Managing Global Resources for a Secure Future

2017 Annual Meeting | Oct. 22-25 | Tampa, FL

Visit Crops Website

105927 Novel Pedometrics-Econometrics Modeling Using VNIR Spectroscopy: Developing Soil Carbon Sequestration Capability Index (Student Poster Content).

Poster Number 1502

See more from this Division: ASA Section: Environmental Quality
See more from this Session: Soil Carbon and Greenhouse Gas Emissions General Poster II (Students' Poster Competition)

Monday, October 23, 2017
Tampa Convention Center, East Exhibit Hall

Katsutoshi Mizuta1, Sabine Grunwald2, Christopher M Clingensmith3, Gustavo M. Vasques4, Won Suk Lee5, Michelle A. Phillips6, Wendell P. Cropper7, Xiong Xiong8 and Brenton D. Myers8, (1)University of Florida, Gainesville, FL
(2)2181 McCarty Hall, PO Box 110290, University of Florida, Gainesville, FL
(3)Soil and Water Science Department, University of Florida, Gainesville, FL
(4)Pedometria e Mapeamento Digital de Solos, EMBRAPA, Rio de Janeiro, Brazil
(5)FL, University of Florida, Gainesville, FL
(6)Warrington College of Business Administration, University of Florida, Gainesville, FL
(7)School of Forest Resources and Conservation, University of Florida, Gainesville, FL
(8)Pioneer Hi-Bred International, Inc., Johnston, IA
Abstract:
A synthesis of pedometrics and econometrics holds potential to assess soil functions, capabilities, and risk in context of soil quality, health and security. We interfaced methods commonly used in pedometrics and econometrics to develop a novel Soil Carbon Sequestration (SCseq) Capability Index (SCI).

The Data Envelopment Analysis (DEA) is one of the pedo-econometrics suited to assess soil capability or efficiency. However, the relationship between soil capability derived from the DEA with analytical lab-based soil measurements and the ones derived from visible-near-infrared (VNIR)-soil estimates is not well known. Our objective was to compare the SCI derived using SOC measured in the laboratory (SOCm) and SOC estimated using VNIR spectra (SOCe) in Florida, USA.

The SCseq rate was calculated based on both historical and current datasets that contained SOCm collected from the top soil (~20 cm depth) in 1965-1995 and 2008-2009, respectively. Some pedogenic and environmental factors relevant to soil carbon sequestration were selected by the Boruta variable selection method as well as Spearman correlations. The DEA was performed with the SCseq and the selected variables to produce the SCI scores. The VNIR spectrum was calibrated using the Partial Least Square Regression method and Random Forest (RF), after 25 different preprocessing methods were applied. The model with the highest prediction accuracy was chosen to generate the SOCe. The DEA was implemented based on these VNIR estimates to predict the SCI scores (SCI-VNIR). The prediction regression of SCI-VNIR using the RF algorithm was fitted well with the observed SCI scores as shown by the highest R2 of 0.7. This implies that the SCI scores are site-specific yet predictable using lab-based spectroscopy data. The successful estimation of the DEA-SCI scores via VNIR spectra promises future applications that assess soil quality, health, and security.

See more from this Division: ASA Section: Environmental Quality
See more from this Session: Soil Carbon and Greenhouse Gas Emissions General Poster II (Students' Poster Competition)

<< Previous Abstract | Next Abstract >>