Masking in MCD19A1CMGL product

[ajwow]

Dear EarthData Forum,

I am using the MCD19A1CMGL product (Daily 0.05° global surface BRF CMG for Land Bands) for BRF and BRFn. I noticed that the data are already masked in many regions, but I could not find a clear description explaining why they are masked.

If my understanding is correct, BRF in 0.05° MCD19A1CMGL is composited from 1 km BRF in MCD19A1 product, which is produced in “cloud-free and clear-to-moderately turbid (AOD (0.47) < 1.5) conditions”. Therefore, the unmasked 0.05° data is the average of 1 km BRF pixels from
1. The selected orbit with coverage fraction > 0.5;
2. Cloud-free and clear-to-moderately turbid (AOD (0.47) < 1.5) conditions.
Is this understanding correct? My main concern is to confirm whether the unmasked data are already cloud-free (It would also be helpful to know the threshold used to define cloud-free).

I am also wondering whether the normalized BRFn and angular parameters (cSZA, cVZA and Rel_AZI) are also averaged from higher-resolution data. Do they match the same higher-resolution pixels used for the unnormalized BRF, and are they also in clear-sky conditions)?

Thank you in advance!

[LP DAAC - dgolon]

Hello @ajwow We checked with the Science Team of this dataset and heard back the following:

The MAIAC surface BRF CMG data are aggregated from the 1-km surface BRF product through a selection process rather than averaging. Please see the last page of the user guide for details: https://lpdaac.usgs.gov/documents/1500/MCD19_User_Guide_V61.pdf. The angular parameters correspond exactly to the BRF fields.

The normalized BRF is first derived from the 1-km BRF, then aggregated to the CMG grid through an averaging process.

All 1-km pixels have been masked for clouds, cloud shadows, etc., so all non-fill-value pixels are valid.