Using Generative Art to Convey Past and Future Climate Transitions

Summary

Here, a protocol is presented to visualize climate data as generative art.

Abstract

The ability to understand modern day climate relies on a foundational understanding of past climate variability and the ways in which the planet is stabilized by interconnected feedbacks. This article presents a unique method for translating records of past climate transitions preserved in deep-sea sediments to broad audiences through an immersive visualization. This visualization is a multimedia installation that incorporates geochemical records of glacial and interglacial transitions and model predictions for future anthropogenic warming to create an immersive experience for viewers, inviting them to engage with and reflect on the subtle, nuanced differences between subsets of Earth's history. This work showcases five intervals of time, beginning with the inception of modern glacial-interglacial cyclicity (~one million years ago), comparing past climate with model results for projected future anthropogenic warming (until 2099). The installation consists of several experimental projections, one for each subset of time, displayed on different surfaces in a room. As viewers move through the space, the projections slowly cycle through different climatic transitions, using animation methods like speed, color, layering, and repetition, all generated through site-specific data to convey the planet's unique behavior as it relates to global climate. This work provides a framework for unique scientific data visualization, with generative animations created using a Perlin Noise algorithm at the center of the installation. Research variables, like sea surface temperature, nutrient dynamics, and the rate of climate change, impact formal outcomes like color, scale, and animation speed, which are all easy to manipulate and connect to specific data. This approach also allows the possibility of publishing data online and provides a mechanism for scaling visual parameters to a wide variety of quantitative and qualitative data.

Introduction

Generative art and the methods employed here allow for the direct translation of quantitative data into animations while preserving the integrity of the data. Artists use generative art to explore perceptions of space and time^1,2, but generative art is not yet commonly used with spatial or temporal scientific data. The work presented here provides a simple framework for using generative visual products to showcase climate data. These products can be widely applied, whether used to create in-person exhibits or as a visual aid for a presentation or online publication.

Using geochemical measurements or estimates to scale elements such as color, shape, size, and speed provides a means of visually conveying rates and magnitudes of change without requiring the viewer to read a paper, interpret a graph, or look through a data table. Alternatively, the randomization of selected variables is used to convey a lack of data or uncertainty, as in the case of future projections. The juxtaposition of geologic past and future is perhaps integral to the effectiveness of these products as science communication tools. Recent experiences often serve as the baseline of comparison for modern climate change, making it difficult to grasp the magnitude of anthropogenic climate change³.

Geochemical measurements visualized in this paper span the mid-Pleistocene transition (MPT; 1.2 million to 600,000 years ago), recording changes near the northern boundary of the Southern Ocean from International Ocean Discovery Program Site U1475^4,5. The MPT data is presented in four animations, which highlight changes in ocean conditions as the planet cools and glacial and interglacial variability is amplified⁶. This provides a geologic baseline revealing the natural rhythm of Earth's climate, emphasizing a long-term cooling trend that starkly contrasts future climate projections. Future temperature estimates are average values of the results of 20 climate models under the forcings of Representative Carbon Pathway 8.5 (RCP 8.5; scenario with a radiative forcing of 8.5 W/m² in the year 2100) for the location New York, NY⁷. RCP 8.5 represents a worst-case scenario of sustained emissions resulting in a 3.7 °C increase in average global temperature by 2100⁸. Thus, this article demonstrates a means of comparing future projections with geologic data to compare rates of climate change and climate variability.

