Spaxel
September 2026
Context
Most space photos show the sky in one set of colors. SPHEREx, a NASA infrared telescope launched in 2025, instead records the entire sky in 102 colors. The pattern of light across those colors reveals what an object is made of and how hot it is, whether it’s a star, a glowing cloud of gas, or a brown dwarf [1]. That makes the dataset a map of the sky that researchers can explore for new discoveries, but only if they can easily pull up the spectrum of any spot they’re curious about.
Spaxel extends upon prior visualization work under the Findercharts program. This tool, built during my time back at Caltech, gives researchers a more accessible and feature-rich way to analyze a far more comprehensive dataset.
Overview
Given an RA/Dec and a field of view, the pipeline queries IRSA (or a SPHEREx mirror on S3), aligns every cutout north-up, and builds a spectral cube. Results come back as FITS products, CSVs, and an HTML viewer where clicking a pixel plots its spectrum and cross-matches nearby stars against Gaia.
The Pipeline
- Acquire. Cutouts covering the target are pulled from IRSA or the S3 mirror and combined into one multi-extension FITS file.
- Build the cube. Each exposure is reprojected onto a common north-up grid using its own world coordinate system, with a wavelength-dependent background estimated and removed. Quality flags such as overflow, non-functional pixels, and missing data are masked, and per-pixel variance is carried through.
- Match wavelengths. Every pixel in the cube also gets a wavelength and bandwidth, since each exposure samples a slightly different wavelength at each position.
- Export. The cube is saved as FITS and CSV products, plus RGB composites with D1/D2 mapped to blue, D3/D4 to green, and D5/D6 to red.
Large fields are wavelength-compressed for the browser, but spectra are always measured against the full-resolution raw layers. Photometry samples each exposure in its own native pixel grid instead of the reprojected one, which avoids the interpolation scatter that reprojection introduces for undersampled point spread functions.
What It Lets Researchers Do
- Click any pixel for its spectrum. An aperture with an optional local annulus background gives a cleaned 0.7 - 5 µm spectrum with uncertainties, exportable as CSV.
- Identify sources. Gaia Data Release 3 stars in the field are cross-matched, with positions propagated for proper motion to the current epoch, and can be filtered by magnitude and parallax. Spectra can be extracted for many Gaia sources in one batch.
- Characterize them. For a selected source the viewer reports brightness magnitudes per band, bolometric flux, and luminosity when a parallax is available. A color-luminosity diagram places every source in the field on one plot.
- Search by spectral feature. A wavelength overlay highlights where a feature is strongest, with presets for polycyclic aromatic hydrocarbon (PAH) emission (3.15 - 3.60 µm), M dwarfs, planetary nebulae, and brown dwarfs, or a custom range. This makes new candidates stand out.
- Explore the cube. Switch between RGB, per-bandpass stacks, a single wavelength layer, and a wavelength-contrast view, with an Aladin sky overlay for context.
- NASA. NASA’s SPHEREx Telescope Sees Menagerie of Brown Dwarfs. Article Link