Using km instead of degrees for regrid?

Hello everyone,

I am trying to create a L3 product from TropOMI NO2 to compare a model VCD and the satellite. My model grid is in the GRIDDESC format (https://www.cmascenter.org/ioapi/documentation/all_versions/html/GRIDDESC.html), so it’s got some transformation elements to center the grid over the US, and it uses km to set the grid instead of degrees.

I’ve managed to use the bin_spatial() operation to create a rough estimate of my grid, but I was wondering if there was a way I could use this product to create an exact match?

Thank you.

HARP only supports grids using lat/lon axes. This is similar to the LATGRD3 = 1 grid type as mentioned at On The Definition of Grids and Coordinates for Models-3.
So, to perform the comparison, you would either need to have your model data on a lat/lon grid or (re)grid the tropomi data to your custom 2D grid (you did not specify what grid type your model uses by the way).

I am personally not that familiar with 2D regridding tools (and am also not sure how much accuracy you would loose when regridding your data). Perhaps something like gdalwarp — GDAL documentation could do this?

Hmmm, thank you for letting me know!

My grid is a 2D custom lambert conformal grid with 1.3 km spacing. I’ve worked around this issue by using the degree import with degree spacing that is nearly the same as my grid, and then I simply used a bilinear interpolation to the 1.3 km grid. There is almost no smoothing or distortion when I used this method, so I think it worked well for my purpose.

If you use 1.3km then you are oversampling the S5P data (S5P is about 3.5 x 5.5 km), so that would work without much distortion indeed.
But be aware that the S5P data is then a more ‘smoothed’ dataset compared to what your model may give.