Differences between (Statistical API/QGIS Sentinel Hub Plugin) and EO Browser Statistical Info

Hi, I´m trying to compute mean chlorophyll index derived from Sentinel 2 (“B8/B3-1”) for a polygon. When I try Statistical info from EO Browser (first image) I get that maximum (peak of the curve) mean value of the index for that polygon is 3.51 in 7th february.

However, when I downloaded the histograms of the index for that polygon the mean value for that date is around 8.5 (second image).

When using QGIS Sentinel Hub Plugin (third image), visualization shows that values are closer to a mean value of 8.5 than 3.51.

When I tried visualizations in other sources (like Google Earth Engine) values for that date are the same as Statistical API/QGIS Sentinel Hub Plugin.

I realised that this happens with all polygons that I tried. Why are those values (Statistical API/QGIS Sentinel Hub Plugin or GEE visualization) so different from values in EO Browser Statistical Info? And how can I get these higher values in the EO Browser Statistical info?

Hello Matias,

EO browser uses the old FIS service, but there shouldn’t be any major discrepancies such as the ones you have described.

Would you mind sharing the polygon, so we can look into it?

Minor differences are of course expected if some statistics were done at different resolution than the others.

Best regards,

Jan

Thanks Jan. Here is the polygon, but as I said, it happens with any polygon.
Jacob.kml (968 Bytes)

I checked, and using the statistics vs. downloading the raw data as a TIFF, I obtained numbers, which were very similar (but differed slightly because of the varying resolutions).

It turns out the issue seems to be that you’re using Sentinel-2 L1C in EO browser, and atmospherically corrected Sentinel-2 L2A elsewhere.

If you check the statistics for that day in EO browser for Sentinel-2 L2A, you can see that the mean value is around 8:

I hope this helps.

Btw you can download the data for the field as 32-bit float TIFF image in Analytical download as well.

Best regards,

Jan

Thanks Jan! That was the problem.