Location: Forage and Livestock Production Unit
Title: Predicting soil bulk density of Oklahoma paddocks with near infrared relectance spectroscopy Authors
|Daniel, John -|
Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: June 1, 2012
Publication Date: N/A
Interpretive Summary: Soils found in native range and croplands have multiple indicators of both physical and biological condition. Bulk density is an important indicator of soil physical condition, and how applied management may impact soils. Accurately describing bulk density can be difficult. The techniques used to measure bulk density require large amounts of manpower and time. Further, bulk density is a complex property involving physical, biological and chemical features of soils, and its distribution is highly variable within landscapes. Such difficulties point to the need for more-timely and less costly estimates. We undertook a study to describe the capacity of near infrared reflectance spectroscopy (NIRS), to provide estimates of soil bulk density in central Oklahoma. Near infrared reflectance spectroscopy measures which infrared light wavelengths are reflected by samples, and relates the reflected wavelengths to chemical bonds in the sample. Soils included in the experiment were collected from one pasture of winter wheat used for grazing and three pastures of native prairie managed under different grazing regimes, during 1978 to 2004. We collected samples from the top 10 inches of soil along an elevation gradient common to all four pastures. The samples were processed, analyzed by traditional laboratory techniques, and scanned by reflectance spectrophotometer to develop data required for NIRS analyses. We then tested the capacity of an equation developed by NIRS to describe bulk density based on reflected wavelengths. The study showed the equation, based on laboratory measured density and reflected wavelengths, functioned adequately and defined 66% of the information related to bulk density. A more accurate equation will require the addition of samples from a wider range of landscapes, soil types, and management.
Technical Abstract: Bulk density (BD) is an important indicator of soil physical condition and potential impacts of management. Accurate description of bulk density can be difficult due to variable distribution patterns, and large sample numbers required for accurate estimates. Near infrared reflectance spectroscopy (NIRS) has potential to provide faster, accurate, and lower cost estimates. This study defined the capacity of NIRS to predict BD of soils in rangeland (n=3) and conventionally tilled wheat (n=1) paddocks in Oklahoma, under different forms of long-term (1978 to 2004) management. Soil samples were collected along 150 m transects in each paddock, situated between a ridge and toe slope. The A horizon was divided into three sections (0-5, 5-10, and 10-25 cm) and, reflectance (R) measurements were collected for all samples (n=1200). A randomly selected subset (n=220) of absorption (log1/R) spectrom in the 750-2500nm wavelength range were used to develop a calibration equation relating log 1/R in a given wavelength to measured BD. The calibration equation was tested with a validation set of observations. The relationship between laboratory values and NIRS estimates (n=84) generated by the calibration equation was significant (R^2 =0.78; P<0.01), and the slope approximated unity (slope=0.97). Application of the calibration equation to the validation dataset (n=145) generated a different slope (0.73), and a lesser but significant relationship (R^2=0.66; P<0.01). Results suggest the calibration equation could provide useful predictions of bulk density for routine determination of soil response to management. A more accurate equation will require samples from a wider range of landscape positions, soil types, and management.