Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 29, 2011
Publication Date: March 1, 2012
Citation: Chaudhary, V.P., Sudduth, K.A., Kitchen, N.R., Kremer, R.J. 2012. Reflectance spectroscopy detects management and landscape differences in soil carbon and nitrogen. Soil Science Society of America Journal. 76(2):597-606. Interpretive Summary: Methods for more efficient estimation of soil properties are needed, especially where many measurements are required to document differences among management systems or across landscapes. One specific need is for efficient soil carbon measurement, as society realizes the importance of tracking how well management practices sequester carbon in soils. Sensors that can estimate soil carbon and other properties without the need for sampling are a promising alternative. One technology that has received considerable attention in this regard is optical diffuse reflectance sensing in the visible and near infrared wavelength bands. Our goal in this study was to apply this approach to detect differences in topsoil organic carbon and total nitrogen among treatments within a long-term cropping systems experiment in comparison to standard laboratory methods. Similar to previous research, we found that optical reflectance sensing provided very good estimates of soil organic carbon and good estimates of total nitrogen. These reflectance-based estimates detected differences among cropping systems but were generally not sensitive enough to detect differences among positions on the landscape. In contrast, data obtained by standard laboratory analysis were able to detect landscape position differences and some additional cropping system differences. Although reflectance sensing was not quite as sensitive as standard laboratory methods, it is more efficient, and may be particularly useful in experiments where data must be collected and analyzed at many points. These results will be useful to scientists who need more efficient ways to estimate soil property variations across experiments and landscapes.
Technical Abstract: Many studies document the successful calibration of visible and near infrared (VNIR) diffuse reflectance spectroscopy (DRS) to various soil properties. However, few studies have reported on the use of VNIR DRS to detect treatment differences in controlled experiments. Therefore, our objective in this study was to investigate the ability of VNIR DRS to detect treatment differences in topsoil organic carbon (SOC) and total nitrogen (TN) compared to standard laboratory analysis. The study area was a long term (since 1991) experiment in central Missouri, where cropping systems (CS) were replicated across a typical claypan soil landscape. Soil samples from two depths (0-5 cm and 5-15 cm) were obtained in 2008 at three landscape positions (LP; summit, backslope, and footslope) for six CS [three grain cropping systems; mulch tillage corn-soybean (MTCS); no-tillage corn-soybean (NTCS); no-tillage corn-soybean-wheat (NTCSW); and three grass systems; conservation reserve program-cool season (CRP-C); conservation reserve program-warm season (CRP-W); and hay crop (HAY)]. Cross validated estimates of SOC by VNIR DRS using field-moist samples were very good, with R2 = 0.87 and root mean square error (RMSE) = 0.22 g kg-1. Estimates of TN were somewhat less accurate (R2 = 0.79, RMSE = 0.022 g kg-1). Field-moist VNIR DRS data were able to detect significant differences among CS. The NTCSW had significantly higher SOC and TN than other two grain CS in the 0-5 cm depth. However, MTCS SOC and TN were higher than NTCSW in the 5-15 cm depth, due to tillage in MTCS moving residue into the soil. Among the grass CS, SOC and TN for CRP-W were generally lower than for the other two systems, although not always significantly. Differences among LP were generally not significant using field-moist VNIR DRS data. This was in contrast to standard dry combustion laboratory analysis where backslope SOC and TN were generally significantly higher than at summit and footslope. Clearer separation among CS was also seen with laboratory data in a few cases. Overall, VNIR DRS was slightly less able to detect treatment differences than standard laboratory methods. However, the efficiencies inherent in this method, particularly when applied to field-moist soil, suggest that it deserves consideration as a tool for determining SOC and TN differences in field experiments.