Skip to main content
ARS Home » Plains Area » Lincoln, Nebraska » Wheat, Sorghum and Forage Research » Research » Publications at this Location » Publication #399633

Research Project: Improving Forage and Bioenergy Plants and Production Systems for the Central U.S.

Location: Wheat, Sorghum and Forage Research

Title: Multi-site evaluation of stratified and balanced sampling of soil organic carbon stocks in agricultural fields

Author
item POTASH, ERIC - University Of Illinois
item GUAN, KAIYU - University Of Illinois
item MARGENOT, ANDREW - University Of Illinois
item LEE, DK - University Of Illinois
item BOE, ARVID - South Dakota State University
item DOUGLASS, MICHAEL - University Of Illinois
item HEATON, EMILY - University Of Illinois
item JANG, CHUNHWA - University Of Illinois
item Jin, Virginia
item LI, NAN - University Of Illinois
item Mitchell, Robert - Rob
item NAMOI, NICTOR - University Of Illinois
item Schmer, Marty
item WANG, SHENG - University Of Illinois
item ZUMPF, COLLEEN - Argonne National Laboratory

Submitted to: Geoderma
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/27/2023
Publication Date: 7/28/2023
Citation: Potash, E., Guan, K., Margenot, A., Lee, D., Boe, A., Douglass, M., Heaton, E., Jang, C., Jin, V.L., Li, N., Mitchell, R., Namoi, N., Schmer, M.R., Wang, S., Zumpf, C. 2023. Multi-site evaluation of stratified and balanced sampling of soil organic carbon stocks in agricultural fields. Geoderma. 438. Article 116587. https://doi.org/10.1016/j.geoderma.2023.116587.
DOI: https://doi.org/10.1016/j.geoderma.2023.116587

Interpretive Summary: Estimating the amount of soil organic carbon (SOC) stored in agricultural fields is an essential monitoring need for farmers that could have economic consequences. Stratified sampling currently is the preferred approach for SOC monitoring protocols. However, more information is needed on how to stratify and quantify the potential benefits. Additionally, it is unknown how stratified sampling performs compared to alternatives like balanced sampling. We evaluated SOC at eight farmer fields in four states using a high-density sampling approach to a depth of over 24 inches. We evaluated stratified and balanced sampling using a previously proposed set of readily-available information from soil surveys, determining the number of samples needed to achieve a given level of accuracy for SOC storage. Compared to simple random sampling, stratified sampling reduced sample size by 18% on average and balanced sampling reduced sample sizes by 28% on average. The performance of both strategies varies across fields, but balanced sampling consistently outperformed stratified sampling, which provided minimal benefits at several sites. We conclude that in order to meet the urgent challenge of efficiently estimating SOC stocks in agricultural fields, balanced sampling may be superior to stratified sampling.

Technical Abstract: Estimating soil organic carbon (SOC) stocks in agricultural fields is essential for environmental and agronomic research, as well as for farm management and policy. Stratified sampling is a classic approach to estimating mean soil properties, and has recently been codified in SOC monitoring protocols. However, for the specific task of estimating SOC stocks in agricultural fields, concrete guidance is needed on how to stratify and quantify potential benefits. Moreover, it is unknown how stratified sampling performs compared to modern alternatives, notably balanced sampling. To address these gaps, we employ a Bayesian evaluation framework and high-density (average of 6 samples ha-1), deep (average of 70 cm) measurements at eight commercial fields across four states in the US Midwest. We evaluate stratified and balanced sampling using a previously proposed set of readily-available geographic, topographic, spectroscopic, and soil survey variables. We examine the number of samples needed to achieve a given level of SOC stock estimation accuracy. Compared to simple random sampling, stratified sampling would likely reduce sample sizes by 18% on average across fields. Balanced sampling would likely reduce sample sizes by 28% on average compared with simple random sampling. The performance of both strategies varies across fields, but balanced sampling consistently outperforms stratified sampling, which provides minimal benefits at several sites. We conclude that in order to meet the urgent challenge of efficiently estimating SOC stocks in agricultural fields, stratified sampling could be replaced by balanced sampling.