Data Processing & Analysis: From Collection to Discovery
The Eclipse Soundscapes (ES) team is committed to openly sharing the process of analyzing the sound data collected during solar eclipses. This page provides a transparent look at the multi-stage data processing workflow, from initial uploads to expert analysis, ensuring participants stay informed about their contributions. Explore the latest updates on site-specific data progress and learn how this research helps uncover the impact of eclipses on natural soundscapes.
The flowchart image and its text description below the image explain how the Eclipse Soundscapes team handles and analyzes eclipse audio data from ES Data Collectors. This page outlines the steps from receiving and sorting data to uploading, sharing, and analyzing. This process will ensure that data is well-organized, analyzed for significant changes, and shared properly. We are currently in stage 3.
Flowchart Image
Data Flow Chart Explained
Text Description
Stage 1
Receipt, Sorting, & Metadata Organizing:
Input: Data from 1,310 volunteers submitting online registrations with site location, calculated eclipse time, percentage of totality, and 494 microSD cards submitted by volunteers, some with written time and location notes.
Process:
- Match each microSD card to the corresponding online location data and Eclipse Soundscapes ID (ES ID#) to create a metadata spreadsheet
- If no microSD card or online location data is available, attempt to review volunteer-submitted offline location notes for manual entry. If written location notes are unavailable, the process halts for that dataset.
- Merge metadata with NASA’s Goddard Space Flight Center eclipse data to determine the eclipse percentage and maximum time for each dataset.
- Organize each dataset into hardcopy backups by site name and ES ID#.
Delays: Manual entry is required when online data is incomplete or missing.
Stage 2
Upload & Backup Raw Data:
Input: Metadata spreadsheet and raw audio data from microSD cards.
Process:
- Ensure the microSD card’s audio data matches its ES ID# in the metadata spreadsheet.
- Verify if the audio data has a timestamp. If no timestamp exists, review volunteer notes to determine date and time for manual tagging, adding this information to the metadata spreadsheet.
- Upload raw audio data and create a backup.
- Zip audio files by site for easier handling.
Delays: Manual entry occurs if timestamps or metadata are incomplete.
Stage 3
Share Audio Data
Input: Processed metadata and zipped audio files.
Process:
- Develop Auidio and metadata sharing API to work with Zenodo data repository
- Prepare metadata and audio files for sharing in Zenodo, using Eclipse Soundscapes (ES)-developed APIs.
- Create a Zenodo template and merge data for streamlined uploads.
- Share rawl data publicly on Zenodo. Click to access the public and free audio data from Eclipse Soundscapes on Zenodo.
Open Source: All ES programs and APIs developed for the ES process are shared openly on the ARISA-Lab-LLC/ESCSP GitHub Repository.
Stage 4
Soundscape Data Analysis
Input: Audio files split into one-minute chunks.
Process:
- Analyze data for “Vocal Richness” and species identification using machine learning tools such as CityNet CNN and BirdNet.
- Publish processed results and findings.
Open Source: All ES programs and APIs developed for the ES process are shared openly on the ARISA-Lab-LLC/ESCSP GitHub Repository.
2024 ES Audio Recording Sites Map
The map below shows the locations where audio data for the 2024 Eclipse Soundscapes project was recorded and shared with the ES team. The recordings are being processed in order to determine which of these can be analyzed for any changes that occurred due to the total solar eclipse on April 8, 2024. The map also illustrates the eclipse path. The ES team is still receiving data, and this map was updated on June 5, 2024. If the map does not load, or if you want to view it in a separate tab, click here.
Not all sites on this map will be analyzed—each recording undergoes processing to determine whether the audio is viable for analysis. It is normal for some sites to be excluded from analysis due to technical issues such as device malfunctions, microSD card damage, or corrupted audio files.
Having a large number of collection sites is crucial because data loss is an expected part of any scientific field study. Even if a site’s audio does not make it to the analysis stage, each contribution remains valuable by helping ensure we have enough high-quality data to study how soundscapes changed during the total solar eclipse on April 8, 2024.
491 MicroSD Cards Being Processed & Prepped for Analysis
- 340 Sites are on the total solar eclipse path
- 151 Sites experienced a partial solar eclipse
To view the ES ID #, latitude and longitude, and eclipse coverage %, click on the pin in the map.
To view the total eclipse sites or the partial eclipse sites in isolation, click on the square in the top left corner of the map to change the map layers.
2024 Data Analysis Sites Map & Info
273 Sites
The Eclipse Soundscapes (ES) team has been hard at work processing audio data collected during the 2023 Annular and 2024 Total Solar Eclipses. After carefully reviewing hundreds of 2024 Total Solar Eclipse recordings, 273 sites met all the requirements to be included in the initial analysis.
This process involved:
Verifying time accuracy – ensuring timestamps were correct and aligned with eclipse events. Checking data integrity – filtering out corrupted or incomplete files. Assessing recording quality – confirming that usable sound data was captured.
