remote sensing - Is atmospheric correction necessary when working with multi-spectral indexes?

When you are working with Landsat Level 1 data and calculate indexes, e.g. NDVI, NDBI, SAVI ..., do you calculate them using raw data (pixel DN value), or atmospherically corrected data (surface reflectance)?

Answer

It depends upon the intended use of the Landsat data. Generally speaking, if you are doing multi-temporal analyses, you need atmospherically corrected data, otherwise DN format is sufficient. I would recommend reading the following landmark paper on the subject:

Song, C., Woodcock, C. E., Seto, K. C., Lenney, M. P., & Macomber, S. A. (2001). Classification and change detection using Landsat TM data: when and how to correct atmospheric effects?. Remote sensing of Environment, 75(2), 230-244.

From the conclusion in Song et al. 2001:

Atmospheric correction is not always necessary for classification and change detection. Both simple theoretical analysis and empirical results indicate that only when training data from one time or place are applied in another time or place is atmospheric correction necessary for image classification and many change detection methods.