12/20 Meeting Notes

[panalluri]

Super-res CNN
- For urel, vrel, qref, look at stats to understand how well CNN learned (e.g. R2 error)
- try predicting the anomaly fields
- plot true and predicted anomaly fields
- mess with SST, slp, tref to get them more right
- look at the last super-res paper dhruv sent --> reproduce these results
- consider using these methods for my work
- look up how to do R2 and correlation to evaluate NN performance

Diagnostics project:

• take mean out of power spectra calc
• figure out how to take care of land cover affecting results
  • choose an area of a few thousand km in the pacific
  • try using a strip in the drake passage
• plot HR/LR w same colorbar
• window=True for cases without periodic behavior --> choose wisely
• NEW RULE: if anything takes over 15 mins, find some alternative way to make it faster
• make shorter movies w much higher time resolution
• make gif movies based off dhruv's suggested code