[GRASS-dev] Unexpected EVI range from "i.vi"

Markus Metz markus.metz.giswork at gmail.com
Mon Jun 24 12:28:32 PDT 2013


Markus Metz wrote:
>
> I am pretty sure that the EVI2 formula in i.vi is not cross-sensor,
> but also tailored to unscaled MODIS input bands:
>
> EVI2 = G * ( nir - red)  / (nir + C1 * red + L)

Hmm, no, it seems that the coefficients for both EVI and EVI2 are
meant for scaled MODIS input bands (top of atmosphere reflectance).

The coefficients for EVI are a mystery. The MODIS algorithm
theoretical basis document refers to Huete and Liu (1996), which does
not exist in the references. Huete et al (2002) specifies L = 1, C1 =
6, C2 = 7.5, and G = 2.5, referring to Huete, Justice, & Liu (1994)
and Huete, Liu, Batchily, & van Leeuwen (1997), but none of these
references explain how these numbers were determined. Interestingly,
the MODIS algorithm theoretical basis document sets G = 2, not 2.5,
without any explanation.

In order to get an EVI in the range [-1, 1], using

EVI = G * (nir - red)  / (nir + C1 * red - C2 * blue + L)

(nir + C1 * red - C2 * blue + L) must be >= G * (nir - red)

which could easily be added to i.vi.

In any case, L needs to be adjusted to the actual input scale.

In summary, rather use NDVI.

Markus M


More information about the grass-dev mailing list