Radiometric Offset after PB04.00 turning values of 1000 to nodata

Dear friends,

As noted in this forum discussion ([INFO] Introduction of additional Radiometric Offset in PB04.00 products), prior to PB 04.00, Sentinel-2 negative values were not retained. Post PB 04.00, a RADIO_ADD_OFFSET = -1000 is applied at L2A.

L2A nodata values are 0. After applying the offset, images may show “valid” values ranging from -1000 to 10000.

If the offset is applied to non-nodata pixels, it may cause pixels with a value of exactly 1000 to become nodata (0).

I’ve observed this issue, particularly in areas such as in the image “S2A_B04_20230110T151701_N0509_R125_T19QDB_20230110T200453”, where a few water pixels have values <1000, and some exactly at 1000. After applying the offset, these pixels become <0 and 0, respectively.

What is the best solution to avoid this kind of pixel loss? Should nodata be changed to another value?