Hello,
I am a 77 year-old physical oceanographer interested by a dramatic low having been observed on the Polynesian Maupiti Island (16.4 °S-152.3 °W) exactly during the SWOT path of Dec. 30, 2023.
I fully rely on the SLA L3 data accuracy at the 250-m pixel level that, I believe, is demonstrated by permanent small scale structures such as the bean-shaped low in the west of the island displayed on the https://swot-calval.oceandatalab.com/ site. Unfortunately, these permanent small scale structures demonstrate that the considered geoid is not fine enough and I would like to perform averages of the SLA L3 data as they are computed and presented to try specifying them and inferring the oceanic signal (considered to be the small scale variable part and not considering the large scale part due to the tides and the atmospheric pressure) in a relatively small latitudinal interval (16.4 to 17.0 °S) where I expect eddies to have possibly occurred
Could it be possible to obtain Excel files of the SLA L3 pixels in the form Lat-Lon-SLA for a dozen descending paths, ideally without any error, in this latitudinal interval?
Since this would necessitate some extra work, I could specify the dates of these convenient paths and send a manuscript presenting the observations and what has already been done with SWOT data (those on Dec. 30, 2023 are the lowest ever observed) and nadir data (that indicate a 200 km eddy nearby) with the help of a colleague from LEGOS who has no more time to help me.
And we could exchange directly using our e-mails...
With my many and sincere thanks in advance.
Sorry for the late reply, there weren't many people available at ODL the last few weeks due to holidays and conferences.
I downloaded files with an even pass number (descending) from the SWOT L3 Unsmoothed (250m) product for Dec. 30, 2023 and extracted "ssha_filtered" values located in the [-17.0, -16.4] latitudinal band. The "ssha_filtered" variable is described as:
Height of the sea surface anomaly with all corrections applied and with calibration, data selection and noise reduction (using Unet model) applied
Saving these values in an Excel file was not convenient, so I stored them as CSV files (should be easy to import in Excel) named after the NetCDF file values are extracted from. As these files are quite big, they have been compressed in a zip archive available here: https://ftp.odl.bzh/odl/sherleda/maupiti...30_sla.zip
I repeated the same process for all descending passes nearby French Polynesia between Dec. 16, 2023 and Jan. 23, 2024 (thus covering two SWOT cycles: 008 and 009), the results are available in this zip archive: https://ftp.odl.bzh/odl/sherleda/maupiti...23_sla.zip
Please note that:
the pass over Maupiti on Dec. 30, 2023 (pass #444) is included in both zip archives
Longitude values are expressed in the [0, 360]° range (same as in the NetCDF files)
I hope I understood correctly what you wanted and that these files will be helpful.
Hi Sylvain,
Many sincere thanks for what you have done for me.
I am able to read data with the two links you provided but it is clear that I did not correctly explain which files I need so that I think it will be more efficient to better explain.
I am preparing a detailed reply I will be able to send in less than a couple of hours... just hopping you are not in summer holidays!
A am very happy,
Sincere thanks,
Claude
(2025-06-30T04:42 AM)Claude Millot Wrote: Hi Sylvain,
Many sincere thanks for what you have done for me.
I am able to read data with the two links you provided but it is clear that I did not correctly explain which files I need so that I think it will be more efficient to better explain.
I am preparing a detailed reply I will be able to send in less than a couple of hours... just hopping you are not in summer holidays!
A am very happy,
Sincere thanks,
Claude
Hi Sylvain,
Hoping you are still "in operation", please consider the attach doc and let me know.
With my very kind regards,
Claude
Please note that there is a limit to 10MB for .doc and .docx files.
If it still doesn't work, feel free to send me the file by email (I sent you a message on Friday so you should have my address, maybe it ended up in the spam folder, let me know if you did not receive it).
(2025-06-30T04:42 AM)Claude Millot Wrote: Hi Sylvain,
Many sincere thanks for what you have done for me.
I am able to read data with the two links you provided but it is clear that I did not correctly explain which files I need so that I think it will be more efficient to better explain.
I am preparing a detailed reply I will be able to send in less than a couple of hours... just hopping you are not in summer holidays!
A am very happy,
Sincere thanks,
Claude
Hi Sylvain,
Hoping you are still "in operation", please consider the attach doc and let me know.
With my very kind regards,
Claude
Hi Sylvain,
On July 1st, I have done what, seems to me, I did on June 30... may be forgetting to "Add Attachment
I have attached a pdf named "To Sylvain on June 30, 2025".
Have a nice day.
Claude
2025-07-02T07:37 AM (This post was last modified: 2025-07-02T07:41 AM by Sylvain Herlédan.)
Hi Sylvain,
Please consider I am not in a hurry and appreciate very much your efforts.
I just want to specify that I read mi e-mails only in the morning.
Feel free to ask me any question or remark.
Au besoin, je vous précise mon adresse-mel (REDACTED) et nous pouvons aussi échanger de vive voix.
Bonne journée,
Claude
Thank you for the document describing the context, it makes it easier to understand what you are trying to achieve.
