So-Rad processing support

[insalt-glitch]

Hi,

my group and I have two So-Rad systems and are considering to use HyperCP for the processing (or at least compare the results to those of the manufacturer PML). I noticed that there is already some processing script for Level 1a but did not so see anything beyond that.

What is the major blocker here? I have written a script that can download the level 0 data (uncalibrated raw counts) in NetCDF (therefore also HDF) format, so we could use that as a basis together with the calibration files from the manufacturer of the RAMSES sensors (TriOS) :). Would you be interested in a solution like that?

Cheers Nils

[oceancolorcoder]

Hi @insalt-glitch ,

The current release (v1.2.17) can process So-Rad L0 HDF5 files. In fact, that was the impetus behind the L0 HDF export option, as well as firmware that allow HDF5 output directly from the platform prior to passing through the geoserver. We are still waiting for updates to the HyperCP README documentation for So-Rad from PML, but the usage is very similar to other autonomous platforms like SolarTracker, pySAS, and DALEC.

I recommend trying out the So-Rad sample configuration and dataset provided with the repository to familiarize yourself. Don't hesitate to reach out if you have questions.

Dirk

[insalt-glitch]

Hi Dirk,

thank you so much for the quick and super helpful reply! Very cool that the So-Rad system is already supported :). I suppose since we do not locally save our data, one can write a script that converts the from the Geoserver back to the expected HDF-format. Do you know whether such a script already exists?

Cheers
Nils

[oceancolorcoder]

Tom at PML (who I am unable to tag here, but will ping in a moment) is currently working with the HyperCP team on the code to get the geoserver to export the raw HDFs directly. He also may have a script for you to convert your existing geoserver output to HDF, but I'm not sure. I will ask him to chime in here.

[tjor]

Hi Nils & Dirk, @insalt-glitch @oceancolorcoder

Nils - Good to hear that you want to reprocess your So-Rad data via HyperCP.
Are these the 2 systems that AWI have?

Dirk has nicely summarised where I am at. I still need to create a `custom download script' (for the monda pypi repo), which harmonises with the new So-Rad L0 HDF5 format. I plan to test this on a few of our own deployments first, before we update the public version. It should be done soon, though, as I need to deliver our own data from this year with uncertainties by the end of October!

We know, in principle, that So-Rad produces some nice results with HyperCP from our work on the TARA TREC cruise: https://essd.copernicus.org/preprints/essd-2026-623. The main difference (from a normal TriOS system) is that So-Rad is typically sampling at a slower rate, but the sensors are synchronised.

Thanks
Tom

[insalt-glitch]

Hi Tom,

Nils - Good to hear that you want to reprocess your So-Rad data via HyperCP. Are these the 2 systems that AWI have?

Yes, the two systems are at AWI (I am with the PhytoOptics group with Astrid Bracher). I believe that the PML processing pipeline is not necessarily bad, so a comparison between the two could be interesting. However, HyperCP appears to (try to) be the community standard for above-water radiometry so believe there is value in reprocessing our data.

Dirk has nicely summarised where I am at. I still need to create a `custom download script' (for the monda pypi repo), which harmonises with the new So-Rad L0 HDF5 format. I plan to test this on a few of our own deployments first, before we update the public version. It should be done soon, though, as I need to deliver our own data from this year with uncertainties by the end of October!

If you look at the script I wrote. It just needs some minor changes to do what we want. If you want we can coordinate...

We know, in principle, that So-Rad produces some nice results with HyperCP from our work on the TARA TREC cruise: https://essd.copernicus.org/preprints/essd-2026-623. The main difference (from a normal TriOS system) is that So-Rad is typically sampling at a slower rate, but the sensors are synchronised.

Thank you for sharing this, I will have a look :)!

Cheers Nils