Earthdata Forum

Hi,
I'd like to better understand differences that we are seeing between the VIIRS thermal anomalies (TA) produced through our direct readout over Europe compared to the FIRMS NRT products which we use for daily validation of the TAs.

Our team produces RT (to use the current definition from the FIRMS FAQ) TA outputs over Europe and then once a day we download the FIRMS NRT TA to compare the two datasets. The FIRMS data is downloaded the morning after the day of interest from nrt4.modaps.eosdis.nasa.gov/api/v2/content/archives/FIRMS/modis-c6.1/Global/MODIS_C6_1_Global_MCD14DL_NRT_ .

What we observe generally is the following:
1. There are usually more FIRMS NRT TAs for the same area.
2. Most of the time, there is <10% difference between our RT outputs and the FIRMS NRT. This is considered 'acceptable' at the moment.
3. There are days where FIRMS TAs are found in regions where no RT TAs are found. These are generally single TAs with low FRP values.

I wanted to understand/discuss the following:
a. What is the most likely cause of the differences observed? Is it the processing workflow (i.e. software and versions) or is it the availability of other satellite metadata which improves the NRT FIRMS processing after full orbits?
b. From the FIRMS FAQ: To facilitate this, a number of changes have been made to the standard processing approach: for the URT and RT, direct broadcast data are used and for NRT, data downlinked from the satellite are sorted, processed, and delivered in an expedited manner (as Session-based Production Data Sets). -> can NRT TA data be updated in the archives/FIRMS/modis-c6.1/Global/after the expected 1-3 hours processing time?
c. Can you think of other reasons for the observed differences/ discrepancies in the number of TAs identified?

Thanks in advance for your help and keep up the great work,
Conrad

@cbielski_riscognition.com
Hi Conrad,
You are looking to understand the differences between TA from a direct readout station and FIRMS. As you point out this could be due to a number of things but most likely the algorithm (including version) that is being used. What algorithm are you using?
Diane

Good morning Diane,
for the VIIRS processing we use the CSPP Active Fires version: 2.1, ProcessVersionNumber = "2.5.4". I would have thought that the results should be pretty 'stable'. Most of the time, we have a few differences here and there (which I could account to differences in versions). But then every so often, there are larger chunks missing/appearing for certain overpasses.

As you can see in the attached figures, that particular day there is a <10% difference. And the map shows where this difference is located. Note that in the attached figures we have more TAs than FIRMS, but the opposite also occurs regularly whereby FIRMS has more TAs than us.

Thanks for your help,
Conrad

Hi Conrad,

It looks like v2.1 is the current version of the VIIRS active fire detection algorithm in CSPP. CSSP uses the NOAA fire algorithm. FIRMS uses the NASA algorithm - I am not sure what the differences in the algorithm. I have made an enquiry but not heard back yet. There will be a difference in the algorithm being used for the Level 1 processing so that could also be the cause.

I will email you directly if I hear more about the algorithms. Alternatively you could try and contact SSEC directly.
best
FIRMS Suport team.

Thanks FIRMS Support Team,
if you happen to get more details that would be helpful. Any reference documents (either ATBDs or articles) would also be helpful.

I will try and find an SSEC forum and bring up the same issue there to see if they have any more comments on the topic. Any updates I'll share here as well.
Conrad

Hi Conrad - I have been in touch with the team on the CSPP LEO forum and understand they are following up with you directly.
Let us know if you have any additional questions for us.