Discriminating Tropical Grass Grown under Different Nitrogen Treatments using Hyperspectral Data Resampled to HYMAP

要旨

The development of techniques to estimate and map grass quality is critical for a better understanding of wildlife feeding patterns. We tested whether canopy reflectance spectra, resampled to HYMAP resolution could discriminate between groups of tropical grass containing different levels of nitrogen concentration. Canopy spectral easurements were taken from Cenchrus ciliaris grass grown under three different nitrogen treatments using a GER 3700 spectrometer. Using the resampled spectra, the red edge position was calculated and continuum removal was applied on the red absorption feature between 550 nm and 757 nm. Canonical variate analysis was used to discriminate between the treatment groups using the transformed spectra. Results show that canonical functions derived from continuum removed red absorption feature, in combination with the red edge position (resampled to HYMAP resolution) can discriminate between the three nitrogen treatment groups. The canonical structure matrix also revealed that greater discrimination power is contained in the red edge slope. The results show the importance of the visible portion of the electromagnetic spectrum (where foliar water effect is minimal) in predicting foliar nitrogen concentration. Overall, the study demonstrated the possibility to up scale the method to airborne sensors such as HYMAP for mapping tropical grass quality.