Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → États-Unis → 1995 → Analysis of high spectral-resolution remote sensing data for quantifying low cover levels of green vegetation

University of Nevada, Reno (1995)

Analysis of high spectral-resolution remote sensing data for quantifying low cover levels of green vegetation

Chen, Zhikang

Titre : Analysis of high spectral-resolution remote sensing data for quantifying low cover levels of green vegetation

Auteur : Chen, Zhikang

Université de soutenance : University of Nevada, Reno

Grade : Doctor of Philosophy (PhD) 1995

The detection of low cover densities ($<$30% cover) of green vegetation using broad-band remotely sensed data continues to be problematic due to spectral variations in background materials (rock, soil, litter). Broad-band data (AVHRR, MSS, TM) are unable to directly detect the chlorophyll red-edge, the spectral feature diagnostic for green vegetation. It is hypothesized that high spectral-resolution (narrow-band) data are capable of detecting the presence and measuring the amplitude of the red-edge under low green vegetation cover conditions, with minimal influence from background spectral variations. To test the hypothesis, an experiment has been conducted using narrow-band PS-2 spectrometer’s reflectance spectra, having 4 nm bandwidths, of a pinyon pine and a sagebrush canopy with five different gravel backgrounds. From these reflectance spectra, narrow- and broad-band conventional red versus near infrared (NIR) vegetation index values (DVI, NDVI, RVI, PVI, SAVI, SAVI$\sb2$ and TSAVI) were calculated. Moreover, derivative-based green vegetation indices (DGVIs) for the narrow-band data have been developed. The performance of these vegetation indices was evaluated based on their capability to accurately estimate Leaf Area Index (LAI) and percent green cover. The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) data having 10 nm bandwidths were also examined for quantifying low green cover levels. Small red-edge amplitudes were detected in both PS-2 and AVIRIS reflectance and derivative spectra. Narrow-band versions of the red versus NIR vegetation indices had only slightly better accuracy than their broad-band counterparts. The background effects were minimized using the DGVIs. The AVIRIS derived DGVIs exhibited effectiveness in the estimation of green vegetation cover in areas having discontinuous plant canopies. Interactions between bandwidths, vegetation index formulations, vegetation species and the predictive power of a vegetation index were revealed. The research indicates that the capability to detect the chlorophyll red-edge, at low levels of green vegetation cover, is only possible with high spectral-resolution data. Derivative approach is an optimal technique to suppress background effects and to improve detection limit of the red-edge based green vegetation signal. Estimation of LAI and percent green cover based on the conventional vegetation indices can be highly inaccurate in broad-band datasets of regions having discontinuous plant canopies and spectral variations in background materials. The capability of detecting low green vegetation cover conditions by narrow-band data will improve hydrologic modelling in arid and semi-arid regions

Mots clés : AVIRIS, Applied sciences, Environmental science, Remote sensing, Hydrology, Health and environmental sciences Geography, Earth sciences, Aerospace materials

Accès au document : Proquest Dissertations & Theses

Page publiée le 26 février 2015, mise à jour le 1er janvier 2017