Kalman filter based tracker
Implementing SVM training/matching against IBTrACS data
Ensemble member to ensemble member matching (i.e. not using best tracks)
Collection of extra fields around vorticity maxima tracks
Improve worker crash handling in pyro_cluster
Improve documentation
Reduce space used by results - compression
Run for 20 years worth of analysis on UCL computers
Settled on tracking vorticity maxima
Get Kevin Hodges TRACK code building (but don’t manage to get it running against data)
Experiment with Support Vector Machine (SVM) implementation in sckikit-learn
Implement a basic Kalman filter and check that it is producing reasonable data
Smoothing and upscaling of vorticity/pressure data