Haskell “non-diffusive atmospheric flow” data analysis posts

  1. Reading NetCDF files Not really part of the main series of articles, but kind of critical!

  2. Outline and plan

  3. Reanalysis data and Z500 Atmospheric physics background.

  4. Exploring Z500 More background.

  5. Pre-processing Basic data processing tasks: subsetting, averaging, anomaly calculation.

  6. Principal components analysis Quick explanation of PCA with Haskell demonstrations.

  7. PCA for spatio-temporal data PCA for larger spatio-temporal datasets.

  8. Flow pattern distribution Using kernel density estimation to calculate a probability density function of atmospheric flows.

  9. Speeding up KDE Using CUDA to speed up kernel density estimation.

  10. Significance of flow patterns Bootstrapped significance calculations for “important” flow patterns found via KDE.

  11. Flow pattern visualisations Looking at the “important” flow patterns.

  12. Dynamics warm-up Autocorrelation and decorrelation before looking at Markov transition matrices.

  13. Markov matrix examples Simple examples of the kind of Markov matrix calculations we’re going to do.

  14. Markov matrix calculations Markov matrix calculations for atmospheric flow data.

  15. Wrap-up What we’ve done and lessons learned.