Deriving hyper spectral reflectance spectra from UAV data collected in changeable illumination conditions to assess vegetation condition

Hyperspectral imaging is a recent development in the evolving field of UAV remote sensing and a new avenue for habitat condition monitoring. We present preliminary results of a pilot study evaluating the use of UAV hyperspectral imaging to detect early stages of Acute Oak Decline (AOD) in a broadleaved forest in the UK. Field observations revealed that, compared to asymptomatic trees, leaves of symptomatic trees show lower levels of water and higher reflectance in the near-infrared part of the spectrum. The observed changes in leaf level reflectance spectra were subtle but statistically significant. Normalised hyperspectral UAV canopy radiance spectra suggest the opposite is occurring: symptomatic trees have lower near-infrared radiances. UAV campaigns suffer from changing illumination conditions and in our case normalizing between image frames is not sufficient. We plan to derive reflectance spectra to enable us to adequately evaluate the observed differences between leaves and canopies