Habitat 2100
Predicting habitat loss in a changing climate
Overview
In my first Harvard MDE studio project, my partner and I created an interactive data visualization tool to examine how climate change impacts species across regions by modeling future habitats based on climate data. This interactive approach enables users to explore areas where species are most at risk and better understand the long-term effects of environmental change on biodiversity.
Toolkit
Miro, Figma, Maxent, Python, Three.js, After Effects
My roles
Research, data analyst, data visualization, wireframing, prototype, user testing, interface design
Team
Developer, designer (team of 2)
Duration
Sep - Oct 2024 (2 months)
Problem
According to the World Count Report, by the year 2100, half of all species will be at critical risk of extinction due to climate change. Once a species disappears, that loss is irreversible.
Our solution
We modeled habitat loss over the next 75 years, from present to year 2100 to examine how individual species will be affected by the changing climate in the context of ecosystem.
Research focus
To begin with, we focused our study on the Cascades mountain range in the Pacific Northwest, with Mount Rainier as a focal point. Centering on Pika, a small mountain-dwelling mammal that lives in the Cascades as they are the climate change indicator species.
But species don’t exist in a vacuum. We studied habitat loss within the context of ecosystems. This gave us a unique understanding of how the habitat loss of one species can threaten the existence of an entire ecosystem.
Gathering insights
We reached out to key individuals and organizations to gather insights, ensuring a well-rounded approach aligned with real-world conservation needs. Their feedback refined our focus areas and highlighted critical gaps, directing our efforts where they’re needed most.
Problem statement
How can we identify areas where species are most at risk from the climate change?
The big idea - predicting future habitat
To identify areas where species are most at risk from climate change, we modeled habitat loss for each species in the Cascades. By combining species location data with historical climate data and applying machine learning, we built a predictive model to define each species' “suitable habitat” under changing climate conditions.
Ideas development
We brought our ideas to life through sketching, prototyping, and multiple iterations, refining each design with insights gained from user testing and feedback.
Initial idea sketches
Gathering and cleaning data
Heatmap generation
Prototype, test, and iterate
Data transforming process
We sourced over 300GB of climate and species data from several sources such as GBIF, WorldClim, and NOAA.
The processed this data in Python and fed it into the Maximum Entropy machine learning model.
From the prediction scores, we generated heatmap images, which we overlaid on a map rendering in Three.js.
Lastly, we overlaid our 3D rendering with more data visualizations and UI components in Figma.
Visual elements
For the visual design, we used a dark navy background paired with bright colors to highlight important areas.
We kept everything as clean and simple as possible, to make sure that the focus stays on the map.
Final design
Timeline exploration
Scroll through a 20-year timeline, observing pika habitat shrinking due to climate change
Habitat interplay
Layer multiple species' habitats to better understand their interrelationships
Best-worst scenarios
Provide different future habitat scenarios based on the predicted climate models
Expanded timeline
Explore past and predicted climate data across various scenarios
Comparison mode
Allow users to compare scenarios on one screen and zoom in for more details
Full demo video
Key takeaways
Pika habitat shrinks
Pikas are forced to higher elevations as temperatures rise
Pikas & foxes diverge
Habitats barely overlap in the worst case scenario
Grasses flourish
Grasses species expand to both higher and lower elevations
Project's impact
Understand
how habitats will transform over time
Highlight
areas that will become unsuitable in the future
Guide
conservation efforts to areas that are most at risk
What I learned
Heatmap generation
Creating heatmaps with climate models and coding for the first time, thanks to support from my project partner!
Turning raw data to visualization
Gaining experience turning complex data into clear and impactful visualizations while preserving data integrity.
Adapting & iterating
Learning the importance of flexibility in design, continuously improving based on feedback and new insights.