Document: ABD-3-89-17

Spatial scaling issues in hyperspectral remote sensing of grass and chaparral ecosystems.

RAHMAN, A.F.*, J.A.GAMON, D.A.SIMS, D.A.FUENTES and H.QIU

California State University, Los Angeles, CA 90032 USA ¹

Abstract:
Hyperspectral (narrow-band) remotely sensed data are useful for studying ecosystem processes and patterns. But spatial characterization of the remotely sensed images is needed to optimize sampling procedures and to address scaling issues. We have investigated the spatial scaling issue in hyperspectral data for canopy and watershed level ecosystem studies in the Santa Monica region of southern California. Several reflectance indices, such as NDVI (a greenness index), WBI (a water content index), and PRI (a photosynthetic index) were used as potential indicators of biomass and physiological fluxes of grass and chaparral vegetation in that region. We found that the low altitude (~4 km above sea level) AVIRIS images were very suitable for studying variations in these indices between individual plant canopies. Semivariogram analyses using ground-based 1-m resolution hyperspectral data and AVIRIS images showed that the 4-m pixels of low altitude AVIRIS images satisfactorily retained the variability associated with individual canopies. On the other hand, 20-m pixels of high altitude (~20 km above sea level) AVIRIS images failed to capture canopy-level variability, but were suitable for studying watershed level processes.

Keywords: Hyperspectral, remote sensing, scaling, semivariogram, spectral indices

This abstract is being presented at: 4:00 PM in session:
Oral Session #64: Remote Sensing.