Munoz, J.D., A. Kravchenko, and A. Finley
Presented at the GLBRC Sustainability Retreat (2010-02-10 to 2010-02-12 )
Soil organic carbon (SOC) is a fundamental indicator of soil quality in crop production, especially in Midwest agricultural systems. Efficient tools for accurate within-field estimation of SOC are imperative for precision agriculture management as well as for assessment of soil C stocks and changes in time. Near infrared (NIR) reflectance spectroscopy has been used recently in aid to the conventional, laborious and expensive soil analyses. However, NIR spectrum data need to be effectively calibrated with conventionally measured SOC and calibration equations may vary depending on soil type and soil conditions. The objective of this study is to compare predictions of SOC in a spatial context by using three sources of auxiliary information: NIR, topographical data and aerial photographs (AP). Soil samples at 0-10 cm depth and NIR data were collected in two sections with slightly management differences of an agricultural field in southwest Michigan on April 21, 2008. Digital elevation models were derived from LiDAR dataset. Topographical parameters were derived from the DEM after point interpolation. An aerial photograph with visible and NIR bands was taken in May, 2008. Soil carbon was analyzed by dry combustion method. Results indicate the relationship between carbon and predictors are field dependant. NIR spectral data produced highest calibration accuracy. However, AP resulted in similar accuracy when combined with topographical data. Prediction accuracy was improved by the use of topography in both AP and NIR spectra.
Back to meeting | Show |