Researchers have characterized the differences between Red, Blue, and enhanced NDVI (ENDVI). A good publicly accessible study was recently conducted by PublicLab. They evaluated blue and red NDVI indices on various organic (i.e. living) and inorganic (i.e. non-living) materials. One would expect that living healthy photosynthetically active plants would have high NDVI values and that inorganic material (i.e. not living) and senesced or dead plants would have very low NDVI values. As can be seen in the table below, Red NDVI had much lower values than Blue NDVI for inorganic (i.e. non-living) materials and for dead or senesced foliage (i.e. dead grass). Both Blue and Red NDVI had high values for healthy green grass. This demonstrates that Red NDVI is more sensitive to crop stress and has a much higher contrast between stressed versus non stressed crop than Blue NDVI. This illustrates how absorption of blue light by anthocyanins and scattering of blue light in the atmosphere interferes with crop stress detection using Blue NDVI.
|Material||Blue NDVI (840/450)||Red NDVI (840/660)|
Table 1 Example study that published Red versus Blue NDVI correlations (http://publiclab.org/notes/nedhorning/10-30-2013/red-vs-blue-filters-for-ndvi)
Sentek Systems has conducted its own internal evaluation comparing Red, Blue, and Enhanced ENDVI. Spectrometer measurements were made on Corn and Soybean leaves. A range of leaves were selected ranging from very healthy leaves, medium healthy leaves showing some discoloration, and highly senesced leaves. A scientific grade spectrometer was used with a calibrated NIST certified reflectance standard. The spectrometer was calibrated in between each data collection period and the spectrometer was used to collect reflectance from each leaf over the spectral range from 350 to 1000 nm. Post-processing was performed on the collected data to calculate the Red NDVI, Blue NDVI, and ENDVI indices. Images of the leaves and the values of the indices are plotted below.
Observe how Blue NDVI and ENDVI pack all measurements into tight groups, while Red NDVI spreads them out. More generally, Blue NDVI and ENDVI do not appear to be good proxies for the health of the foliage.
Below we show false-color renderings of aerial Red NDVI, Blue NDVI, and ENDVI mosaics. This imagery was collected with the GEMS sensor and corresponds to a test plot from nitrogen trial on corn. In the Red NDVI imagery differences between plots are quite visible, with weaker plots showing stress quite clearly. In the Blue NDVI mosaic the differences between plots are not as clear, and some plot boundaries are no longer discernible. In the “enhanced” NDVI mosaic the different plots look almost uniformly healthy. Plot boundaries are no longer visible at all. These images illustrate how Red NDVI is more sensitive to plant stress (stress caused by N deficiency) than Blue or Enhanced NDVI. Furthermore, Enhanced NDVI appears completely incapable of distinguishing between healthy and stressed corn plots.