EVI

The Enhanced Vegetation Index (EVI) is a spectral vegetation index designed to address key limitations of the widely used NDVI. Developed as part of NASA's MODIS (Moderate Resolution Imaging Spectroradiometer) program, EVI incorporates atmospheric correction coefficients and a canopy background adjustment directly into its formula, producing more reliable vegetation assessments in challenging conditions. Unlike NDVI, which uses only red and near-infrared (NIR) bands, EVI adds a blue band to decouple atmospheric aerosol scattering from the vegetation signal. This makes EVI particularly valuable in regions with heavy aerosol loading, dense tropical canopies, or high biomass where NDVI tends to saturate.

EVI is calculated as: EVI = G × (NIR − Red) / (NIR + C₁ × Red − C₂ × Blue + L). In the standard MODIS implementation, the coefficients are G = 2.5 (gain factor), C₁ = 6, C₂ = 7.5 (aerosol resistance coefficients), and L = 1 (canopy background adjustment). The C₁ and C₂ coefficients use the difference between blue and red band reflectances to estimate aerosol influence — because aerosol scattering cross-sections are larger in the blue wavelength region, higher aerosol concentrations produce a larger blue-red reflectance gap. This information stabilizes the index against variations in atmospheric aerosol concentration.

The L factor adjusts for soil brightness effects that can distort the vegetation signal, especially in areas with partial canopy cover. The blue band (~0.45–0.52 μm), while critical for aerosol correction, has a notably lower signal-to-noise ratio over land surfaces due to very low reflected energy in that spectral region. This led to the development of EVI2, a two-band variant that drops the blue band and performs comparably under most conditions.

The blue band required for full EVI calculation has inherently low signal-to-noise ratios over land surfaces, which can introduce noise in individual scenes. Not all satellite sensors include a blue band suitable for EVI computation, limiting its applicability compared to NDVI. EVI is more computationally complex than NDVI due to its multi-coefficient formula and the need for atmospherically corrected surface reflectance inputs — top-of-atmosphere reflectance is insufficient. In snow-covered or water-dominated pixels, EVI can produce unreliable values. Additionally, while EVI reduces atmospheric effects, it does not eliminate them entirely, and residual aerosol contamination can still affect results in extreme haze or smoke conditions.

EVI was developed in the mid-1990s by Alfredo Huete and colleagues at the University of Arizona as part of the MODIS science team's effort to improve upon NDVI for the new Earth Observing System (EOS) satellite program. The index was formally introduced in the MODIS vegetation index algorithm theoretical basis document and first operationally produced when the Terra satellite launched in December 1999. The foundational paper by Huete et al. (2002) in Remote Sensing of Environment established EVI as a standard product. In 2008, Jiang et al. published the EVI2 variant, which removes the blue band dependency while preserving most of EVI's advantages, expanding its use to sensors like Landsat's earlier missions that lacked a suitable blue channel.