Skip to main content
ARS Home » Research » Publications at this Location » Publication #139183
Title: USE OF SPATIAL VARIANCE INFORMATION FROM REMOTE SENSING IMAGERY TO MAP VEGETATION FOLIAGE DENSITY

Author
item Walthall, Charles
item Timlin, Dennis
item Pachepsky, Yakov
item Dulaney, Wayne
item Daughtry, Craig

Submitted to: American Geophysical Union
Publication Type: Abstract Only
Publication Acceptance Date: 12/1/2002
Publication Date: 12/11/2002
Citation: WALTHALL, C.L., TIMLIN, D.J., PACHEPSKY, Y.A., DULANEY, W.P., DAUGHTRY, C.S. USE OF SPATIAL VARIANCE INFORMATION FROM REMOTE SENSING IMAGERY TO MAP VEGETATION FOLIAGE DENSITY. AMERICAN GEOPHYSICAL UNION. 2002.

Interpretive Summary:

Technical Abstract: Maps of foliage density expressed as leaf area index (LAI) are used for natural resources inventories, land surface-atmosphere interaction modeling, and hydrologic modeling. Remote sensing imagery can be used to produce these maps by relating spectral vegetation indexes (SVIs) to LAI calibration samples acquired at selected locations on the surface. This approach traditionally uses ordinary least squares (OLS) relationships between the surface measurements and the SVIs, and does not fully take advantage of the spatial information content of the imagery. Spatial information inherent in a semivariogram of the imagery may provide additional information for mapping LAI patterns. This is demonstrated using a spatially dense sample of corn LAI and calibrated airborne imagery. An LAI map is produced by interpolating surface measurements with a semivariogram from the imagery. The resulting LAI map captures the main spatial features of a LAI map produced by interpolating the surface LAI data with its semivariogram. The image semivariogram approach also provides a product that has less noise characteristic of OLS-based remote sensing methods. The use of the image semivariogram with the surface LAI calibration samples suggests that the spatial domain information can complement spectral information for improving LAI maps especially at high spatial resolution where OLS methods may not perform well.