I worked on this project with together with 2 developers.
Climate change is one of the major global challenges of the 21st century. If we do not curb greenhouse gas emissions and keep global warming well below two degrees Celsius, the world will change permanently. We are already seeing global impacts such as shifting vegetation zones, rising sea levels and increasingly extreme weather periods. An important aspect of combating climate change is education and information. It is necessary to show how the climate has already changed over the past 250 years - how the climate is likely to change based on our actions today. Climate scenarios are an important tool for this. Climate scenarios allow us to make statements about possible futures. Based on complex ecological and socio-economic models, we can estimate the positive as well as negative impacts of our current actions. However, understanding and interpreting climate scenarios is complex and difficult for non-experts to access. To make climate scenarios more understandable and usable, we want to create data visualizations that allow different user groups to explore, understand, and use the scenarios.
The goal was to be able to compare several SSPs ("Shared Socioeconomic Pathways", different future scenarios) at a glance. We wanted to make this possible with the help of an online tool that would allow experts in particular to communicate the concept of SSPs.
When we came across the SSP database and were presented with some of the data sets, we quickly realized that it was particularly difficult to compare the individual SSPs or scenarios. For example, it was possible to see at a glance which scenario predicted more or less gas consumption or population growth. However, as soon as more complexity was allowed, the graphs lost clarity and were not displayed appropriately. By adding more SSPs or sectors, the visualization became overloaded and too complex. In addition, the operation was far from optimal.
We tried to solve this problem in our group, and in the first weeks of our collaboration we were primarily concerned with finding the form for our SSP fingerprint. Which forms of representation are best suited? At what point does it become too complex? Which colors can potentially bring value to our fingerprints?
Right from the start, we were particularly interested in a modified star diagram. The combination of a distinctive shape and the potential of overlaying (and thus the ease of comparing multiple) convinced us, and so we developed an initial prototype to test with a real data set whether our hunch was correct.
After the first round of feedback in the course, it quickly became clear that several iterations were needed to fix the weaknesses in the display format:
The so-called RCPs (divergent versions of the SSPs that result from policy decisions) could not be represented efficiently.
The representation felt too simple, and feared that we were representing the subject matter too low-complexity.
It was back to the drawing board for us, breaking away somewhat from the star diagram and trying to develop our own individual version of the star diagram. For this we worked simultaneously in analog, in sketch and in code.
In the end, we settled on two forms of representation:
Our star plot, which, with the help of "split arms," the RCPs could also display.
A modified version of the box plot, which could also represent the RCPs.
We wanted to display these diagrams in the style of the Small Multiple in an online tool. This should allow to compare the different SSPs (including their RCPs) at a glance.
We took a very experimental approach to the implementation of the project. Since our design was dependent on the implementation and the actual data sets, we kept working in the code in parallel with the design process. This allowed us to validate and reject our assumptions over and over again. At times, this iterative method threw us backwards, and we had to find new approaches.
For technical implementation and prototyping, we used React, styled-components, D3, and react-faux-dom.
The result is an online tool that allows the exploration of our Small Multiple. We also designed a poster for the 2018 Werkschau that documented our result.