Protocol

1. Playing the existing visualizations

1. Download coding and visualization software (see Table of Materials).
   1. Download the data and code. This article uses 'degrees of uncertainty' with data from Marcks et al.⁴ and Cartagena-Sierra et al.⁵ on the age model from Starr et al.⁹.
      NOTE: The 'degrees of uncertainty' contains five coding files, Supplementary Coding File 1, Supplementary Coding File 2, Supplementary Coding File 3, Supplementary Coding File 4, and Supplementary Coding File 5, with contents pertaining to each time period of visualization (MPT 1, MPT 2, MPT 3, MPT 4, and Future, respectively). Each of these contains coding libraries¹⁰ used for visualizations as well as 'Script' folders containing downloaded data in .csv format, code used to generate visuals 'particle.js', and an index file 'index.html' which links all relevant data and code together.
2. Open the code editing software from the 'degrees of uncertainty'.
3. Drag a file (MPT 1, MPT 2, MPT 3, or MPT 4) into the code editor to visualize it.
   1. The files appear in the EXPLORER menu on the left-hand side of the window. Check the procedure for visualizing data from the 'Future' folder in step 1.7.
4. In the EXPLORER menu, click on the folder (MPT 1, MPT 2, MPT 3, or MPT 4) to reveal a drop-down menu, click on script, and then click on index.html.
   NOTE: The code appears on the right-hand side of the window.
5. Left-click the portion of the window with code for 'index.html' and select open with live server from the menu.
   NOTE: An internet browser window opens and begins playing the visualization.
6. Closing and reopening the code editor may be necessary between visualizations when loading a visual from a different subset of time. Repeat steps 1.4-1.6 for each subset of time.
7. To view the visualization based on future projections, open the 'Future' folder on the computer and drag either the 'Accumulation' or 'Transition' folder into the code editor. The difference between animations is described in the results section.
8. Select the folder name in the EXPLORER window and click on index.html. Left-click the portion of the window with code for 'index.html' and select open with live server from the menu.
   ​NOTE: An internet browser window opens and begins playing the visualization, which can be saved locally on a computer by screen recording.

2. Editing the visualizations

NOTE: To edit the visualizations, follow steps 1.1-1.4 above, as necessary, to load the relevant data.

1. Select the folder of interest in the EXPLORER window of the code editor and open the main script file by clicking on sketch.js.
   NOTE: The 'sketch.js' file in the MPT 1 (Supplementary Coding File 1) contains the most detailed annotations; thus, this file may be the most useful for familiarizing the code.
   1. The code appears on the right-hand side of the code editor window. Perform any edits to visualization parameters within this code. Look for code annotations with detailed descriptions of the code and its function following double slashes "//" and further identified by green text (Supplementary Figure 1).
   2. Define the variables that will be linked to data or used to customize visual parameters (Supplementary Figure 1).
   3. Load the data into the workspace (Supplementary Figure 2).
   4. Define the visual parameters of the canvas. Use a 'for' loop to link data to specific characteristics; here, size is linked to nitrogen isotopic value 'd15N' (Supplementary Figure 3).
   5. Use a for loop to define a tail length for each orb. The tail refers to the length of time that the orbs remain on screen after appearing, creating an accumulation of color as the visual progresses (Supplementary Figure 4).
      NOTE: Here, the tail length is scaled to the accumulation rate of alkenones' c37.
   6. Finally, draw the animation, applying a Perlin Noise algorithm¹¹ to define the shape of the visuals (Supplementary Figure 5).
      ​NOTE: Here, a circle is used as the base shape with noise applied to the points along the circumference of the circle. These will 'wiggle' the boundary of the circle, yielding an organic orb-like shape that deviates from a circle in an amount defined by the 'wiggle' command.
   7. Edit the code as necessary using annotations to aid the alterations.

3. Saving the edits

1. Save the edits by pressing the command and S keys at the same time.
2. View updated visuals by navigating to the 'index.html' file in the EXPLORER window, left-clicking, and selecting open with live server from the menu.
   NOTE: An internet browser window opens and begins playing the visualization, which can be saved locally on the computer by screen recording.

Results

This work produces six visualizations corresponding to five unique intervals of geologic time, with visual aspects scaled to quantitative data either measured on deep-sea sediment (Figure 1, Figure 2, Figure 3, Figure 4, Video 1, Video 2, Video 3, and Video 4) or modeled from the Intergovernmental Panel on Climate Change's (IPCC) RC...

Discussion

This work highlights the utility of generative art for the purpose of science communication. The workflow can be used to translate existing data to elements within an animation. While the animation outputs from this work are unique in that each time the code is run a different version of the animation is created, the visual elements are scaled to geochemical and climate model data; thus, elements such as color, speed, and size remain constant, so long as the input data remains the same. This also allows for the direct co...

Materials

Name	Company	Catalog Number	Comments
Easel	Uline	H-1450SIL	Telescoping easel to hold foam core board
Foam Core Poster Board	Royal Brites	#753064	Foam core board used as a canvas for projection
Live Server	Microsoft; Publisher: Ritwick Dey	Version 5.7.9	Software extension for Visual Studio Code which allows for viewing of animations in a browser window. Downloaded at: https://marketplace.visualstudio.com/items?itemName=ritwickdey.LiveServer
Throw Projector	Optoma	796435814076	Any model throw projector which will work for projection surface/distance desired
Visual Studio Code	Microsoft	Version 1.74 for MAC OS	Software for code editing and execusion. Downloaded at : https://code.visualstudio.com/