|Reeves iii, James|
Submitted to: Spectroscopy Letters
Publication Type: Peer reviewed journal
Publication Acceptance Date: 5/16/2005
Publication Date: 12/1/2005
Citation: Madari, B.E., Reeves III, J.B., Coelho, M.R., Machado, P., De-Polli, H., Coelho, R.M., Benites, V.M., Souza, L.F., McCarty, G.W. 2005. Mid- and near- infrared spectroscopic determination of carbon in a diverse set of soils from the brazilian national soil collection. Spectroscopy Letters. 38(6):721-740. Interpretive Summary: The use of light in the near- and mid-infrared spectral ranges has come to be used to rapidly determine the composition of many organic containing materials including, more recently, soils. However, many questions still remain as to how to relate the spectral information to the measure of interest, e.g., carbon content in soils (Calibration development). For example, only limited ranges in soil types and composition have often been examined. Soils in the United States often contain carbonate minerals, but those from the tropics do not. The objective of this study was to examine a wide range of soils representative of those found in Brazil. Calibrations were developed for the determination of carbon as determined by two standard methods: oxidation or combustion, the method of choice today, and by chromate oxidation (Walkley-Black carbon), an older method. Results using mid-infrared spectra were found to be superior to those based on near-infrared spectra especially for the chromate oxidation derived carbon. More importantly, all results indicated that the best calibrations result when limited ranges in carbon content are considered but that calibration development based on other soil properties or soil groupings, like texture, mineralogy, or soil type, is going to be necessary in order to obtain the most accurate calibrations. Finally, comparisons to results obtained with similar soils based on carbon content indicate that the presence of carbonates can cause serious problems with calibration development originally attributed to the range of carbon present.
Technical Abstract: Calibrations for soil carbon content measured by combustion (total carbon, TC) and chromate oxidation (Walkley-Black carbon, WBC) based on 1135 and 1014 samples respectively, from the Brazilian National Soil Collection were made using Fourier transform near (NIRS) and mid-infrared diffuse reflectance (DRIFTS) spectroscopy. Calibration sets of sample populations of different carbon ranges were established. For TC ranges between 0.4 ' 555.0, 0.4-99.1, and 0.4-39.9 g kg-1, and for WBC 0.2-401.0, 0.2-66.0, and 0.2-30.0 g kg-1. Good calibrations were obtained for the largest carbon ranges (R2=0.95 and 0.93 using DRIFTS for TC and WBC respectively), but lower accuracy was found for the narrower carbon ranges (R2<0.85 and 0.90 for TC and WBC respectively). The RMSD (root mean squared deviation between actual and predicted) values, however, were smaller for the lower carbon ranges (15.38, 5.87, and 4.18 for the large, narrower and narrowest TC range, and 10.88, 4.24, 2.62 for the large, narrower and narrowest WBC range using DRIFTS). DRIFTS always gave better calibrations than NIRS, except for the high TC samples when these were examined separately from the other samples (Histosols and Spososols). These results demonstrated that while calibrations can be developed using either DRIFTS or NIRS for even a very diverse set of soil samples, which will determine C over a wide range of concentrations inherent in such a diverse set, such calibrations do not appear to be the best procedure to use. Developing calibrations for ranges of soil C content did not enhance R2 of calibrations. The lower the TC or WBC range, the lower the R2 obtained, however, the error of the calibrations (RMSD/Mean TC or WBC) also decreased. Results indicate that calibration development based on other soil properties or soil groupings, like texture, mineralogy, or soil type are necessary to use in order to obtain the most accurate calibrations.