160-8 Prediction of Particle Size Distribution with Visible Near-Infrared Spectroscopy.

Poster Number 1527

See more from this Division: SSSA Division: Soil Physics and Hydrology
See more from this Session: Environmental Soil Physics and Hydrology Student Competition: Lightning Orals with Posters: II

Monday, November 16, 2015
Minneapolis Convention Center, Exhibit Hall BC

Cecilie Hermansen1, Maria Knadel2, Per Moldrup3, Mogens H. Greve4, Dan K. Marning4 and Lis W. de Jonge2, (1)Dept. of Agroecology, Aarhus University, Tjele, DENMARK
(2)Agroecology, Aarhus University, Tjele, Denmark
(3)Civil Engineering, Aalborg University, Aalborg, Denmark
(4)Dept. of Agroecology, Aarhus University, Tjele, Denmark
Abstract:
The Rosin-Rammler distribution function is a two-parameter (α and β) function which can be used to describe the particle size distribution (PSD) curve of the soil mineral fractions. The α-parameter is the particle size corresponding to the 63.2 percentile of the cumulative probability distribution when expressed as a log-normal distribution. The β-parameter is related to the shape of the PSD. Knowledge of the PSD is important for evaluating soil quality and functional properties for determining transport of water, gas and nutrients in soil. However, detailed analysis of the PSD is time consuming and expensive.

Visible near-infrared reflectance spectroscopy (vis-NIRS) is a fast and non-destructive method for soil analysis. Absorptions in specific wavelength intervals of this spectral range are characteristics of organic and inorganic matter.

In this study vis-NIRS covering the spectral range between 400 and 2500 nm, was used to predict the Rosin-Rammler model parameters with the aim to indirectly predict the full PSD curve. Additionally, vis-NIRS was applied to determine model parameters (γ and S) of a newly developed and error-function based PSD model which assumes log-normally distributed data. The γ and S are related to the shape of the PSD.

A total of 349 soil samples from seven differently-textured Danish fields (clay: 0.03-0.45 kg kg-1, organic carbon: 0.011-0.084 kg kg-1), were collected, analysed for PSD, and scanned with a vis-NIRS spectrophotometer (DS2500, Foss, Hillerød, Denmark). The parameters of the two PSD models were correlated to spectral data with partial least squares regression analysis. The samples were randomly divided into independent calibration and validation sets. Based on the results, the vis-NIRS showed very successful predictive abilities of α, β, γ and S for the PSD across the full texture range.

See more from this Division: SSSA Division: Soil Physics and Hydrology
See more from this Session: Environmental Soil Physics and Hydrology Student Competition: Lightning Orals with Posters: II