Create NetCDF file with irregular rotated lat-lon grid for QGIS

Question

I want to write data to a NetCDF file with an irregular rotated lat-lon coordinate grid and open it in QGIS. The data is on an irregular grid which follows a satellite swath. The grid is irregular in the sense that the distances between points can vary. Basically, I have a matrix of wind speed values ws10 at certain latitude and longitude coordinates. How do I properly save these ws10(lat, lon) to a NetCDF file, so that they can be viewed in QGIS projected on a world map?

This is what my NetCDF file header currently looks like:

netcdf waves {
dimensions:
        y = 52 ;
        x = 60 ;
variables:
        float lat(y, x) ;
                lat:standard_name = "latitude" ;
                lat:long_name = "latitude" ;
                lat:units = "degrees_north" ;
        float lon(y, x) ;
                lon:standard_name = "longitude" ;
                lon:long_name = "longitude" ;
                lon:units = "degrees_east" ;
        float ws10(y, x) ;
                ws10:long_name = "Wind Speed at 10 m" ;
                ws10:units = "meters per second" ;
                ws10:coordinates = "lon lat" ;
}

If I open the file in QGIS and add it as a raster or mesh layer (based on Dave X's comment) it takes the x-y-indices as coordinates. For some reason the grey-scale raster layer is flipped:

Yet Panoply projects the raster as intended: Based on other questions, I have tried to add the CRS information to the file as shown below, but QGIS still uses the x-y-grid. How can this be solved?

netcdf test_man {
dimensions:
        y = 52 ;
        x = 60 ;
variables:
        int crs ;
                crs:spatial_ref = "GEOGCS[\"GCS_WGS_1984\",DATUM[\"WGS_1984\",SPHEROID[\"WGS_84\",6378137.0,298.257223563]],PRIMEM[\"Greenwich\",0.0],UNIT[\"Degree\",0.017453292519943295]]" ;
                crs:grid_mapping_name = "latitude_longitude" ;
                crs:longitude_of_prime_meridian = 0. ;
                crs:semi_major_axis = 6378137. ;
                crs:inverse_flattening = 298.257223563 ;
        float lat(y, x) ;
                lat:standard_name = "latitude" ;
                lat:long_name = "latitude" ;
                lat:units = "degrees_north" ;
        float lon(y, x) ;
                lon:standard_name = "longitude" ;
                lon:long_name = "longitude" ;
                lon:units = "degrees_east" ;
        float ws10(y, x) ;
                ws10:long_name = "Wind Speed at 10 m" ;
                ws10:units = "meters per second" ;
                ws10:coordinates = "lon lat" ;
                ws10:grid_mapping = "crs" ;
}

Answer 1

I suggest using UGRID Convention NetCDF format to be fully supported by QGIS Mesh Layer, where you need to define your grid elements. Examples can be found in MDAL test suite. MDAL is C++ Library that QGIS uses to read the data for Mesh Layer.