Why does this matter? Some recordings were inevitably lost due to device malfunctions or storage failures, but having data from a large number of sites ensures a strong, reliable dataset for studying how soundscapes changed during the eclipse.
Now that processing is nearly complete, we’re moving into the early stages of analyzing these sites with BirdNET, a powerful tool that detects and identifies vocalizing species. Our research focuses on how wildlife responded to the sudden darkness of the eclipse, including changes in vocalization patterns and the timing of behavioral shifts.
Data Analysis Sites Map
Explore the map of the 273 ES sites that are part of the 2024 data analysis. Click on a site to get more information. Sort the sites by clicking on the rectangle to the left of the orange A in the map header.
Data Analysis Sites Info by ES Data Collector Groups
The sites below were included in the 2024 data analysis. Many more people participated as ES Data Collectors than is represented below.
Individuals
Click on an ES ID Number below to view more details:
Museums, Schools, & Libraries
Click on an ES ID Number below to view more details:
Parks
National Parks, State Parks, and several other natural area collaborated with ES. Click on an ES ID Number below to view more details:
Sounds of Nature
Members of the Sounds of Nature community research project collaborated with ES and collected audio data at the 66 sites below. Click on an ES ID Number below to view more details:
SciAct Partners
NASA Science Activation Partners, including Nationwide Eclipse Ballooning Project (NEBP), Gulf of Maine Research Institute (GMRI), NASA at My Library, NESEC Globe Eclipse, collaborated with ES to collect audio data. The NEBP and GMRI sites that were included in analysis are below. The included NASA at My Library sites are in the library tab. NESEC Globe Eclipse community members participated on their own as individuals. Click on an ES ID Number below to view more details:
Data Analysis: Uncovering the Sounds of the Eclipse
The Eclipse Soundscapes (ES) team is entering the next phase of our research—analyzing audio data collected from 273 verified sites during the 2024 Total Solar Eclipse. This analysis aims to reveal how wildlife, particularly birds, responded to the sudden darkness of an eclipse.
Why BirdNET?
After evaluating multiple analysis tools, we have prioritized BirdNET, a powerful machine learning tool designed to detect and identify bird species through their vocalizations. BirdNET provides clear, tangible insights into how different species behave during the eclipse.
While we may explore CityNET and other tools in the future—especially for analyzing insect sounds—our current focus is on bird vocalizations, as they provide an accessible and well-documented way to study changes in the soundscape.
Key Research Questions
Using BirdNET, we are investigating:
How does vocalization activity change during an eclipse? Which species are vocalizing—or going silent—during totality? When do these behavioral changes begin before totality, and how long do they last?
Thanks to the precise timestamps in our dataset, we have a unique opportunity to examine the exact timing of behavioral responses. This level of detail will allow us to explore minute-by-minute changes, offering new insights into how birds react to rapid shifts in daylight.
Early Findings: Watch the Video!
While our analysis is still in its early stages, we are excited to share some preliminary findings in the video below. This sneak peek highlights some of the first patterns emerging from the data and sets the stage for deeper exploration in the months ahead.
Stay tuned as we continue this exciting research, refining our findings and expanding our analysis. Your contributions to Eclipse Soundscapes are helping to build a groundbreaking understanding of how eclipses affect the natural world!
Data Sharing
The Eclipse Soundscapes (ES) audio data consists of recordings collected by ES volunteers participating as Data Collectors during solar eclipses. These recordings are publicly available and can be used for research, educational, artistic, or other purposes. Uploading the data is a time-intensive process and will continue throughout 2025 with more and more sites being added throughout the year.
Where to Find the Data
The audio recordings are available for public access on the Eclipse Soundscapes Community page on Zenodo. Users can explore the data in two ways:
- By browsing the Zenodo ES Community page, where datasets are categorized for easy access.
- By referencing the list below, which is pulled from the Zenodo website.
Explore the uploaded ES Audio data on Zenodo:
Zenodo – Eclipse Soundscapes Community
Criteria for Inclusion in the Public Dataset
To ensure the reliability and accuracy of the recordings, only data that meets the following criteria will be included:
-
Accurate Time and Date- The AudioMoth device must have been properly set with an accurate time and date.
- If the time and date were recorded manually in the ES Data Collector notes.
-
Device and Data Integrity
- The audio data must be free of issues caused by device malfunction, microSD damage, or other technical problems that could compromise the quality or completeness of the recordings.
-
Verified Location
- A valid latitude and longitude must be available for the recording site.
Accurate Time and Date
Device and Data Integrity
Verified LocationNote on Research vs. Public Data Sharing
For ES-led research, researchers required recordings from at least the two days prior to the eclipse and the eclipse day for inclusion in their studies. However, this was not a requirement for public sharing on Zenodo, as others who utilize this data may have different research or project needs. All available high-quality recordings, regardless of duration, are shared to allow for broader use and exploration.
By following these standards, the Eclipse Soundscapes project ensures that all publicly available data is well-documented and accessible for a variety of uses.
List of All ES Sites Currently Available on Zenodo
Failed to load Zenodo records.