Quantifying soil particulate organic matter using image analysis

Nowicki, Michael Joseph

Permalink

https://hdl.handle.net/2142/129648 Copy

Description

Title

Quantifying soil particulate organic matter using image analysis

Author(s)

Nowicki, Michael Joseph

Issue Date

2024-07-16

Director of Research (if dissertation) or Advisor (if thesis)

Wander, Michelle M

Committee Member(s)

• Grift, Tony
• Ugarte, Carmen

Department of Study

Natural Res & Env Sci

Discipline

Natural Res & Env Sciences

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Particulate Organic Matter
• Image Analysis
• Soil Analysis

Abstract

Rapid and cost-effective quantification of soil organic carbon (SOC) fractions that are responsive to management will improve soil health assessment and associated decision support tools. The particulate organic matter (POM) fraction of SOC is of interest due to its sensitivity to management and significant contributions to soil structure and biological integrity. Conventional laboratory methods for assessing POM include size and density fractionation, processes that are laborious and time-consuming. This study evaluated the potential for a rapid image-based POM quantification method. This novel approach only requires lightly prepared samples to be saturated then placed in a table-top imaging box, where a series of images are taken. Preliminary tests were performed on a series of assembled soil samples with POM contents ranging from approximately 6.5-50 g/kg soil. A series of images enabled differentiation of POM from surrounding microaggregates using image analysis. Illumination was provided by Light Emitting Diodes (LED) spanning limited sections of visible to near-infrared (Vis-NIR) spectrum. Estimates of POM area were regressed against known concentrations. Initial regressions were performed on separate POM types and size fractions. Grass residue (POMG) model fits of R2 = 0.83, 0.91 and 0.90 for estimates POM< 0.75 mm, 0.5 to 1 mm, and 1.0 - 2.0 mm. Broadleaf POM (POMB) estimates R2 = 0.90, 0.89 and 0.69 for size fractions 0.1 - 0.5 mm, 0.5 - 1.0 mm and 1.0 - 2.0 mm. This study found the predictive power decreased with the inclusion of multiple POM size fractions, with model fits of R2 = 0.53 for grass, and R2 = 0.38 for broadleaf. A regression combining broadleaf and grass lowered predictive strength further with R2 = 0.31. Despite a decline in fit for combined size sample estimates, there remained a promising capacity to reproduce estimates for size medium (average particles) for both POMG and POMB fractions.

Graduation Semester

2024-08

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/129648

Copyright and License Information

Copyright 2024 Michael Nowicki

Owning Collections