GEDI level4A, predictor_limit_flag

[flowww]

Hi,

In the guide (https://daac.ornl.gov/GEDI/guides/GEDI_L4A_AGB_Density_V2_1.html) the following is stated for predictor_limit_flag & response_limit_flag :
"Predictor value is outside the bounds of the training data (0=in bounds; 1=lower bound; 2=upper bound)"
However, in my data I encounter values of 255, what do they stand for? Did I miss out on an additional document that expands further on flag value meanings?

Thanks a lot :)

[rupeshornl]

Hi @flowww, Thank you for contacting us. We have reached out to the GEDI science team. We will post here when we hear back. The variable descriptions are provided in the user guide and the ATBD.

[flowww]

Hi @rupeshornl,
Thanks a lot for your reply! I had looked through the guide and the algorithm document, but didn't find any info on my question.

I have a second question:
Is the ancillary data table provided with each granule (h5 file, eg. GEDI04_A_2019107224731_O01958_01_T02638_02_002_02_V002.h5) the same for all granules? With ancillary data table, I mean tables 6 to 8 in the user guide (https://daac.ornl.gov/GEDI/guides/GEDI_L4A_AGB_Density_V2_1.html).

[rupeshornl]

Yes, the ancillary tables (ANCILLARY/model_data, ANCILLARY/pft_lut, and ANCILLARY/model_lut) are same for all granules. This tutorial also explores the data structure of GEDI L4A.

[flowww]

As an afterthought:
Could the response_limit_flag == 255 have arisen from a typing problem?
E.g. that the type of response_limit_flag is defined as 1 byte storage class (e.g. uint8) which can as a maximum take the value of 255 and somehow this maximum is often exceeded and thus 255 is reported?
Thanks :)

[rupeshornl]

Hi @flowww, Sorry for the delayed response. The `predictor_limit_flag` value of 255 should be interpreted as a null value. This occurs when no valid rh_metrics were recorded and the `predictor_limit_flag` could not be computed. Thank you for reaching out!

[flowww]

Hi @rupeshornl,

Thanks a lot for your reply.
I am somewhat confused by the answer, however.
In my sample there are many GEDI footprints that have a valid abgd estimate and l4_quality_flag ==1 (thus I assume there would be valid rh_metrics as input, or is this my wrong assumption?) and still have predictor_limit_flag == 255 and response_limit_flag == 255.

One example for this is the footprint with row index 43718 (indices starting at 0) in GEDI granule
GEDI04_A_2022135204023_O19385_01_T01130_02_003_01_V002.h5:
agbd: 3.335451
l4_quality_flag: 1
predictor_limit_flag: 255
response_limit_flag: 255

Thanks a lot for the clarification :)

[rupeshornl]

We've reached out to the GEDI Science Team for clarification. We will post as more information becomes available.

[flowww]

Any news from the science team?

Thanks a lot :)

[flowww]

Hi @rupeshornl,

Could you give me an update for my question? The answer would still be relevant for me.

Thanks a lot :)