Subramanian, Senthil Kumar (Department of Crop & Soil Sciences, Michigan State University, 520A, Dept. of Crop & Soil Sciences, PSSB, Michigan St. University, East Lansing, MI, 48823; Phone: 517-980-2741; Fax: 517-355-0270; Email: subram20@msu.edu)
 
Quantifying soil carbon using NIR spectroscopy: A study on its consistency at different soil carbon levels
 

S.K. Subramanian*, A.N. Kravchenko, X. Huang, J. QI

Quantitative assessment of soil carbon has gained increased interest in recent years, since such information helps farmers and land managers in developing strategies for increasing soil sustainability and productivity and helps researchers to find out the best land use practices that sequester more carbon per land unit. Deriving such information through standard procedures of soil sampling and laboratory analyses is expensive, time-consuming, and labor-intensive. The recently developed methods of Near-Infrared (NIR) spectroscopy have potential in augmenting the current procedures by reducing the time and the cost involved. However, it has been reported that accuracy of soil carbon prediction by NIR might vary at different levels of soil carbon. The objective of our study is, thus, to assess the consistency of soil carbon predictions using NIR under different soil carbon levels. A set of 1,200 samples has been collected from the Long Term Ecological Research site (LTER) at the Kellog Biological Station, MI. The samples were analyzed for total carbon/nitrogen contents and spectral measurements for each sample was collected in the laboratory conditions with analytical spectral device (ASD). First, we used a portion of the samples to test the statistical significance of employing NIR in soil carbon prediction across all observed levels of carbon (from <0.8 to >3.0 %) based on principal component and regression analyses, and found the regression model to be significant (r2=0.82). Then, we segregated the samples into different groups based upon the soil carbon values (e.g., <1.0%, 1.0-1.5%, 1.5-2.0%, 2.0-2.5%, 2.5-3.0%, >3.0%). The comparisons of how accurate the NIR predictions are in every soil C group will be presented and discussed. The results will help us in assessing the reliability of NIR spectroscopy in soil carbon prediction and thereby will provide the researchers and decision makers with an idea on how far such predictions can be relied on.