Replication of sen2cor correction to L1C data

s2tbx sen2cor
#1

Dear Step Forum community,

I am new in remote sensing use and I aim only at doing very simple NDVI mapping and analyses. I am focusing on a zone that is included in one tile: T37NBA (in Kenya).

I want to create a NDVI time serie and to do so I have downloaded data from the map on the Copernicus Scihub. My time serie extend before and after 2018. This means that I have been able to download L2A products from January 2019 but only L1C products before it. I then wanted to reproduce the corrections manually with sen2cor in SNAP in order to have complete my dataset. My problem is that I am a bit lost in the version to use and the settings to have. Also, as many people already highlighted, the GIPP xml file cannot be found.

My wish is to replicate the exact same correction as the one that have been done on the Scihub L2A products for my tile. In some other topics it seems the settings are all default (Sentinel-2 L2A products available on Copernicus Open Hub). I wonder what is default exactly because the default resolution in SNAP is 60m while in my L2A products it seems that I have all resolutions and I am working at 10m bands.

So my question is the following: with which version of sen2cor and with what settings should I correct my L1C data in order to have a consistent dataset with my L2A data?

Many thanks for your help,
Sincerely yours,

#2

Why not, instead, download L2A products from the Google Cloud - starting from here:

https://console.cloud.google.com/storage/browser/gcp-public-data-sentinel-2/L2/tiles/37/N/BA

gcp-public-data-sentinel-2/L2/tiles/37/N/BA

#3

Dear gbrelstaff,

Thanks for your answer.
This does not resolve my problem. I have already downloaded the 2019 and 2020 data in Copernicus (which are the same as the one on the Google Cloud). My problem is that, as you can see, no L2A exists before December 2018. So I have to download the L1C data, which I did, and I just want to be sure to apply the same sen2cor procedure to these L1C products but I don’t find the version of sen2cor and the settings that have been used for 2019 and 2020 data.

Thanks

#4

The way Google sort the product (not using proper date order)
makes it hard to find the data from 2018 yet here is an example of one folder:

https://console.cloud.google.com/storage/browser/gcp-public-data-sentinel-2/L2/tiles/37/N/BA/S2B_MSIL2A_20180522T074719_N0208_R135_T37NBA_20180528T145257.SAFE

there are many more

#5

Dear gbrelstaff,

Many thanks again. Actually, when I look at it, I only find some products from December 2018 and the one you spotted from May 22th. Do you spot other ones from other months? It seems to me that the products here corresponds to the products available in Copernicus and it doesn’t resolve my problem…

#6

Okay you’re right there aren’t very many for 2018 at Google cloud. And that one from May is mostly cloud!

#7

Well, the people who run the EO Browser seem to have copies of the L2A products you want.
For example:

but as far as I recall they charge for access to the full products rather than just their visualization in the browser.

#8

Many thanks, I will have a look at this.

#9

Hello drozenrechels,

unfortunately you will find on SciHub no homogenous time-series as you look for. The data you can download are processed with very different Sen2Cor-versions and especially in 2018 there had been major changes. You can find more details in [B. Pflug et.al.,“Next updates of atmospheric correction processor Sen2Cor”, Proc. SPIE 11533, Image and Signal Processing for Remote Sensing XXVI, 1153304 (20 September 2020); https://doi.org/10.1117/12.2574035]. If this is not sufficient don’t hesitate to contact me again.
Data on Google or other archives are just a copy of the data on SciHub. Therefore these archives cannot solve your problem. Reprocessing your data with the latest Sen2Cor version is the only option you have. As always, I recommend to do it not from SNAP, but from command line tool. You have complete control about configuration for your processing and it can run in background at the same time as you are analysing other data in SNAP.
Data archived at SciHub are processed with the same default configuration as is provided with the public Sen2Cor. The only difference is that you cannot use Planet DEM as is used for the products on SciHub. However, you can use SRTM-DEM which should be equivalent for Kenya.
But reprocessing your time series with SRTM-DEM may be a problem because there was a format change in summer. This means, that nothing happens if you have already downloaded the DEM tile with earlier processing. If not, then the only option is waiting for release of Sen2Cor 2.9 which is planned for Q1/2021 actually.
I am wondering that you don’t find the L2A_GIPP.xml configuration file. It should be located in you Sen2Cor home directory.

I hope that I could you help you a little
BdPg

#10

Dear BdPg,

Many thanks for this answer.
I will study all of this. It seems to me reading your response that the best I can do is download L1C data only and process all of them myself to ensure an homogeneous atmospheric correction across months and years to allow relative comparison.
The L2A_GIPP file that I didn’t find was the one I thought I was suppose to find in the GRANULE > AUX data of the L2A products after processing. But of course I have the one linked to sen2cor and I tried the program with command lines and it works fine.

Many thanks again, I will go forward with this, and if I face an issue again, I will come back to you.
Best,
D