The Sunlight Spectrum Catalog: A Citizen Science Atlas of Atmospheric Signatures

The Sunlight Spectrum Catalog is a global project designed to put professional-quality spectroscopy into the hands of anyone with a smartphone and a cardboard box. Participants follow step-by-step video tutorials to assemble a simple diffraction-grating spectrometer using a CD, a cereal box, and black tape. They then point it at the sun (using safe projection methods), capture the rainbow spectrum on their phone camera, and upload the photo to our open-source platform. Our backend automatically extracts the wavelength-intensity profile, corrects for camera response, and flags interesting features like Fraunhofer lines, water vapor bands, or particulate scattering. The collected spectra are compiled into an interactive atlas, searchable by location, time, weather, and air quality index. Over time, the catalog reveals patterns: how dust from the Sahara reaches Brazil, how volcanic ash alters the sky, or how urban smog shifts the color balance. The project is deliberately built for absolute beginners—no physics background, no math beyond basic arithmetic, and no expensive equipment. Every step is explained in plain language with visual guides. Our long-term dream is to have a spectrum from every populated square kilometer on Earth, creating a living portrait of our planet's atmosphere.

Understanding Earth's atmosphere requires expensive, specialized spectrometers operated by trained scientists. This creates a data apartheid: the vast majority of atmospheric observations come from a few hundred well-funded institutions, leaving huge spatial and temporal gaps. Meanwhile, millions of curious people own smartphones with decent cameras and would love to contribute to real science. The itch is that there is no accessible, affordable way for ordinary people to collect scientifically meaningful atmospheric data. The Sunlight Spectrum Catalog scratches that itch by turning a cheap disposable item—a CD—into a precision instrument, and by providing a clear pipeline from snap to science. It democratizes atmospheric research and gives beginners a tangible sense of discovery.

• Design and validate a low-cost spectrometer that any beginner can build in under 30 minutes.
• Create a step-by-step tutorial series (videos, PDFs, infographics) that requires no prior science knowledge.
• Launch a web platform that accepts photo uploads, extracts spectra, and visualizes results in real time.
• Collect at least 10,000 unique spectra from 100+ countries within the first year.
• Publish the first public version of the Sunlight Spectrum Atlas, showing seasonal and geographic trends.
• Write a methods paper with 20+ citizen co-authors (the top contributors) describing the dataset and its potential.

• Not aiming to produce lab-grade spectral resolution (we target <5 nm, enough to resolve major features).
• Not replacing existing scientific instruments—this is complementary crowd-sensing.
• Not monetizing the data; everything is open access under Creative Commons Zero.
• Not requiring contributors to own a computer; the upload process works on smartphones.
• Not excluding people who can't safely observe the sun; we provide alternative methods using reflected light.

Materials for spectrometer: cardboard, CD-R (unused or scratched), black electrical tape, razor blade, ruler, smartphone (any with a camera). Construction method: cut slit in cardboard, tape CD at 60°, project spectrum onto flat surface, photograph. Software: custom Python script using OpenCV to extract pixel values to wavelength (calibrated via known line positions), Flask web app for uploads, PostgreSQL database, interactive frontend with Mapbox for atlas. Data format: raw images (JPEG), extracted spectral curves (CSV), metadata (GPS, time, ISO, exposure). Guide materials: recorded screen-captures, PDF blueprints, narrated slideshows.

The project is structured as a three-tier system: (1) Field tier—the physical spectrometer kit and smartphone app for capture. The app provides a live histogram to help users expose correctly and a built-in solar compass to aim safely. (2) Cloud tier—when a user uploads, the image passes through a pipeline: denoising, distortion correction (from lens vignette), wavelength calibration via known Fraunhofer lines (e.g., Hydrogen-alpha at 656.3 nm), and intensity normalization. Then the spectrum is stored tagged with GPS, timestamp, and weather info pulled from OpenWeatherMap. (3) Atlas tier—an interactive globe where each pin shows a thumbnail spectrum. Users can overlay air quality, humidity, or aerosol index layers to see correlations. The architecture is designed so that a contributor can see their data appear on the map within 5 minutes of uploading.

• Data quality: beginners may produce blurry or poorly exposed images. Mitigation: in-app real-time feedback, and automated rejection of spectra with low signal-to-noise. Human reviewers can also flag bad data.
• Safety: looking at the sun can damage eyes. Mitigation: the tutorial exclusively uses projection (look at the cardboard, not directly at the sun). We add a mandatory safety quiz before first upload.
• Seasonal/geographic bias: most early participants may be from temperate regions. Mitigation: partner with schools in tropical regions, provide simple sun-finding charts.
• Abandonment: without gamification, users may lose interest. Mitigation: badges for number of spectra, leaderboards, and monthly 'discovery highlights' featuring unusual spectra.

• What is the simplest way to automate wavelength calibration without requiring a known lamp? We're exploring using the sun's 589 nm sodium doublet as a built-in marker.
• How should we handle spectra taken indoors through windows? The glass alters the spectrum—do we include them as 'indoor' vs 'outdoor' categories?
• Can we train a simple neural network to classify spectra by weather condition (clear, cloudy, hazy) without needing metadata?
• What is the minimum number of pixels needed across the spectrum to achieve 5 nm resolution? This affects smartphone compatibility.

We built the first prototype and captured about 200 spectra from friends and volunteers, but then the core team got busy with day jobs and the calibration pipeline needed more work than anticipated. The project is deeply researched and ready to launch, but we never found the two consecutive months of uninterrupted time needed to finalize the upload platform and release the tutorials. It remains a beloved 'one day' idea.

The project was originally inspired by the Public Lab spectrometer design, but we wanted to make something even simpler and focused specifically on sunlight. The name 'Sunlight Spectrum Catalog' is deliberately low-tech and inviting. We hope someone else picks it up and runs with it—all plans and code are open-source and waiting on GitHub.