ASSESSING CLIMATE, SOIL AND LANDSCAPE PROCESSES AFFECTING AGRICULTURAL ECOSYSTEMS
Title: Infrared sensors to map soil carbon in agricultural ecosystems
Submitted to: Book Chapter
Publication Type: Book / Chapter
Publication Acceptance Date: February 15, 2009
Publication Date: July 1, 2010
Citation: McCarty, G.W., Hively, W.D., Reeves, J.B., Lang, M.W., Lund, E., Weatherbee, O. 2010. Infrared sensors to map soil carbon in agricultural ecosystems. In: Viscarra-Rossel, R., McBratney, A., Minasny, B., editors. Proximal Soil Sensing, Progress in Soil Science Volume 1. New York, NY: Springer Science. 14:165-176.
Interpretive Summary: Increasing carbon dioxide content of the Earth’s atmosphere has stimulated research to assess the role of terrestrial ecosystems in the global carbon cycle. The terrestrial biosphere is an important component of global carbon budget, but estimates of sequestration in terrestrial ecosystems are partly constrained by the limited ability to assess the distribution of soil C storage. Agricultural croplands have a great potential for sequestering atmospheric C if C-positive farming methods such as no-till, organic, and perennial cropping are adopted, but current technologies for monitoring soil carbon in terrestrial ecosystems are not cost effective, or they depend on laborious methods. We explore rapid methods of measuring soil carbon using near-infrared (NIR) and mid-infrared (MIR) diffuse reflectance spectroscopy. We determine that these methods perform well and demonstrate that airborne hyperspectral images of bare soil can be used to generate high resolution maps of soil carbon. This demonstration has shown the power of remote sensing approaches for assessing distribution patterns of soil carbon in terrestrial ecosystems and understanding soil carbon dynamics.
Rapid methods of measuring soil carbon such as near-infrared (NIR) and mid-infrared (MIR) diffuse reflectance spectroscopy have gained interest but problems of accurate and precise measurement still persist resulting from the high spatial variability of soil carbon within agricultural landscapes. Tillage-based and airborne-based NIR sensors offer opportunity to effectively capture the spatial structure of soil carbon within agricultural landscapes. We evaluated an airborne spectral sensor covering the range from 450 to 2450 nm at 2.5-m spatial resolution and a tillage sensor covering the range from 920 to 2225 nm. We intensively soil sampled five tilled agricultural fields within the flight path of the airborne sensor. The test fields were located on the Delmarva Peninsula in Maryland, USA. The information quality of spectral data acquired by these field-based sensors was compared to laboratory-acquired spectral data in both NIR (1000 to 2500 nm) and MIR (2500 to 25000 nm) spectral regions for the soil samples taken at 304 geo-referenced locations within the fields. The Partial Least Squares (PLS) regression models developed from the three NIR spectral data sources were very comparable, indicating that these two field-based NIR sensors performed well for generating spatial data. Although the laboratory based MIR calibration was found to be substantially better than the laboratory derived NIR calibration, current instrumentation limitations favor the use of NIR for in field measurements. A 2.5 m resolution soil carbon map was produced for an agricultural field using the airborne hyperspectral image using the PLS calibration. This new approach for mapping soil carbon will permit better assessment of soil carbon sequestration in agricultural ecosystems by use of improved landscape models that account for biogeochemical and soil redistribution processes that occur within often complex topographic and management settings.