1

I am hoping to derive a variety of vegetation indices for grasslands in Peru using Sentinel 2 data for one cloud free image in June 2016. But most of my indices are not in the range of the index. For example,

CRI 1- theoretical range is 0-15, range determined is -0.012 to 0.010
EVI- theoretical range is -1 to 1, range determined is -10755 to 10515
EVI 2- theoretical range is -1 to 1, range determined is -0.84 to 2.016
SAVI- theoretical range is -1 to 1, range determined is -1.31 to 1.31
MSAVI- theoretical range is -1 to 1, range determined is -5.506 to 1.43

Using raster calculator in ArcMap, I have appropriately used Float() and also multiplied all variables to 1.0 to overcome the data type problems. Also, I have checked to see if extreme index values pixels are clouds, or edges etc, but they happen to be within the image itself i.e for example EVI 2 values >1.0 are green vegetation pixels itself.

How do I determine is the indices I have calculated are correct? If not correct, what is the right to calculate each.

1

@HDunn is right, you have to convert to reflectance using the quantification value (10000) which is in the product metadata (not the granule metadata).

For instance, EVI = G.(Nir-Red)/(Nir+C1.Red-C2.Blue+L)

If all the reflectances are multiplied by 10000, adding L=1 does not have much effect.

  • @O.Hagolle- you mean divide each band by 10000, not multiply. – tg110 Sep 15 '16 at 12:52
  • To convert to reflectance, you need to divide by 10000. But if you don't convert, then your reflectances are multipled by 10000. – O. Hagolle Sep 15 '16 at 16:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.