-
Notifications
You must be signed in to change notification settings - Fork 50
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Adapt graph generation script to general limited areas #4
Comments
I am trying out LAM for an academic project on one specific state of India. Looks like creating a mesh for the state will require me to provide a static nwp_xy.npy. Could you please help me out with how to generate this file for the area that I am concerned with? Does it contain Lambert projection? The values in the example nwp_xy.npy file clearly do not look like lat/lon coordinates |
Sounds like a nice use case! Indeed, the coordinates in
|
Hi, I am absolutely new to geographical projections. Could you direct me to resources to understand which projection I should use for epsg:7779? |
(Small caveat: I am probably not the right person to advice you on this, as I am far from an expert in map projections or GIS. You should probably ask someone who knows these kinds of details about your dataset.) I guess you could just feed your EPSG code to cartopy: https://scitools.org.uk/cartopy/docs/latest/reference/generated/cartopy.crs.epsg.html#cartopy.crs.epsg and get the right projection. A quick google gave me this: https://epsg.io/7779 which seems to indicate that it's a transverse mercator projection. Note however that the projection specified in |
Hi @nandini-21, I had to create this file too when trying Neural-LAM on my data. The steps to feed your data into Neural-LAM would be:
For my part, step 1 was made easy because I also use a Lambert projection, so I just had to edit the variable For the step 2, here is the snippet of code I used (inspired from this) to convert my grid coordinates from regular lon-lat (EPSG:4326) to the appropriate Lambert projection coordinates. It relies on
|
For the g2m and m2g we could use vector quantization (initial guess with known topology) or Kohonen SOM as suggested by @leifdenby Starting out with this regular mesh node distribution and topolgy |
The graph generation script
create_mesh.py
is currently written for the MEPS area. Without huge changes this should be possible to change to a generic script that can work for general (quadratic in their projection) areas. Some more arguments would probably have to be introduced.It might be a good idea to turn the graph generation into a function, and let
create_mesh.py
only be a utility script for calling this function with some command line arguments.The area definition object proposed in #2 could be useful as input to such a generic script.
The text was updated successfully, but these errors were encountered: