Submitted to: Agriculture Ecosystems and the Environment
Publication Type: Peer reviewed journal
Publication Acceptance Date: 10/21/2004
Publication Date: 5/2/2005
Citation: Beeri, O., Phillips, R.L., Liebig, M.A., Carson, P. 2005. Alternate Satellite Models for Estimation of Sugar Beet Residue Nitrogen Credit. Agric. Ecosys. Environ. 107:21-35. Interpretive Summary: Sugar beet leaves are often high in nitrogen (N) content, and sugar beet canopy N is often a factor when fertilizer application rates are calculated by agronomists. Canopies high in leaf N content could contribute enough soil N to mitigate fertilization the following year without impacting yield, and satellite data can assist growers with canopy N content assessment. Satellite imagery has historically been used to indicate relative differences across the field using indices such as the Normalized Vegetation Index (NDVI). However, methods for quantitative assessment of sugar beet N are lacking. We determined how satellite data might be used to estimate canopy N for sugar beet growers who wish to avoid the economic and environmental impacts of over-fertilization. We modeled IKONOS satellite data and compared measured to estimated canopy N content. On average, canopy N measured was 350 kg ha-1 and satellite estimated canopy N was 300 kg ha-1. Results suggest canopy N content can be estimated with satellite data to support more accurate agronomic recommendations.
Technical Abstract: Satellite assessment of aboveground plant residue mass and quality is essential for agro-ecosystem management of organic nitrogen (N) because growers credit a portion of residue N towards crop requirements the following spring. Precision agriculture managers are calling for advanced satellite models to map field-scale residue mass and quality. Remote sensing has proven useful for assessing the concentration of foliar biochemicals under controlled laboratory conditions, but field-scale satellite model validation for quantitative, landscape-scale N assessment is needed. We addressed this problem by building ground-truth models for sugar beet N-credit and testing these models with alternate satellite sensor imagery. We recorded spectral reflectance and measured leaf carbon (C) and N in situ at leaf and canopy levels near the end of the growing season using 1 nm bandwidth spectroradiometer. We performed univariate correlation analyses between spectral reflectance and the variables N, C:N ratio and biomass to determine spectral signature models for leaf quality and spectral signature models for plant biomass. The 1 nm hyperspectral data were convolved to fit Landsat 5, SPOT 5, Quick-Bird 2, and Ikonos 2 multi-spectral satellite bands and models created using stepwise linear regression. Biomass formulae for each sensor were applied to satellite imagery acquired at peak season, while leaf quality formulae were applied to imagery acquired just prior to harvest. August sugar beet fields in the St. Thomas, ND vicinity were identified and aboveground biomass mapped with 10–20% error, depending upon the sensor. Sugar beet leaf N was similar for all sites and varieties tested (31 mg g-1 dw), so biomass primarily influenced N-credit estimates. Measured C:N ratio variability was identified and mapped to delineate areas where C:N ratio was outside the normal distribution. The general model for each sensor maps N-credit per unit area and delineates aberrant, low leaf quality areas as zones with high C:N ratio. In summary, we provide separate spectral models for N-credit and leaf quality applicable to available multi-spectral sensors for precision sugar beet N management.