna, A., Starks, P.J., Nelson, A.M., Steiner, J.L. 2019. Prediction of soil carbon fractions using a field spectroradiometer equipped with an illuminating contact probe. Soil Systems. 3(4):71. https://doi.org/10.3390/soilsystems3040071.
DOI: https://doi.org/10.3390/soilsystems3040071

Interpretive Summary: The definition of soil health is the capacity of a soil to function within ecosystem boundaries to sustain biological productivity, maintain environmental quality, and promote plant and animal health and is measured using a suite of chemical, physical and biological indicators. We chose to measure several soil health indicators that included total soil carbon and nitrogen (TSOC, TSN), particulate organic matter C and N (POMC, POMN) and acid resistant C (RCAH) that make up portions of total soil organic carbon. Because these measurements are time consuming, and in some instances hazardous, we explored the use of soil reflectance measurements, acquired using a portable field spectroradiometer, to predict concentrations of TSOC, TSN, POMC, POMN, and RACH. The study site is located at the USDA-ARS Grazinglands Research Laboratory El Reno, OK and is comprised of eight 1.6 ha watershed treatments which includes paddocks that are representative of native warm season grasslands and winter wheat (Triticum aestivum) managed by either conventional or no-till systems. A baseline set of soil samples was taken at 0-5, 5-15 and 15-30 cm depths. Laboratory measurements included TSOC, TSN, POMC, POMN, and RCAH. Soil reflectance of these soil samples was measured in the 350 to 2500 nm region of the electromagnetic spectrum using a portable field spectroradiometer. The soil spectra were matched to the relevant laboratory measurements and divided into model development (70% of the data) and model validation (30% of the data) data sets. The calibrated models were applied to the validation data sets and the statistical analysis revealed that the prediction efficiencies of the soil reflectance-based models were highly quantitative. The findings from this research suggest that: 1) field spectroradiometry was useful in parameterizing the spatial variability of soils at our study site, 2) that spectroradiometry was useful in identifying changes in C and N fractions due to effects of land management, and 3) that soil reflectance measurements have the potential to reduce the time, effort, and cost of acquiring measurements of TSOC, TSN, POMC, POMN, and RCAH.

Technical Abstract: The inherent heterogeneity of soil and added variation resulting from soil sampling makes in-situ measurements of edaphic soil properties highly desirable. This research compares the accuracy of laboratory reference measurements of soil carbon (C) and nitrogen (N) fractions, indicators of soil health, to that determined from soil reflectance spectra acquired using a portable field spectroradiameter fitted with an illuminating contact probe. The study site is located at the USDA-ARS Grazinglands Research Laboratory El Reno, OK and is comprised of eight 1.6 ha watershed treatments which includes paddocks that are representative of native warm season grasslands and winter wheat (Triticum aestivum) managed by either conventional or no-till systems. We are currently establishing a soil health baseline on the study site that will integrate a diversified adaptive crop livestock system. Landform complexes serve as replicates within and among 1.6 ha sized watershed treatments that serve as paddocks (3-4% slope, westerly exposure) allowing us to monitor biological, chemical and physical indicators of soil health. A baseline set of soil samples was taken at 0-5, 5-15 and 15-30 cm depths. Measurements included total soil organic carbon (TSOC), total soil nitrogen (TSN), residual C of acid hydrolysis (RCAH) and particulate organic matter C (POMC) and N (POMN). Soil reflectance of these soil samples was measured in the 350 to 2500 nm region of the electromagnetic spectrum using a portable field spectroradiometer. The soil spectra were matched to the relevant laboratory measurements and divided into model development (70% of the data) and model validation (30% of the data) data sets. The calibrated models were applied to the validation data sets and the statistical analysis revealed that the prediction efficiencies of the soil reflectance-based models were highly quantitative. Coefficients of determination (R2) near 1 and ratios of predicted values to the measured standard deviation (RPD) greater than 2 indicate good predictive models. For our study, TSOC for whole and ground (R2 = 0.98, RPD = 7.6 and 0.94, 3.3 respectively); TSN for whole and ground soil (R2 = 0.92, RPD = 3.3 and 0.91, 3.3 respectively); RCAH (R2 = 0.91, RPD = 3.23 respectively) and POMC (R2 = 0.90, RPD = 2.70 respectively) and POMN (R2 = 0.94, RPD = 3.98 respectively). The findings from this research suggest that: 1) field spectroradiometry was useful in parameterizing the spatial variability of soils at our study site, 2) that spectroradiometry was useful in identifying changes in C and N fractions due to effects of land management, and 3) that soil reflectance measurements have the potential to reduce the time, effort, and cost of acquiring measurements of TSOC, TSN, POMC, POMN, and RCAH.