O3Scope (Ozone Analysis and Prediction Using Nasa DataSet)

Tracking the Ozone Layer: Interactive Maps, City Insights, and Future Predictions.

Comment

Inspiration

The depletion of the ozone layer is one of the most pressing environmental challenges, affecting UV radiation exposure, human health, and ecosystems.
I was inspired by the idea of making satellite data tangible, turning complex NASA ozone measurements into interactive insights that anyone can explore, from global trends to city-level risks.

What I Built

I created a Hex-powered interactive app that allows users to:

  • Explore global ozone distribution on a world map.
  • Select specific years (2012–2026) and cities to see ozone levels in Dobson Units (DU).
  • Classify cities into Normal, Warning, and Severe Depletion zones, providing a human-centric perspective on ozone risk.
  • Analyze historical trends and predict 2027 ozone levels using a simple linear regression model powered by Hex AI.

How I Built It

  1. Data Collection
    I used NASA OMPS datasets. Each HDF5 file contains latitude, longitude, and ozone column amounts.

  2. Data Processing & EDA

    • Extracted 2D ozone arrays for each year.
    • Calculated global and city-level averages.
    • Identified patterns and anomalies over time.

  3. Visualization

    • Used matplotlib and cartopy to create interactive maps.
    • Added city markers and color-coded ozone levels.

  4. Prediction

    • Trained a Linear Regression model on historical yearly averages to predict 2027 ozone:

$$ \hat{Ozone}_{2027} = \beta_0 + \beta_1 \times Year $$

  • Displayed trend lines and predicted values directly in the app.
  1. App Interactivity
    • Created Hex dropdowns and checkboxes for selecting cities and years.
    • Made outputs dynamic, showing maps, tables, and textual insights.

Challenges Faced

  • Data size and format: HDF5 files required careful handling to extract meaningful 2D arrays and align them with latitude-longitude grids.
  • Visual clarity: Displaying global ozone data in a readable way while highlighting city-level information took several iterations.
  • Prediction limitations: With limited data points, predictions needed to be accompanied by clear explanations of uncertainty.

What I Learned

  • How to handle, clean and analyze complex datasets like NASA OMPS.
  • How to work with HDF5 satellite datasets and extract meaningful information.
  • How to map gridded geospatial data to cities using latitude-longitude coordinates.
  • How to build interactive Hex apps for real-world scientific storytelling.
  • The importance of communicating uncertainty and real-world implications in predictive modeling.

Reflection

This project reinforced that by combining visualization, interactivity, and prediction, users can explore ozone trends in a way that is both educational and actionable.

Built With

Share this project:

Updates