Submitted to: Proceedings of American Society of Agricultural Engineers
Publication Type: Proceedings
Publication Acceptance Date: August 12, 2005
Publication Date: October 1, 2005
Citation: Yang, C., Everitt, J.H., Bradford, J.M. 2005. Using multispectral imagery and pixel unmixing techniques for estimating crop yield variability. American Society of Agricultural Engineers. Paper No. 05-1018. St. Joseph, Michigan. 2005 CDROM.
Interpretive Summary: Vegetation indices calculated from remotely sensed data are commonly used to estimate biophysical attributes of interest, but most of these indices only use two spectral bands in the data. This study evaluated linear spectral unmixing techniques, which can use all the bands in the data, for estimating grain sorghum yield variability from airborne multispectral imagery. Linear spectral unmixing models were applied to five time-sequential airborne multispectral images collected from a grain sorghum field to generate crop plant and soil abundances. The abundance values were significantly related to grain yield and produced equivalent or better correlations compared with the normalized difference vegetation index. These results indicate that linear spectral unmixing techniques can be used alone or in conjunction with vegetation indices for quantifying crop canopy cover and mapping crop yield.
Vegetation indices derived from multispectral imagery are commonly used to extract crop growth and yield information. Spectral unmixing techniques provide an alternative approach to quantifying crop canopy abundance within each pixel and have the potential for mapping crop yield variability. The objective of this study was to apply linear spectral unmixing techniques to airborne multispectral imagery for estimating grain sorghum yield variability. Five time-sequential airborne multispectral images and yield monitor data collected from a grain sorghum field were used for this study. Both unconstrained and constrained linear spectral unmixing models were applied to the images to generate crop plant and soil abundances for each image and all 26 multi-image combinations of the five images. Yield was related to spectral abundances and significant correlations were found between yield and abundances. For comparison, yield was also related to the normalized difference vegetation index (NDVI) and the green NDVI (GNDVI). Results showed that although unconstrained plant abundance didn’t provide as good correlations with yield as GNDVI for three of the five images, it had better correlations with yield than NDVI for each image. Moreover, unconstrained plant abundance provided better overall correlations with yield than constrained abundances, and the unconstrained model was not as sensitive to the variation in endmember spectra as the constrained model. Multi-image combinations improved the correlations with yield and the best three-image combination resulted in the highest overall r2-value (0.813) between yield and unconstrained plant abundance. These results indicate that linear spectral unmixing techniques can be a useful tool for quantifying crop canopy cover and mapping crop yield.