Urban Biodiversity Sound Monitoring Science Project

ISEF Category: Environmental Engineering

Subcategory: Other  ·  Difficulty: Intermediate  ·  Setup: School Lab  ·  Time: 1 to 2 Months

The Hook

A city can sound healthy or stressed before you ever see a map. Birds, insects, and frogs leave audio clues all day, even when people miss them. You can turn those clues into data and compare different urban zones. That gives you a real way to study how pollution and habitat stress change biodiversity.

What Is It?

This project uses sound as a biology signal. Instead of counting every bird by hand, you record the soundscape and let a classifier help identify species. A classifier is a program that sorts audio into labels, like a very fast helper that listens for you. Bird richness, which means how many different species you detect, can then become a measure you compare across places.

The main idea is simple. Healthy habitats usually support more kinds of calls, while noisy, paved, or heavily polluted areas often support fewer. Think of each location like a playlist. Some places are full of different tracks, while others sound flattened and repetitive. Your job is to test whether that pattern changes with distance from highways or industrial zones.

The ESP32 and INMP441 mic logger collect the sound data. BirdNET-Lite can help identify species from the recordings. You can then compare species richness, call frequency, or acoustic diversity across sites. That turns a messy city soundscape into a clear environmental dataset.

Why This Is a Good Topic

This is a strong science fair topic because you can measure a real environmental pattern without needing a wet lab. You can change one thing, like distance from traffic, and compare it to biodiversity signals in the recordings. That makes the project testable, local, and connected to air pollution, noise pollution, and habitat quality. You can also learn field sampling, audio classification, data cleaning, and basic statistics.

Research Questions

• How does bird species richness change with distance from highways?
• What is the effect of proximity to industrial zones on acoustic biodiversity?
• Does noise level predict the number of bird calls detected in a recording?
• To what extent does time of day change species richness near busy roads?
• Which urban habitat type, park edge, street tree corridor, or vacant lot, has the highest acoustic diversity?
• How does the classifier's species count compare between manually reviewed recordings and BirdNET-Lite output?

Basic Materials

• ESP32 microcontroller board
• INMP441 microphone module
• microSD card module and microSD cards
• breadboard and jumper wires
• USB cable for programming and power
• laptop or desktop computer
• outdoor-safe enclosure for the logger
• smartphone with GPS or maps app
• spreadsheet software for data tables
• decibel meter app or handheld sound meter.

Advanced Materials

• ESP32 microcontroller board
• INMP441 microphone module
• weatherproof microphone housing
• calibrated sound level meter
• field laptop or tablet for metadata logging
• external battery pack or regulated power supply
• microSD cards with backup copies
• reference audio library for local bird species
• basic GIS software or mapping platform
• access to BirdNET-Lite or a similar bird-call classifier.

Software & Tools

• BirdNET-Lite: Identifies bird species from audio clips and helps you turn recordings into species counts.
• ImageJ: Measures visual plots or spectrogram screenshots when you need a quick way to compare signal patterns.
• Python: Cleans audio metadata, organizes outputs, and runs your summary statistics.
• Audacity: Lets you inspect clips, trim recordings, and check whether background noise swamped the signal.
• QGIS: Maps your sampling sites and helps you compare biodiversity patterns with location data.

Experiment Steps

1. Define the environmental gradient you will test, such as distance from traffic, industrial activity, or park size.
2. Choose a recording plan that keeps site selection, timing, and device setup as consistent as possible.
3. Decide how you will turn each audio file into a biodiversity metric, such as species richness, call count, or acoustic diversity.
4. Build your control strategy so you can separate true habitat effects from noise, weather, and time-of-day changes.
5. Plan how you will clean classifier output, remove false positives, and compare automated identifications with a manual check.
6. Set up your analysis before fieldwork so you know which graphs, maps, and statistical tests will answer your question.

Common Pitfalls

• Placing recorders too close to one another, which makes each site capture the same bird community.
• Recording at inconsistent times, which confounds traffic noise, bird activity, and daily calling patterns.
• Treating every classifier label as true, which inflates richness when BirdNET-Lite mistakes background noise for a species.
• Ignoring windy or rainy sessions, which lowers detection quality and hides real site differences.
• Comparing sites with different habitat structure without measuring vegetation or open space, which makes pollution effects hard to separate from habitat effects.

What Makes This Competitive

A strong version of this project does more than count birds. You can test how well acoustic biodiversity tracks a real pollution gradient, then check whether your metric agrees with traffic, land use, or sound level data. You can also strengthen the project with careful controls, repeated sampling, and a manual validation set for the classifier. If you map the results and compare several urban habitat types, your analysis starts to look like real environmental monitoring.

Project Variations

• Test whether acoustic biodiversity differs between schoolyards, city parks, and roadside green strips.
• Compare BirdNET-Lite results with manual call counts to measure classifier accuracy in noisy urban sites.
• Study whether dawn recordings and evening recordings give different richness patterns near the same highway corridor.

Learn More

• BirdNET: Search the BirdNET project site for documentation on bird-call classification and field use.
• NOAA Climate and Weather Data: Search NOAA for local weather records that can help you filter recording days.
• USGS ScienceBase: Search USGS for urban ecology and land cover datasets.
• Cornell Lab of Ornithology: Search the eBird and bird identification resources for species range and call references.
• PubMed: Search review articles on urban noise pollution, biodiversity, and acoustic ecology.
• MIT OpenCourseWare: Search for introductory environmental data analysis and GIS materials.