Output

The output of the FootNet model contains footprints, and reference information for matching corresponding receptors.

The output of the FootNet is a time-integrated footprint with dimensions (400, 400) with units of ppm /(µmol m-2 s-1). The output footprint has a spatial resolution of 1 km.

The footprints are provided as numpy arrays for easier integration with existing inversion frameworks or an option to save the footprints on disk.

Output format options

  • It is recommended to use numpy arrays if using FootNet on-the-fly in an inversion.
  • If the user prefers storing the footprints then it is easier to use NetCDF format.
  • User can also write their own storage functions to add any additional storage format.

Surface Measurement Example

from SurfaceFootNet import SurfaceFootNet

# Loading FootNet model
model_path = "../models/SurfaceFootNet_in_sample.pth"
model = SurfaceFootNet(model_path=model_path)

foots, reference_indices, reference_timestamps, reference_rlons, reference_rlats = model.run_inference(receptors, input_met)

Column Measurement Example

from ColumnFootNet import ColumnFootNet

# Loading FootNet model
model_path = "../models/ColumnFootNet_in_sample.pth"
model = ColumnFootNet(model_path=model_path)

foots, reference_indices, reference_timestamps, reference_rlons, reference_rlats = model.run_inference(receptors, input_met)

check out the complete code on GitHub