Vegetation indices in crop assessments

Abstract

Vegetation indices (VI), such as greenness (GVI), perpendicular (PVI), transformed soil adjusted (TSAVI), and normalized difference (NDVI), measure the photosynthetic size of plant canopies and portend yields. A set of equations, called spectral components analysis (SCA), that interrelates VI or cumulative seasonal VI (∑ VI), leaf area index (I), fractional photosynthetically active radiation (FPAR, dimensionless), cumulative daily PAR energy absorbed (∑ APAR, MJ/ m 2), above-ground dry photomass (DM, g/m 2), and economic yield (Y, g /m 2) are presented and used to analyze data from two studies conducted in 1989. In one study we made boll counts and percent plant cover measurements in a salt-affected cotton field on 60 m grid intervals and calculated GVI, PVI, TSAVI, and NDVI at the grid intersections for SPOT-1 HRV and videography scenes. The four VI from SPOT accounted for more of the variation in the lint yield estimated from the boll counts (75–76%) than those from videography (61–63%), but VI from the different systems each accounted for 67% of the variation in plant cover. All relations were linear. In the other study, reflectance factors and FPAR were measured periodically during the season in corn planted at three densities (7.7, 5.4, and 3.1 plants/m 2). FPAR could be estimated from NDVI and PVI, respectively, by FPAR = −0.344 + 0.229 exp (1.95 NDVI) ( r 2 = 0.973) and FPAR = 0.015 + 0.036( PVI) ( r 2 = 0.956). Methods of obtaining FPAR and their effect on the efficiency of conversion of APAR to DM are illustrated and discussed. The data demonstrate how SCA unifies and strengthens the scientific basis of VI interpretations.