Brainstorming

1. first idea to use spatial distribution for the definition of features like variability, entropy, etc. to be used as features - maybe do this over quarters of the image
2. we should use the time series data, since rhythmical properties are expected to be in it.
3. interest is in the deviations from the average yearly shape. We decided to use an acoustic representation of the data series conditioned to that benchmark curve as follows:
   • if value is higher than average -> high pitched dust
   • if value is lower than average -> lower pitched dust
   • amplitude scales with absolute deviation from average
   • dust rate scales with absolute rain values
4. we can sonify different locations at the same time by assigning the sonifications of different locations to different channels. This should give us temporal rhythms if there are systematic dependencies between the locations.
5. data conversion is nontrivial in sc3 (as everywhere :-)