Visualizing future of climate science,�one dataset at a time

- Valerio Pascucci, Aashish Panta, Amy Gooch, Giorgio Scorzelli

Held annually as part of the IEEE VIS Conference

Acknowledgements

Cyberinfrastructure

Climate Data

Redundant Storage

$1,000 Cash Prize !!!!

OSDF

NSDF and WIRED awards

What is the SciVis Contest?

• Annual contest part of the IEEE VIS Conference, with a focus on scientific visualization

​

• Encourages interdisciplinary collaboration across fields such as data science, climate science, and computer graphics.

​

• 2026 theme: Visualization of a peta-scale climate data from NASA, addressing critical topics like climate change, oceanic dynamics, and extreme weather events.

​

• Partially based on 2024 IEEE LDAV Best Paper Award: ‘Web-based visualization and analytics�of peta-scale data: Equity as a tide that lifts all boats’

​

​

• Dataset Focus:
• DYAMOND Models: Global integration of ocean-atmospheric models .
• NASA's ECCO LLC4320 Model: Oceanic circulation data at high temporal and spatial resolution.
• NEX-GDDP-CMIP6 data: Downscaled climate projections derived from the CMIP6 models, providing finer spatial details suitable for regional climate impact studies

​

• Challenges:
• Real-time rendering of Peta-scale datasets
• Interactivity and user-driven exploration
• Addressing visual complexity in climate simulations
• Best Submission wins cash prize of $1000! (sponsored by ViSOAR)

​

​

​

Focus of 2026 Challenges

• Three Major Datasets, totaling 4 petabytes

​

• NASA ECCO LLC4320 Data

​

• NASA DYAMOND Data

​

• NEX-GDDP-CMIP6 Data

Data Available

• DYnamics of the Atmospheric general circulation Modeled On Non-hydrostatic Domains (DYAMOND) data
• Total Size: 1.8 PB
• Timeline: 14 months, starting from prescribed initial conditions on January 20, 2020.
• Dataset includes:
1. C1440 configuration of the Goddard Earth Observing System (GEOS) atmospheric model
2. LLC2160 configuration of the MITgcm with 2–4-km grid spacing, ocean model

​

NASA DYAMOND Data

• GEOS model
1. Uses cubed-sphere grid; 6 faces available for each fields

​

• Dimensions: Latitude, Longitude, Depth

​

• Shape: (1440, 1440, 51)

​

• Available Fields:
1. mid-level pressure, wind velocity in 3 directions, air temperature, carbon monoxide, carbon dioxide, liquid cloud particle, potential temperature due to convection, mass fraction of cloud ice water,�mass fraction of cloud liquid water.

​

​

NASA DYAMOND Data (Contd.)

Image taken from https://thatsmaths.com/2012/10/18/carving-up-the-globe/

GEOS model is atmospheric data model projected on a �cubed-sphere.

Users will query each face individually based on their needs.

​

​

​

DYAMOND Geos Data

Image taken from https://thatsmaths.com/2012/10/18/carving-up-the-globe/

How to access DYAMOND Geos Data

1. Import all required libraries (install them with pip if not found)

​

1. Set the variable, timestep, face, �depth and resolution of the data you want to access

​

1. Apply the read function; the output�will be a numpy array and can be �used in any analysis notebooks

• LLC2160 MITgcm model
1. 2–4-km grid spacing and 90 vertical levels, on a lat-lon-cap grid. 12 faces, each of shape 2160*2160, attached together in a specific way on a lat-lon grid. Depth from 0.5m to 6km down the sea level

​

• Dimensions: Latitude, Longitude, Depth

​

• Shape: (8640, 6480, 90)

​

• Available Fields: east—west velocity, north-south velocity, vertical velocity, salinity, temperature

DYAMOND LLC2160 Data (Oceanic)

Grid structure:

1. Install required python libraries using pip:

​

1. Set the variable, timestep, depth and resolution of the data you want to access

​

1. Apply the read function; the output�will be a numpy array and can be �used in any analysis notebooks

How to access DYAMOND LLC2160 data

https://sciviscontest2026.github.io/examples/

• A 14-month global simulation of the ocean (September 2011 to November 2012) that resolves to capture complex ocean dynamics
• Total Size: 2 PB, each timestep: >30GB
• Dimension: 90* 17280*12960
• The dataset includes various oceanographic variables, each available at 90 different depth levels
• Fields: vertical velocity, temperature, salinity

NASA ECCO LLC4320

Follows the grid-structure of LLC2160

from MITgcm Model, but twice the resolution

1. Install required python libraries using pip:

​

1. Set the variable, timestep and resolution of the data you want to access

​

1. Apply the read function; the output�will be a numpy array and can be �used in any analysis notebooks

​

1. We are still in the process of moving all fields for LLC4320 data

How to access LLC4320 data

https://sciviscontest2026.github.io/examples/

Overview:

