I have sample points (shape file converted from a .xls file which contains XY coordinates,nnx,nny and canopy density) and a landsat etm+ image. First I applied 'SpatialAnalyst Tools>Extraction >Sample' in ArcGIS in order to extracted spectral signatures of all sample points from Landsat image. I then exported output to a DBF file and opened it in Excel. The DBF file has XY coordinates of sample points and NP_2001102,NP_2_30513,NP_2_30514,NP_2_30515,NP_2_30516 and NP_2_30517. I assume e.g. NP_2_30513 is band 3 and NP_2_30514 is band 4.

So the question is: 1), are they really band 3 and band 4? 2), if they are, and if I want to calculate NDVI values and then do linear regression between canopy density and NDVI in Excel, what should I do to get correct reflectance values before the calculation?

Examples of values of NP_2_30513 are 11, 9, 13, 8...etc., examples of values of NP_2_30514 are 10, 7, 7, 4, ...etc.

I know these values are not really reflectance values, but how can I get the correct values for my calculation and regression?

  • Could you please provide the source of your landsat imagery and provide an example of the naming convention used in the scene? Thanks.
    – Aaron
    Sep 27, 2014 at 19:31
  • I really don't know the source of the image. I got it from my professor and it is too big to upload here. I thought I need to apply e.g an equation after extracted signatures for getting real band values, like you need to *gian when its NOAA image. Sep 28, 2014 at 6:58

2 Answers 2


it´s seems you have digital numbers, if you want to transform it to reflectance read this article. For atmospheric correction you can use dark subtraction or Flaash (not in ArcGis). I think that ArcGis is not the best software for atmospheric correction.

If you are doing a temporal analysis you must perform an atmospheric correction.

  • Thanks for the reply. I used equation (1) given by the article, but still got the wrong reflectance values. For example, the 1st sample point has values: NP_2_30513:11, NP_2_30514:10.After I applied the equation (1),I got NP_2_30513:1.471, NP_2_30514:0.894 (ETM+ high gain). But according to my professor, the same point has values 56(band 3) and 93(band 4). He also applied resampling processes(nearest neighb and cubic convolution)before calculate regression. I have no idea how he did this. Any tips? Thanks! Oct 1, 2014 at 11:19
  • reflectance values are always between 0 and 1; and the data type must be float (or with decimals). In certain case you can obtain a value greater than 1. The same case for NDVI. But you have to consider that with this paper you obtain TOA reflectances, if you want to do temporal analysis you must apply atmospheric correction.
    – Pau
    Oct 1, 2014 at 11:41

I writte almost same question about working with multi-spectral indexes have a look. Is atmospheric correction necessary when working with multi-spectral indexes?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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