Ross S. Bricklemyer1, David J. Brown1, Sam M. Clegg2, and James E. Barefield2. (1) Washington State University, Dept. of Crop and Soil Sciences, 201 Johnson Hall, Pullman, WA 99164-6420, (2) Advanced Diagnostics and Instrumentation Group, Los Alamos National Laboratory, P.O. Box 1663, MS J565, Los Alamos, NM 87545
There is growing need for rapid, accurate, and inexpensive methods to measure, and verify soil organic carbon (SOC) change for national greenhouse gas accounting and the development of a soil carbon trading market. Visible and Near Infrared Spectroscopy (VisNIR) and Laser Induced Breakdown Spectroscopy (LIBS) techniques have the potential to fill that need. LIBS and VisNIR are fundamentally different and complementary technologies. LIBS provides precise elemental analysis of soils, but cannot distinguish between organic C and inorganic C. VisNIR has been used to characterize soil mineral and molecular composition and can distinguish between organic C and inorganic C. In this study we evaluated the accuracy and precision of a simulated
in situ VisNIR/LIBS proximal sensing array for soil profile carbon measurement.
We simulated the in situ system by scanning 120 intact soil cores (3.8 cm x 50 cm) from three agricultural fields in north central Montana, USA. VisNIR measurements were taken using an Analytical Spectral Devices (ASD, Boulder, CO, USA) Agrispec® spectrometer. LIBS measurements involved focusing a Nd:YAG laser onto the surface of the soil core to produce a plasma. The spectra from emitted plasma light was detected with a charge-coupled device. LIBS and VisNIR measurements were taken concurrently in 2.5 cm increments along each intact soil core. Subsamples of soil (~ 25g) were taken from interrogation points, and then each core was cut into 10 cm increments for carbon analysis. All samples were analyzed for SOC and SIC using standard methods.
A representative subset (~600 samples) of the interrogation point samples were used to calibrate and validate several VisNIR/LIBS models to 1) compare VisNIR /LIBS vs. standard carbon analysis and 2) examine the transferability VisNIR/LIBS measurements to a larger volume of soil. We constructed local, regional and hybrid calibration models for SOC and SIC using partial least squares regression on several transformations of the VisNIR and LIBS spectra (ex. 1st derivative, absorbance).
