Meeting Notes 12/27

[panalluri]

Diagnostics project:

• Fix colorbars in movies
• Make Drake Passage sample taller and go around the whole earth
• Figure out power spectra x and y axes units + describe to Dhruv
• Average over more timesteps (5-10 spaced out by 5 days) for power spectra
• inspect power spectra behavior --> use wavenumbers to match to features on maps
• Make velocity movies with a divergent colormap (red, white, blue)
• Make power spectra, movies, snapshots for HR for various filtered scales
• Pick 2-3 variables (e.g. sst, wind) --> plot power spectra at various filter scales on the same plot
• Calculate air-sea fluxes for HR case in the drake and pacific regions

Super-resolution project:

• Deep Learning course

• use evaluation metrics that are sensitive to behavior across many scales

• plot power spectra of the target-prediction pairs

• adjust kernel size so frame around prediction maps goes away

• finish mini lit review -->

  • come up with next steps for super-res project
  • figure out which paper to reproduce results from
  • compare and contrast papers against each other

• for reference: steve brunton --> stats and ML videos on youtube (look at his probability and stats)