Relations between directional spectral vegetation indices and leaf area and absorbed radiation in Alfalfa

Abstract

Sensors on satellite platforms with extreme view angles have been increasingly used to analyze regional and global vegetation cover and productivity because of frequent observations. This study, using experimental and theoretical methods, analyzed variations in vegetation indices with sun-view geometry as a means of understanding the sensitivity of relations beween vegetation indices and the biophysical properties, the leaf area index (LAI), and the instantaneous fraction of absorbed photosynthetically active radiation (fAPAR). Canopy bidirectional reflectance factors (BRFs) of an alfalfa crop were measured and simulated at a variety of solar and view zenith angles. Also, fAPAR, LAI, and leaf optical properties were measured. Measured and simulated canopy reflectances agreed generally within 1% (absolute). Normalized difference and simple ratio vegetation indices (NDVI and SRVI, respectively), derived from BRFs, varied with view and solar zenith angles. The minimum for near-infrared (NIR) BRFs and relatively high red BRFs generally occurred near nadir, resulting in some of the lowest vegetation index values. Highest VI values were generally obtained at forward view angles. Variation of NDVI with sun-view-geometry was greatest at LA's <2, whereas the range in SRVI was greatest for LAIs>2. Measured reflectances indicate that relations between NDVI and LAI and between SRVI and fAPAR were curvilinear across all solar and view zenith angle combinations in the solar principal plane, whereas relations between SRVI and LAI and between NDVI and fAPAR varied from linear to curvilinear. Analyses revealed that vegetation indices at large view zenith angles were poorly correlated with fAPAR, whereas those at small zenith angles were strongly correlated. In general, vegetation indices were more sensitive to fAPAR than to LAI, which is attributed to the fact that fAPAR is a radiation quantity, whereas LAI is nonlinearly related to radiation. Regression of fAPAR with VI values derived from combinations of red and NIR BRFs from similar and nonsimilar directions indicates that the highest correlation is in near-nadir and backscatter directions. However, further investigation into variations of relations between remotely sensed observations and canopy attributes and into the usefulness of off-nadir in extracting information is recommended.