Satellite Methane Plume Ranking Project

ISEF Category: Earth and Environmental Sciences

Subcategory: Atmospheric Science  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

Methane traps far more heat than carbon dioxide over short time spans. That means one leaking landfill or dairy can matter a lot more than its size suggests. Satellites now let you look for those leaks from space. Your project can turn raw orbital data into a ranked list of likely emitters.

What Is It?

This project asks a simple question with a big atmosphere behind it, where does methane seem to be escaping, and how strong does that signal look? Sentinel-5P is a satellite that measures gases in the air. You can think of it like a camera for chemistry. Instead of taking a photo of a field, it takes a measurement of what gases sit above that field.

A matched-filter retrieval is a way to pull a weak signal out of noisy data. Imagine trying to hear one flute in a crowded gym. The filter looks for the pattern methane should make in the satellite data and boosts that pattern while reducing the rest. After that, a Bayesian hierarchical model helps you compare many facilities at once while handling uncertainty. That model does not just say, “this site looks strong.” It helps you estimate which sites are likely stronger than others, even when the measurements are messy.

Why This Is a Good Topic

This is a strong science fair topic because it has clear data, clear math, and a real-world purpose. You can test whether satellite data can separate large methane sources from weaker ones, and you can study how stable those rankings are across time or location. The topic also connects to climate, waste management, and dairy emissions, so your results matter outside the classroom. You can learn remote sensing, signal extraction, uncertainty, and statistical modeling in one project.

Research Questions

• How does the choice of matched-filter baseline affect methane plume detection near landfills and dairies?
• What is the effect of pixel selection radius on the detected methane enhancement above each facility?
• Does a Bayesian hierarchical ranking change when you group facilities by land use type?
• To what extent do wind speed and wind direction improve methane plume detectability in Sentinel-5P scenes?
• Which facility attributes best explain the strongest methane signals, landfill size, herd size, or nearby waste handling activity?
• How does the number of satellite passes used in the model affect the stability of facility rankings?

Basic Materials

• A computer with internet access.
• Sentinel-5P methane data from the Copernicus or NASA Earthdata portals.
• Google Earth or QGIS for viewing facility locations and satellite footprints.
• A spreadsheet program for logging facility metadata and outputs.
• A list of candidate landfills and dairies from public maps or state databases.
• Weather data from NOAA for wind screening and scene selection.
• A notebook for tracking assumptions, exclusions, and uncertainty decisions.

Advanced Materials

• A university workstation or cloud compute account.
• Python with geospatial and Bayesian libraries.
• Sentinel-5P Level 2 methane products.
• Facility shapefiles or parcel layers for landfills, dairies, and surrounding land use.
• NOAA or reanalysis wind fields for plume screening.
• A calibration or validation dataset from published methane emission studies.
• A version-controlled project folder for code, figures, and model runs.

Software & Tools

• Python: Handles data cleaning, plume screening, map making, and Bayesian analysis.
• QGIS: Lets you inspect facility footprints, satellite scenes, and plume locations on a map.
• Google Earth Engine: Helps organize large geospatial datasets and compare scenes over time.
• ImageJ: Can help inspect visual patterns in exported plume images and masks.
• R: Supports statistical summaries, uncertainty checks, and ranking comparisons.

Experiment Steps

1. Define a facility list and decide which public data fields you will use to compare sites.
2. Select the satellite scenes that best match your question and set your exclusion rules for clouds, winds, and bad geometry.
3. Design the matched-filter workflow so you can turn each scene into a methane enhancement map.
4. Build a standard comparison method for each facility so rankings are based on the same spatial window and time window.
5. Plan the Bayesian model structure that separates site-to-site differences from scene-to-scene noise.
6. Decide how you will test uncertainty, sensitivity, and ranking stability before you trust the final results.

Common Pitfalls

• Mixing scenes with very different wind conditions, which can make one facility look stronger for reasons that have nothing to do with emissions.
• Using inconsistent search radii around facilities, which changes the amount of background methane included in each comparison.
• Treating cloudy or low-quality satellite passes like clean observations, which adds noise and weakens the plume signal.
• Comparing raw methane columns without a baseline correction, which hides small but real enhancements near the source.
• Ranking facilities from too few observations, which makes the Bayesian model look confident when the data are still thin.

What Makes This Competitive

A strong version of this project goes beyond simple plume spotting. You would test how sensitive your results are to scene selection, background choice, and model assumptions. You would also compare multiple facility groups, not just one site, and report uncertainty on every ranking. That kind of careful analysis turns a map exercise into a real atmospheric science study.

Project Variations

• Compare methane signals from landfills, dairies, and oil and gas sites using the same retrieval pipeline.
• Test whether rankings change when you switch from Sentinel-5P scenes to another public methane product.
• Analyze seasonal differences in methane detectability by comparing winter, spring, summer, and fall observations.

Learn More

• NASA Earthdata: Search for Sentinel-5P methane products, user guides, and download tools through the Earthdata portal.
• Copernicus Sentinel-5P documentation: Read the mission overview and product descriptions on the Copernicus data space and Sentinel-5P pages.
• NOAA Global Monitoring Laboratory: Find background methane information, atmospheric context, and related greenhouse gas resources.
• PubMed: Search for review articles on methane detection, remote sensing, and emission quantification from landfills and dairies.
• MIT OpenCourseWare: Look for free materials in atmospheric chemistry, remote sensing, and Bayesian statistics courses.