Latent Datacapes
Through an interview with JPL research scientist Anthony Bloom, who focuses on the carbon data model framework (CARDAMOM), I learned that his team wanted to present carbon cycles in a way that felt fresh and accessible to the public. Conducting research based on data and effectively communicating it to a general audience require entirely different skill sets. Recognizing this gap, I explored various visualization methods that could transform complex carbon cycle data into something visually compelling and intuitive.
To understand the project
As a non-native English speaker, I found the terminology used by the scientists unfamiliar and the logical flow between concepts quite complex, which made it difficult for me to fully grasp the conversation. In order to synthesize the information, I had to listen to the recording multiple times and revisit parts of the discussion repeatedly. I often lacked confidence in whether I truly understood what was being said, which made communication challenging.
That said, our main client, Anthony, has repeatedly expressed a strong desire for the visualizations to incorporate the image of a tree.
“What most Cardamom users end up doing is making plots that look more like box and arrow..
Even though they look sciency, they're not intuitive at all. And then translating it into something like a tree, that's more like a friendly way to see
the carbon budget, is something that we probably do too late”
- From Anthony, during the weekly interview
Even though they look sciency, they're not intuitive at all. And then translating it into something like a tree, that's more like a friendly way to see
the carbon budget, is something that we probably do too late”
- From Anthony, during the weekly interview
Lower Engagement Caused by Complexity and Poor Accessibility
Daily meetings with teammates and mentors helped me become familiar with the terminology and core concepts. After several rounds of interviews, I understood Anthony’s reasoning for incorporating a tree into the visualization.
Scientists may feel they understand the data without visualizations, but visuals help them imagine circulation more clearly. For the general public, understanding the tool’s purpose and function proved challenging. Because the data spans multiple dimensions, it’s best to start with the big picture before zooming in to detect subtle or uncommon signals.
“How can we design CARDAMOM to be understandable and accessible to
non-experts, while still providing meaningful insights for climate science researchers?”
Hover to see CARDAMOM Data Variables and Structure
Narrowing down the Complexity
We aimed to create a visualization that would be professional enough for scientists to use, while also accessible to those without a scientific background.
To achieve this, we decided to make the design as intuitive as possible, using clear visual hierarchies and interactive elements to unpack and communicate the complex relationships within the data.
Diving into Particles
A key concept we came up with was visualizing carbon as particles to represent its movement and accumulation, drawing inspiration from natural elements like trees and petals to evoke the origin behind the numbers.
Bringing into Digital Format
I moved from static paper sketches to digital formats in order to capture the directionality and velocity inherent in flux data. Now, I’ve started creating visualizations using actual data as input.
Materializing the DATA
Among the tasks, I dedicated a lot of time to the tree visualization. At a high level, I visualized the massive dataset by categorizing it into two major flows — GPP and decomposition — which respectively represented the growth and decay of the tree.
However, the initial 3D version proved difficult for scientists to use in terms of numerical analysis, so I shifted toward a 2D approach that prioritized clarity, with numerical data displayed alongside the visualization.
Clove
: Interactive Data Visualizations for Carbon Cycle Science
Residence Time Tree
The Residence Time Tree uses a tree-shaped visual metaphor to illustrate carbon storage and turnover in ecosystems. Carbon pools like foliage, wood, roots, litter, and soil are spatially arranged to reflect their ecological relationships. The amount of carbon is shown through particle count, and turnover rate is represented by particle speed—faster particles indicating shorter residence times—making complex carbon dynamics easier to understand intuitively.
Capabilities
- Intuitive Mapping: Visualize variables on a tree diagram
- Residence Times/Turnover Rates: Show carbon dynamics
Visual Encoding
- Tree Diagram: Segmented carbon pools
- Particle Count: Carbon amount
- Particle Speed: Residence time/turnover rate
Shell Relationship Diagram
The Shell Relationship Diagram is a directional chord diagram designed to reveal how C moves between ecosystem components. C pools appear as vertically stacked circular nodes, while C fluxes are visualized as curved chords. Flux direction is spatially encoded.
Capabilities
- Establish Mental Model: Understand variable relationships and interactions
- Examine Overview: Observe high-level system changes
Visual Encoding
- Kriskogram: Directional chord diagram
- Circle Node: Carbon pool
- Semicircle Arc: Carbon flux
Aggregate Flux Petal Plot
The Aggregate Flux Petal Plot visualizes five carbon flux categories—respiration, production, decomposition, fire, and disturbance—using petal-shaped elements. It compares current values (filled petals) with historical monthly (dashed outlines) and overall averages (solid outlines), helping researchers identify seasonal patterns and anomalies at a glance.
Capabilities
- Relative Flux Changes: Track relative changes across flux groups over time
- Comparison/Anomalies: Compare to averages for anomaly detection
Visual Encoding
- Colored Fill Petal: Flux category (e.g., Respiration, Fire, Mortality, Production, Disturbance)
- Dotted Stroke Petal: Monthly average
- Solid Stroke Petal: Overall average
Garden Petal Plot
Garden Petal Plot presents a chronological grid of
monthly carbon flux snapshots across multiple years. Each petal represents a distinct flux category, with petal size indicating magnitude. The layout enables quick visual comparison across time to reveal seasonal trends, year-to-year variation, and unusual flux behavior.
Capabilities
- Time-Series Comparison: glanceable view of relative flux group variations over time
Visual Encoding
- Petal Plot Film Strip: Time series of individual petal plots