To a software engineer (remember that I am not an oceanographer, I have no expertise in altimetry), your approach seems sound, computing a mean over time should approximate the constant part of the signal, including the unaccounted geoid error hinted by the plots shown in your document.
As you mention being able to recover the SWOT data by yourself, can you be more precise on what kind of help I can provide? Computing the mean map and sending you the resulting values in CSV format like before?
PS: I redacted your email address to prevent it from being collected by spam bots (I already had it, I sent you an email last Friday to notify you about the forum reply, the mail subject was "Données de SLA SWOT Maupiti 2023-12-30", so you should have my email address as well).
I have already agreed to help you, but you said you were not in a hurry and there have been several incidents last week that I had to deal with urgently, so I had to prioritize.
Also, I started getting the required SWOT data on Friday, but downloading them from the AVISO website takes quite a long time (maybe Vinca had a more direct access), about 30 minutes for each pass. Downloads have completed during the week end and now I have all the descending swaths passing over Maupiti, so I will be able to start working on them and produce the maps you want during this week if everything goes well.
2025-07-07T08:35 AM (This post was last modified: 2025-07-07T08:45 AM by Claude Millot.)
Hi Sylvain,
Many sincere thanks for all what you are doing for me.
I am sorry to be unaware of all the difficulties you encounter to get the adequate files.
I hope what we are doing will be expressive, giving a realistic geodesic structure (from the the mean) and a realistic oceanic structure (from the corrected Dec 30 swath) ... and conclusive, allowing to hypothesize a kind of oceanic phenomenon.
Because I am a lucky guy (as demonstrated by the fact I met a so kind, open minded and efficient "software engineer"), I am very confident.
Best regards,
Claude
Sylvain,
Discussion with a person like you allows me clarifying my ideas.
In what I proposed, it is expected that all small scale geoid features are measured during all swaths (computing a mean makes sense) and that a major oceanic event was measured only on Dec. 30., so that oceanic signals are relatively low only during the other swaths.
I thus think that it can be easily justified to compute the mean from all available swaths... except the one on Dec. 30. (the mean-geodesic signal will thus be modified as less as possible by the oceanic signals).
2025-07-07T08:04 PM (This post was last modified: 2025-07-07T09:03 PM by Sylvain Herlédan.)
Hi Claude,
1) Computation of the mean SLA
I downloaded pass #444 for the 28 cycles currently available in the SWOT L3 250m product.
Out of these, two of them had no valid value at all for the sla_filtered variable, so I left them aside:
2024-02-09
2024-06-14
The file for 2023-12-30 has not been included in the computations for the mean SLA.
For the remaining 25 files, the values have been interpolated on the same grid as the 2023-12-30 pass, here is the result:
The mean SLA takes into account the fact that satellite passes had missing values in different locations. Here is a plot of the number of samples included in the mean computation for each pixel: https://ftp.odl.bzh/odl/sherleda/for_cla...231230.png
2) Original SLA on 2023-12-30
The content of the sla_filtered variable found in the SWOT L3 250m file for pass #444 on 2023-12-30 is available here as a map and a CSV file:
3) (SLA - mean) on 2023-12-30
When the mean map computed in step 1) is subtracted from the 2023-12-30 SLA extracted in step 2), only small scale variations remain:
All values are negative, so the initial range is not suitable anymore. I made another plot after adapting the min/max to see more details (previous plots use [-0.05, 0.20] whereas this one uses the [-0.25, 0.0] range):
One last version where the range of the colormap has been narrowed down to [-0.15, -0.05] so that details in the immediate vicinity of the island stand out more: https://ftp.odl.bzh/odl/sherleda/for_cla...trange.png
Please let me know if there is something I misunderstood, or if these plots need to be changed/improved (I could add titles and units to make them easier to interpret).
2025-07-08T11:40 PM (This post was last modified: 2025-07-09T06:39 AM by Sylvain Herlédan.)
Hi Claude,
The plots have been generated again with titles and units, both for the left and right swaths (they are named as such in the SWOT documentation). In order to simplify the visualisation of he plots, I put them here: https://ftp.odl.bzh/odl/sherleda/maupiti_sla/2swaths/
Please note that since the satellite is heading toward South during descending passes, the left and right swaths are respectively the eastern and western swaths once represented on a map.
What you did and show as "Both Swaths" for Dec.30 is perfect.
Now that we have a correct (I think) idea on the shape of the lowering on Dec.30, we could easily, considering what you already did, i.e. considering the same mean, compute and display the corrected data for the closest descending paths on Dec.12 and Jan.9 (if these are the correct dates). Maybe we will see an especially intense lowering somewhere and will find arguments to make a link between a lowering on 2 dates, thus infer a trajectory.
Therefore, what do you think about providing what you entitled "Both Swaths", for two other dates with the same color scales?
I kept the same scales for now, as the small scale SLA does not seem to saturate too much in the first plot (using [-0.1, 0.1] m range), while the other two plots can reveal more details for negative anomalies (using [-0.15, -0.05] m range). I can change that later on if needed.