• Downscaled Climate Projections: Derived from CMIP6, downscaled to 0.25° x 0.25° (~25 km).
• Time Range: Historical (1950-2014) and Future Projections (2015-2100).
• Local Topography: Captures finer climate gradients influenced� by geographical features.
• Resolution: 0.25° x 0.25° horizontal grid (~25 km)
• 30 models, 3 scenarios, 8 variables
• Shape: (365, 600, 1440)
• Size: 38 TB

​

Variables: 8 variables including air temperature, precipitation, �relative humidity, shortwave/longwave radiation

​

​

​

​

​

NEX GDDP CMIP6

1. Install required python libraries using pip:

​

1. Set the model, scenario, variable, timestep and resolution of the data you want to access

​

1. Apply the read function; the output�will be a numpy array and can be �used in any analysis notebooks

​

1. Access the data from the dashboard: https://nasa-arset.nationalsciencedatafabric.org/

​

How to access NEX GDDP CMIP6 data

https://sciviscontest2026.github.io/examples/

Task 0: Exploring climate change with NEX_GDDP_CMIP6 data

​

Warmup Objective: Visualize how the air temperature varies over time from 1950 to 2100

​

• Participants should try creating simple visualizations to show how the air temperature changes over time

​

• See how the humidity varies across different regions on the globe

​

• Check out this dashboard for more details: https://nasa-arset.nationalsciencedatafabric.org/

​

• Dataset: NEX GDDP CMIP6

Task 1: Visualizing Wind Patterns in the Atmosphere

​

Objective: Visualize global atmospheric wind patterns using the eastward (U) and northward (V) wind velocities.

​

• Participants will generate vector field visualizations showing wind direction and magnitude near the Earth's surface or at mid-atmospheric levels.

​

• Deliverable: A dynamic wind field visualization showing atmospheric circulation patterns.

​

• Dataset: DYAMOND GEOS Data

​

• Key Variables: U (eastward wind), V (northward wind), W (vertical Velocity)

Task 2: Visualizing the Interaction Between Atmosphere and Ocean in Extreme Weather Events

​

Objective: Explore and visualize the interaction between atmospheric dynamics and ocean conditions during extreme weather events, such as tropical storms or hurricanes.

​

• Participants will focus on a region and time period where extreme weather events occur (e.g., tropical cyclones or monsoons) and visualize the interplay between atmospheric wind patterns, temperature, and ocean currents or temperature.

​

• Dataset: NASA DYAMOND data

​

• Deliverable: A visualization or interactive tool that shows the coupling between oceanic and atmospheric dynamics during extreme weather events

​

• Sample usecase here: http://chpc3.nationalsciencedatafabric.org:11957/run
• More on next slide

Task 2: Visualizing the Interaction Between Atmosphere and Ocean in Extreme Weather Events

​

• Atmospheric Data: Use variables such as U (eastward wind), V (northward wind), P (mid-level pressure), and T (air temperature) to analyze the atmospheric structure of the storm.

​

​

• Ocean Data: Use variables like theta (sea-surface temperature), u, v (east-west and north-south ocean velocities), and salt (salinity) to show how ocean conditions influence or respond to atmospheric changes.

​

​

• Participants should emphasize how atmospheric wind patterns interact with sea-surface temperatures and currents, exploring questions such as:
1. How do sea-surface temperature anomalies fuel extreme storms?
2. What are the effects of atmospheric wind shear on ocean currents during a storm?

​

​

Task 3: 3D Ocean Circulation Patterns with Depth

​

Objective: Visualize the complex 3D structure of ocean currents by mapping velocity fields (u, v, w) at different depths.

​

• Participants should generate layered visualizations that show how ocean currents change with depth. They can create 3D vector field visualizations or layered 2D slices at various depths.

​

• Encourage visualizations that highlight major oceanic circulation systems, such as thermohaline circulation, and their vertical structure.

​

• Key Variables: w ( vertical velocity), theta(temperature), salt (salinity)

• Feel free to explore different variables across different regions
• Objective: Participants are invited to explore any of the provided datasets in creative and innovative ways. The goal is to come up with novel visualizations that offer new insights into the data, without being restricted to a specific problem or task.
• Use any variables, dimensions, or time ranges from the datasets. Participants can focus on a single dataset or combine multiple datasets for a multi-disciplinary visualization.
• Deliverable: A unique and innovative visualization that provides fresh insights into the datasets, presented with a brief explanation of the scientific or creative approach behind the visualization.

Task 4: Explore and Innovate (Open-Ended Task)

• Submissions will be judged on:
• Usability: Can the visualization be easily understood and manipulated by users?
• Creativity and Innovation: Are there novel visual techniques that provide new insights into climate data?
• Scientific Accuracy: Does the visualization accurately represent the underlying data and phenomena?

​

​

​

We encourage submissions with any number of tasks completed!

​

Judges TBD…

​

Evaluation Criteria

Acknowledgements

Scan this Code For More Information:

​

​

Or visit this link: https://sciviscontest2026.github.io/

​