From AVHRR-NDVI to MODIS-EVI: Advances in vegetation index research

Zhengxing WANG, Chuang LIU, Huete Alfredo

UTS ePRESS (University of Technology Sydney) · 2003 · 160 citations · 0 references

Concepts

Abstract

Global AVHRR-NDVI data sets have been widely applied to many fields from land cover change to the extraction of various biophysical vegetation parameters. Yet there still remain some limitations in the NDVI product: (1) NDVI saturates in well-vegetated areas, partly a result of saturation in the Red channel and partly due to the ratio-based NDVI equation; (2) The effect of canopy background on NDVI has not been considered; (3) The ratioing properties of the NDVI along with the Maximum Value Composite (MVC) procedure does remove some sources of internal and external noise, but there still remain significant noise in the final NDVI products; (4) The MVC cannot guarantee the selection of the clearest pixels and smallest view angles. All of these limitations are improved to some extent in the Enhanced Vegetation Index (EVI) product from the Moderate Resolution Imaging Spectroradiometer (MODIS). The MODIS-EVI has several advantages over the AVHRR-NDVI; (1) The MODIS atmosphere correction scheme includes the effect of atmospheric gases, aerosol, thin cirrus clouds, water vapor, and ozone, whereas there are only corrections for Rayleigh scattering and ozone absorption in the AVHRR-NDVI product. This reduces the need for ratio-based vegetation indices, such as the NDVI, that remove some atmospheric noise at the cost of saturation; (2) The influence of residual aerosol is removed by the Atmosphere Resistant Vegetation Index (ARVI), which is based on the difference of Red and Blue aerosol scattering; (3) The influence of the canopy background is reduced by the Soil Adjusted Vegetation Index (SAVI); (4) The concepts behind the ARVI and SAVI are coupled together to form the Enhanced Vegetation Index (EVI), which removes both atmosphere and background noise simultaneously and; (5) A Constrained-View Maximum Value Composite (CV-MVC) algorithm is applied to select the clearest pixels with smallest view angles and a BRDF compositing scheme is being tested to further improve the seasonal depiction of vegetation dynamics. The MODIS-EVI has improved its linearity with vegetation, particularly in well-vegetated regions.