30 year rolling mean

Design Rationale & Development

When iterating for the ideal final visualization, our main goal was to make complex environmental data more accessible and understandable to the readers through clear visual encodings as well as interactivity. For data transformation, we plotted the anomalies with the 18050-1900 baseline subtracted, which centered it at 0°C to compare to the historical average. We used a line chart as our primary mark, with the vertical position representing the temperature anomaly in Celsius over time (horizontal position). To emphasize the long-term trend over time, we created an interactive smoothing system for the reader to be able to turn the data into an estimation to make the overall trend more clear. We used this with a dual line system as the actual data was shown in faint grey and the reader’s selection (smoothed or not), in a bolder purple. Below the plot, we added Warming Stripes that use color as a channel which use a blue to red scale to instantly associate to cold and hot temperatures. The chart is fully interactive as the reader can drag across a specific time period and the corresponding histogram of anomaly temperatures (colored in the same cold-blue red-warm scheme) will automatically adapt to show the selected range. Interactivity is an essential design choice for our visualization as it uses brushing and linking to the Period Statistics chart. Lastly, a key feature is the Paris Agreement checkbox which adds a 1.5°C reference line in red and dynamically scales the y-axis. This gives the reader the control to view the data scaled and unscaled, while also being able to make comparisons between the 30-year data and the 1.5°C line. We had initially started with a singular line chart, realizing it was more effective to display both smoothed and unsmoothed data on the same plot. While we initially intended to simply show the average anomaly for the selected time period, our last iteration included the addition of a histogram to better represent the selected data.

Our development process was a collaborative effort which required around 20 people hours to complete the project within the two-week scope. We divided up the work based on technical strengths while all doing our best to adhere to the rubric. One team member cleaned and grouped the data (as well as setting up the GitHub & formatting the page) to be used by two team members who developed the main D3 code to create the interactive visualizations, and the last team member updated the page with the final write-up. After finally figuring out how to access the dataset, the most time consuming aspect was developing the histogram to be updated instantly upon line plot brushing by the reader.