Details

Autor(en) / Beteiligte

• Wilson, Helene

Titel

Landsat measurement of green vegetation cover in a heterogeneous landscape

Ort / Verlag

ProQuest Dissertations & Theses

Erscheinungsjahr

1990

Link zum Volltext

• ProQuest

Quelle

ProQuest Dissertations & Theses A&I

Beschreibungen/Notizen

• Remotely sensed study of vegetation is increasingly addressing large geographical areas. If quantitative relationships between satellite-derived visible/near-infrared vegetation indexes (VNVIs) and vegetation quantity can be extrapolated to regional, continental, or global areas, important contributions to monitoring and modelling of climatic and hydrologic processes, biological production, and biogeochemical fluxes would ensue. However, the stability of VNVI-vegetation relationships across different landscapes is not known. The use of VNVIs as universal measures of vegetation quantity is based on the assumption that green vegetation has unique and unvarying spectral features. Previous empirical studies have not adequately tested the geographic generality of this assumption because of limitation to simple, homogeneous landscapes. One indicator of the stability of VNVI measurement of vegetation with surface variation is the strength of the VNVI-vegetation relationship in a heterogeneous landscape. This study examined the relation between two representative Landsat MSS VNVI transformations and fractional green vegetation cover in a New England urban-suburban summer landscape characterized by a wide range of land cover diversity at sub-pixel scale. The Landsat image was precisely registered to 1:12,000 color infrared aerial photography, from which surface characteristics were densely sampled for 504 pixels. Surface variations in the sampled pixels included variable species, morphology, shadowing, and spatial pattern of green and non-green vegetation, as well as variable ground surface type, brightness, and shadowing. Despite this variability, a simple linear relation explained 84 and 87 percent of the variance between the n-space greenness vegetation index and the normalized difference vegetation index, respectively, and total fraction of a pixel covered by green vegetation, including herbaceous, woody, and shadowed components. This strong relation lends support to the idea that a VNVI is, to the first order, a measure of green vegetation that transcends vegetation and other surface categorizations. These results suggest that similar studies should be conducted in other types of heterogeneous landscapes as a further step in exploring the geographic generality of VNVIs.