Repeated scan times in PACE OCI CLDMASK V3.1 and cloud-mask choice for land smoke

[tingshuntian]

I am investigating land wildfire-smoke observations and found inconsistent scan-time fields in:

PACE_OCI.20250110T203503.L2.CLDMASK.V3_1.nc

The file is 7,152,211 bytes. My local SHA256 is 79956240771ccbbc385152bc7102194f57be78384385ff2c451c393f5a968dbb.

A second streaming retrieval from the CMR-provided NASA download endpoint on 2026-09-09 returned the identical size and SHA256.

Of 1710 rows, scantime at zero-based rows 30+28k and 31+28k (k=0 to 59) repeats the times at rows 2 and 3 of the same-start-time AOP V3.2 product. The other 1590 scantime values match AOP time exactly. The year/day/msec fields also appear displaced and are not a reliable substitute: their reconstructed time differs from scantime by up to 295.495 seconds. Navigation arrays of the two products are not identical, so I have not copied AOP time into CLDMASK.

Could you confirm whether this is a known issue and identify the appropriate validated time field or reprocessed product?

Separately, I see substantial spatial disagreement between AOP CLDICE and direct CLDMASK cloud_flag over this smoke scene. AOP records an empty cloud_mask_file, cloud_thresh=0.027 and cloud_wave=870. The cloud-mask ATBD discusses thick-smoke overscreening. Is there a recommended procedure for basic/dilated masks in land aerosol applications, and a way to distinguish their intended use from the AOP flag?

I am retaining both products as diagnostics and have not interpreted either as cloud truth or repaired the original files.

[amsayer]

I am a developer of the OCI CLDMASK product. You mention you're working over land, but then mention AOP files, by "AOP" do you mean "Apparent Optical Properties" (an ocean product) or something else? The ocean products have their own cloud mask. I am wondering if you mean something else given you mention smoke over land. Could you give the exact name of the AOP file in question?

I can't comment on the time issue but if you are concerned about mismatch I would verify that the latitudes/longitudes for the same scan lines and pixels match.