SWOT-data-analysis

⚠️ Formalizing a collection of scripts I (Tatsu) have been using to analyze SWOT data. More to come hopefully. Contributions welcome!

Getting started

1. Create the conda environment from the provided spec: conda env create -f swot_python12.yml.
2. Set up your data-access credentials by running python gen_netrc.py, which writes a .netrc for NASA Earthdata and Copernicus Marine (used by the SST and regridding examples). To download SWOT data you will also need an AVISO+ account.
3. Download the SWOT orbit shapefiles with bash src/download_swot_orbit.sh (they land in orbit_data/). These are needed by the examples below to locate swaths.

Quick Examples

1. Find swaths using orbit shapefile. The first notebook gives an example of using the SWOT orbit shapefiles to find specific orbits. You can also use the .kmz orbit files provided by NASA to look at the tracks using Google Earth (our group's example here).

   

2. Download 1-day repeat swaths for the Kuroshio region. An example notebook for pulling SWOT data from the AVISO+ server over sftp. This works, but sftp is rather slow — see the THREDDS example below for a faster alternative.

3. Download swaths via the AVISO THREDDS server. A faster way to pull the same data using HTTP GET requests against the AVISO THREDDS catalog, which also supports parallel downloads. This is the recommended download path.

4. Plot 1-day repeat swaths with bathymetry. Once you are able to download the data plotting the actual swaths is fairly straightforward using cartopy and matplotlib. This notebook also gives an example of using some helper scripts to open and automatically subset the SWOT data (referred to as "data_loaders").

   

5. Calculate along-swath spectrum. Once we have a swath loaded we can go ahead and calculate the along-swath SSH spectrum to get a sense for SWOT's effective resolution and how the SWOT noise floor changes over time. I still need to add some plots here and get the notebook working smoothly, but the code is here if you would like to try playing around with it.

6. Regrid a swath section to a regular grid. Next, I've added some helper scripts from Scott Martin's NeurOST project to resample SWOT data to a regular lat-lon grid. These scripts will technically work on any xarray dataset with lat-lon coordinates.