search

radical-collaboration / facts

Repository for the Framework for Accessing Changes To Sea-level (FACTS)
MIT License
21 stars 10 forks source link

Develop emulandice2 module #253

Open bobkopp opened 11 months ago

bobkopp commented 11 months ago

We hope that emulandice2 will work natively with FACTS climate input and output FACTS-compative output. Will have more sectors for the AIS and GrIS that need to be tracked, ideally flexibly, and will have associated fingerprints. Tamsin will work with Alex on this.

bobkopp commented 9 months ago

@AlexReedy I think the Python code for this module can be pretty light, given that emulandice2 now works directly with FACTS-type files. We ought to be able to use a single project script with different modes for AIS, GrIS, and glaciers, and a single postprocess script that uses a configuration file to map ice source + region to fingerprint files. We ought to be able to pass info between the scripts using the netcdf files outputted by the project script, as opposed to a pkl file. The postprocess script set up in this generic way also helps pave the way for a general grd fingerprint module.

bobkopp commented 9 months ago

We have a prototype project script for emulandice2/GrIS that seems to be working, but needs to be adjusted as @tamsinedwards adds in a simpler way of specifying the location of the climate_data_file, the emulator file, and the output name (including pipeline name in the filename) -- the prototype uses cludges for working with these right now.

I added some thoughts on the structure of the new postprocess module to the emulandice_postprocess.py file.

bobkopp commented 9 months ago

emulandice_postprocess.py seems to run, though results have not been checked.

bobkopp commented 9 months ago

Need to add functionality to emulandice2_project to (1) interpolate onto years defined by the --pyear parameters and (2) rebase to the year specified by --baseyear.

bobkopp commented 9 months ago

Interpolation functionality added.

bobkopp commented 9 months ago

Current status: I think the Python code may be done, but requires inspection of output vs expected output. The Box.com modules-data directory has working built emulator Rdata files for 19 RGI glacier regions, but we need these for AIS and GrIS.

bobkopp commented 8 months ago

Current status: Box.com modules-data directory has working built emulator Rdata files, but these are priors, not conditioned on observations. Code itself should work but diagnostics need to be generated.

bobkopp commented 1 month ago

@AlexReedy Is this all working in the development branch now?