Is there a problem with ndvi function in ArcMap 10.2? Not getting expected results.

[dougMacfarland]

The imagery i am using is from landsat 8. My process is:

1. bring in Bands 3 and 4 (red and infrared) to data view.
2. Then from Windows-Image Analysis,assign the bands (red 3 IR 4) in options.
3. then highlight the two bands and execute the NDVI function (green leaf). 

The resulting single band is added to the table of contents.  At that point, the scale is -1 to 1, which is a typical NDVI scale.  But when I attempt to classify the pixels, the results are in the hundredths (ie, 0.02586...) with a range of only about 0.2 instead of normal readings which would be in the tenths (0.21 ,0.32 ,0.43....) with a range of 2.0 (-1.0 to 1.0). I have also done the calculation manually in raster calculator with the float expression and gotten the same results as the NDVI function.

Is this an issue with the Landsat tiffs?  The NDVI function?

Thanks again!

[EricRice · Accepted solution]

Hi Doug,

For Landsat 8, the Red band is 4 and the NIR band is 5.

Eric

[CodyBenkelman]

Doug

Eric is correct - with the addition of the deep blue band 1, the Red and NIR bands moved up one band number vs. Landsat 7 & previous.  So your NDVI needs to be configured for Bands 5 & 4. 

How does Landsat 8 differ from previous Landsat satellites

I just wanted to add a note to be sure you were aware you can get instant access to Landsat 8 via the ArcGIS Online image services (you mention "tiffs").  One of the raster functions available on this service is NDVI.  If you are interested, see Landsat Imagery | Fast and Easy Access to Landsat Imagery

Cody B.