I spent some time looking at the two datasets, the control year, and the RED model where sea ice was reduced by 80% leading to several months where the Arctic would have no ice cover whatsoever.
I was considering how best to represent the areas where the model data was behaving most radically, and for clues as to how I could present these mutations musically.
It came as a surprise - though on reflection, of course, it should not have - to observe that the variations were, on paper, very small. The difference between maximum precipitation figures was the difference between 3.12 and 2.28 - the latter figure presenting a decline in precipitation. Similarly, the surface temperature in the model was 1K higher, from 275.8 in the control year to 276.8 in the RED.
Climatically speaking these changes are significant - but there is not a lot of movement in the core music system to represent this.
I discussed this with Lukrecia later in the day, whilst we both awaited out respective country’s games in the Euros. (We both drew. Croatia against the Czechs, and Scotland against England. I think our (Scoltand’s) draw was slightly more significant - but it will be curious to see what happens to our spirit of collaboration when our teams meet each other on Tuesday)
She first arrived me that these changes may look numerically small, but that they were most definitely significant. She also instructed me on how these were daily average readings for the whole of the Arctic, and that those averages are constructed with 6 hourly readings across a grid of reference points. Within that system there would be much more clear, sharp and defined readings. She grew a little concerned that our system of creating daily averages for the region was having the effect of smoothing out the information and making it too soft focus to be impactful.
She also reminded me that within these min / max ranges we had to also consider that the levels were different across periods of time - not just the min / max points. A sustained variance in the readings contains a lot of climatic impact.
We arrived at a number of possible solutions to address this. We considered switching the control of amplitude to the daily difference between the control and the RED reading - using increased volume to alert the ear to where the biggest (loudest) differences were. This could work, but it would be at the expense of amplitude showing where readings were generally low or high. I think we’ll progress by taking a separate data stream showing the difference, as a percentage, perhaps? - between the control and the RED, and I will use that figure to control an effect, perhaps reverb or echo, or some kind of modulating filter. The greater the discrepancy between control and RED - the more obvious and active the effect would be. This is exciting because it allows a new dimension to come in to play, rather than a repurposing of an existing function.
We also decided to create some new data exports which focus on particular individually notable variations. These will be smaller sketches where I run the control data and the RED data in parallel, with the two sounds running together to show how far apart from each other they become at certain points in the year, depending on whether or not there is ice coverage in the Arctic. We will be focussing on activity in northern Europe to specifically speak to the project directive to surface local impact.