Thanks for your question. I conferred with our support team.

The Surface Reflectance (SR) product is provided as a 16-bit GeoTIFF image with reflectance values scaled by 10,000. No need to convert from DN to SR values.

For clarity, you might revise Planet Surface Reflectance Technical White Paper, pp. 2. Available at Surface Reflectance Products here. 

Hope this is helpful, if you have additional questions let me know. Have a nice day!

@idilgumus Thank you for this information.

Hello,
I am now trying to compare the reflectances of Planet images with spectrometer data. If I divide the DN values by 10000, for the case of vegetation in the NIR I get values of about 0.3, which seems to me low for theoretical values in spectral signatures, assuming a total reflectance as 1.

Can you tell me if I am assuming something wrong?

Thanks for your help

Hi,

If you are using the top-of-atmosphere (TOA) data, you need to convert the DN values to reflectance. To do so, there are conversion coefficients for each band in the metadata. But for the surface reflectance product, if you divide the DN values by 10,000 that should give you the reflectance in n0 1] range.

Gracias Milad, 

Estoy usando "analytic_8b_sr_udm2", este dato según los docs, es dato de reflectancia BOA, solo habría que dividir por 10000. Pero me genera dudas por las bajísimas reflectancias resultantes de la vegetación en el NIR después de hacer el conversión.

Thanks for the clarification. The NIR reflectance of vegetation is normally high but this also depends on the season and the status of vegetation. If you haven’t done it so far, you may compare the reflectance with other data which are collected close to the Planet data acquisition time (Sentinel-2, in-situ spectroradiometry, etc.).  

I have also same query. The downloaded surface reflectance dataset’s range is not between 0-1. How should I rescale it. In metadata file, coefficients are given for each band to rescale from DN to TOA radiance and DN TOA reflectance. But there is no information regarding surface reflectance.

Hi ​@Avinash Kumar Ranjan 

Here is the technical documentation on Surface Reflectance. 

​@elyhienrich Thank you for your response.

Hello.
I recently downloaded a series of images from PlanetScope. When I downloaded them, the following options appeared:
Surface reflectance - 4 bands:
Corrected for surface reflectance: Recommended for most analytical applications, includes NIR and RGB.

Analytical radiance (TOAR) - 4 bands:
Calibrated according to the radiance of the upper atmosphere: includes NIR and RGB.

For this, I chose the first option, that is, surface reflectance - 4 bands. It is worth mentioning that the images will be used to calculate different vegetation indices: NDVI, SAVI, EVI.
For the calculation of the NDVI, the values ​​obtained are within the expected range (-1, 1), the other indices present strange values, very far outside this range. The study area is part of the Valle del Cauca, Colombia, and those images that do not present cloudiness were considered for downloading.
Have I made any mistake when selecting the type of product to download (Surface Reflectance - 4 bands, Analytical Radiance (TOAR) - 4 bands)? Do I need to make any kind of correction to the downloaded images? Are the values ​​recorded in pixels scaled by 10000 and do I first need to perform the pixel value/10000 calculation?
I would like to point out that the vegetation indices used to calculate are for determining the Kc crop coefficient, so precision and accuracy are required in the calculated values.

Thank you very much, best regards!

Yes I think so. You need to divide the pixel values by 10000.