Skip to main content
ARS Home » Research » Publications at this Location » Publication #228944
Title: Specular Reflection and Diffuse Reflectance Spectroscopy of Soils

Author
item Reeves Iii, James
item Francis, Barry
item SMITH, S - MICHIGAN STATE UNIV

Submitted to: Electronic Publication
Publication Type: Other
Publication Acceptance Date: 7/21/2008
Publication Date: 7/21/2008
Citation: Reeves III, J.B., Francis, B.A., Smith, S.K. 2008. Specular Reflection and Diffuse Reflectance Spectroscopy of Soils. Electronic Publication.

Interpretive Summary: Near-infrared spectroscopy has been used for decades for the rapid and inexpensive analysis of forages and other agricultural products. More recently this method has been applied to the analysis of soils both for general analysis and specifically for carbon. Until the last decade and a half, it was believed that mid-infrared spectroscopy could not be used for similar analysis without intensive sample preparation, something not needed for near-infrared. This was due to the fact that mid-infrared spectra of ground, but otherwise, untreated materials can be distorted due to a mirror like reflection effect called specular reflection. With the advent of studies showing that mid-infrared spectroscopy could be used to accuracy analyze ground, but otherwise, untreated samples in the same manner as near-infrared, the question of what the effects of the specular reflection really are arises. Results of these studies on the occurrence and effects of specular reflection in mid-infrared spectra of soils have shown that distortions due to specular reflection occur for both organic (humic acid) and non-organic fractions (carbonates, silica, ashed fraction of soil). The results demonstrated explain why the spectrum of calcium carbonate in limed soils did not match that of spectra seen in many books, etc., and also offer an explanation as to why the presence of inorganic-carbon interferes with the development of mid-infrared calibrations for organic-carbon in soil, but not for near-infrared calibrations and may have far reaching implications for the use of mid-infrared spectra for quantitative and qualitative analysis of soils. For example, commercial libraries of spectra of minerals collected by means other than diffuse reflectance would be largely useless for comparing mineral spectra to soil spectra. Finally, with forages and grains, it has been found necessary to develop separate calibrations for different products in order to obtain the best results. With soils, this has not seem to be a problem for many diverse sets of samples where the carbon contents were in a range of 0 to 5%. Mid-infrared calibrations have also appeared to be more robust than the corresponding near-infrared calibrations in that fewer outliers. However, the results discussed here may indicate that at least for some soil types (e.g., really different mineralogy or carbon contents) separate calibrations or the use of non-linear calibrations methods will be necessary.

Technical Abstract: Studies on the occurrence and effects of specular reflection in mid-infrared spectra of soils have shown that distortions due to specular reflection occur for both organic (humic acid) and non-organic fractions (carbonates, silica, ashed fraction of soil). The results demonstrated explain why the spectrum of CaCO3 in limed soils did not match that of spectra seen in many books, etc., and also offer an explanation as to why the presence of inorganic-C interferes with the development of mid-infrared calibrations for organic-C in soil, but not for near-infrared calibrations and may have far reaching implications for the use of mid-infrared spectra for quantitative and qualitative analysis of soils. For example, commercial libraries of spectra of minerals collected by means other than diffuse reflectance would be largely useless for comparing mineral spectra to soil spectra. Finally, with forages and grains it has been found necessary to develop separate calibrations for different products in order to obtain the best results. With soils, this has not seem to be a problem for many diverse sets of samples where the C contents were in a range of 0 to 5%. Mid-infrared calibrations have also appeared to be more robust than the corresponding near-infrared calibrations in that fewer outliers. However, the results discussed here may indicate that at least for some soil types (e.g., really different mineralogy or C contents) separate calibrations or the use of non-linear PLS or other non-linear calibrations methods will be necessary.