Neighborhood Smoke Source Mapping With Cheap Sensors

By /

Neighborhood Smoke Source Mapping With Cheap Sensors

ISEF Category: Earth and amp; Environmental Sciences

Ready to Turn This Idea Into a Real Project?

This guide was put together with the help of AI research tools to give you a solid starting point.But a competitive science fair project lives in the details: refining your research question, fine-tuning your variables, analyzing your data, and presenting your findings like a seasoned scientist.

For next steps tailored to your interests, skill level, and timeline, work one-on-one with a MehtA+ mentor. Learn more about MehtA+ Science & Engineering Research Mentorship →

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

The Hook

A tiny sensor can tell a bigger story than a single smoke reading. When PM2.5, CO, and NO₂ rise together, they can point to what people are burning nearby. That means you can study local heating patterns without ever opening a chimney. You are basically turning air data into a neighborhood fingerprint.

What Is It?

This project asks you to infer where pollution comes from by comparing how several pollutants move together. PM2.5 means tiny particles in the air. CO is carbon monoxide, a gas made by incomplete combustion. NO₂ is nitrogen dioxide, a gas that often rises with fuel burning and traffic. When these signals co-vary, they can act like clues from the same crime scene.

Non-negative matrix factorization, or NMF, is a math method that looks for hidden patterns in data that cannot go below zero. Think of it like sorting a mixed box of colored beads into piles by shared color patterns. In your case, the “piles” may correspond to source types such as wood smoke, gas heating, or traffic influence. You are not proving the exact source with perfect certainty. You are building the best explanation from the pattern in the data.

Why This Is a Good Topic

This makes a strong science fair topic because you can collect real environmental data, compare multiple sensors, and use a serious analysis method without needing a university lab. It connects to air quality, home heating, and exposure in neighborhoods, so the question has clear real-world value. You can learn sensor calibration, time series analysis, and basic source apportionment, which are all skills that matter in environmental research. A careful student can make a project that feels original, local, and data-rich.

Research Questions

  • How does nighttime co-variation of PM2.5, CO, and NO₂ differ between blocks with different heating density?
  • What is the effect of weather conditions on the sensor patterns linked to local fuel burning?
  • Does adding NO₂ improve the separation of source patterns compared with using only PM2.5 and CO?
  • To what extent do indoor and outdoor sensor trends match during likely heating periods?
  • Which neighborhood features best predict the strongest mixed-fuel signature in sensor data?
  • How does the number of hidden factors chosen in non-negative matrix factorization change the source interpretation?

Basic Materials

  • Low-cost PM2.5 sensor with data export, such as Plantower-based monitors or a PurpleAir unit if available.
  • Low-cost CO sensor with logged readings.
  • Low-cost NO₂ sensor with logged readings.
  • Laptop with spreadsheet software.
  • Weather data source from NOAA or a local weather station.
  • Notebook for site notes and timestamps.
  • Outdoor-rated power bank or battery setup for portable logging.
  • Fixed mounting hardware or a safe bracket for consistent placement.

Advanced Materials

  • Reference-grade PM2.5 monitor for cross-checking sensor bias.
  • Electrochemical CO and NO₂ sensors with calibration interface.
  • Multiparameter meteorological station for wind, humidity, and temperature.
  • Data logger or microcontroller with synchronized timestamps.
  • Field calibration gases or a supervised calibration setup through a school or university lab.
  • GIS software or mapping tool for spatial analysis.
  • Air inlet tubing and identical housings for side-by-side comparisons.
  • Particle count or black carbon instrument if a partner lab can provide access.

Software & Tools

  • Python: Processes time series, cleans data, and runs non-negative matrix factorization.
  • pandas: Organizes sensor logs and weather data into usable tables.
  • scikit-learn: Runs non-negative matrix factorization and related clustering tests.
  • QGIS: Maps sensor locations and compares neighborhood patterns.
  • ImageJ: Helps inspect sensor filter images or calibration plots if you collect them.

Experiment Steps

  1. Define the neighborhood question and decide what count as likely source zones, such as areas with dense home heating or heavy traffic.
  2. Choose a sensor placement plan that keeps location, height, and shielding as consistent as possible.
  3. Build a data table that aligns PM2.5, CO, NO₂, and weather variables on the same time axis.
  4. Decide how you will screen out periods with rain, strong wind shifts, or bad sensor behavior.
  5. Test a small set of factor numbers in non-negative matrix factorization and compare which solution makes physical sense.
  6. Link the math output back to site notes, weather, and known local features so your source story is not just a chart.

Common Pitfalls

  • Moving the sensor between sites without a fixed height or housing, which adds fake differences between locations.
  • Treating traffic spikes and heating spikes as the same source, which can blur the source mix.
  • Ignoring humidity and temperature effects, which can distort low-cost PM2.5 readings.
  • Choosing too many NMF factors, which can split one real source into several meaningless parts.
  • Using unsynchronized timestamps, which makes the pollutant co-variation look weaker or stronger than it really is.

What Makes This Competitive

A stronger project goes beyond a simple before-and-after comparison. You can test whether one sensor set can separate source types better than another, or whether the factor patterns match known neighborhood features with spatial evidence. You can also compare several NMF models and defend why one fits best using error, stability, and interpretability. That kind of careful model choice, plus good field design, can move the work from a basic monitoring project to a research-style analysis.

Project Variations

  • Compare a residential neighborhood with a nearby commercial corridor to see how the source mix changes by land use.
  • Focus on winter evenings versus daytime hours to test how heating patterns change the factor structure.
  • Add black carbon or wind direction data to see whether the source apportionment becomes clearer with an extra air-quality clue.

Learn More

  • NOAA Air Resources Laboratory: Search for guidance on air quality data, atmospheric measurement, and local weather context.
  • US EPA Air Sensor Toolbox: Find free background on low-cost sensor behavior, calibration, and field use.
  • NASA Earthdata: Look for articles and datasets on aerosols, smoke, and air quality patterns.
  • PubMed: Search review articles on low-cost air sensors, source apportionment, and residential combustion exposure.
  • MIT OpenCourseWare Environmental Engineering courses: Use free course notes to learn environmental data analysis and atmospheric basics.
  • USGS Water and Environmental Science resources: Read about field sampling, data quality, and how to think like an environmental scientist.
Shopping Cart