Hi Astana,

For any TOA radiance data, it’s vital to convert to TOA reflectance before calculating NDVI or similar parameters.  NDVI calculated from radiance will give very different values than if it’s calculated from TOA reflectance.  Planet’s ortho_analytic_4b or ortho_analytic_8b assets are TOA radiance, not TOA reflectance.  The parameters to convert are in the associated XML file, or you can use the Orders API to apply the conversion as part of delivery.

Note that this is separate from an atmospheric correction and converting to surface reflectance (SR).  Ideally, you’d use SR data, but there are cases where the SR correction can cause problems and you want to use top of atmosphere data instead. The key part is making sure the data is in units of reflectance, regardless of whether or not atmospheric correction has been applied.

The reason it’s important to use reflectance and not radiance in band ratio calculations has to do with what radiance represents.  Radiance is in units of energy (watts per square meter).  Shorter wavelengths have more energy.  So the radiance values for each band vary systematically.  If you look at a histogram of a radiance-based asset for each band, you’ll see that the histograms do not overlap a ton, and there’s a systematic shift that depends on wavelength.  Compare that to a histogram of reflectance for the same data, and you’ll notice that the histograms for each band overlap much more and the systematic shift is no longer present.

This difference affects any calculation that uses band ratios, including NDVI.  If you calculate NDVI on radiance data, you’ll get a very different number than if you calculate it on reflectance data for the same scene. NDVI will be systematically lower when calculated on radiance data than it is on reflectance data.  The standard is to use it for reflectance, and if you’re comparing to expected values (e.g. bare soil should be somewhere around 0.1 - 0.3 ish), it’s vital that reflectance is used in the calculation.