ThisCalculator

Everyday

Verify an AI Exoplanet Candidate with NASA Data

Updated · Standard formula, shown below

This guide shows you how to take an exoplanet candidate flagged by an AI model and confirm it against real NASA mission data. You’ll learn to cross-check the candidate’s transit signal, validate its host star, and assess whether the signal is a genuine planet or a false positive. No prior astronomy experience is needed, but you should be comfortable with basic web navigation and have a candidate’s coordinates or identifier ready.

What You’ll Need

  • The candidate’s sky coordinates (RA and Dec) or its TIC ID (from the TESS Input Catalog).
  • A web browser and internet connection.
  • Access to NASA’s Exoplanet Archive and MAST (Mikulski Archive for Space Telescopes) – both free, no login required.
  • Optional: A basic understanding of what a transit light curve looks like (a periodic dip in brightness).

Step-by-Step

  1. Identify the candidate’s coordinates. If you have a TIC ID, write it down. If you only have coordinates (e.g., RA 18:36:25, Dec +38:47:01), convert them to decimal degrees if needed (e.g., using an online converter).
  2. Open the NASA Exoplanet Archive. Go to exoplanetarchive.ipac.caltech.edu. In the top menu, click “Transit Spectroscopy” or “Periodic Tables” – but for a quick check, use the “Confirmed Planets” table. Search for your candidate’s coordinates or TIC ID using the “Search” box. If the object is already a confirmed planet, it will appear here with a status like “Confirmed.”
  3. Search the TESS Objects of Interest (TOI) catalog. In the same archive, go to “TESS” → “TOI” (or use the “TESS” tab). Enter your TIC ID or coordinates. This catalog lists planetary candidates from the TESS mission. If your AI candidate is a real TESS detection, it will show up with a TOI number and a “Disposition” (e.g., “Candidate” or “Confirmed”). Note the period and transit depth listed.
  4. Check the light curve on MAST. Go to mast.stsci.edu. Use the “TESS” data search. Enter your target coordinates or TIC ID. You’ll see a list of TESS observations (sectors). Click on “Light Curves” for the appropriate sector. Download the FITS file or use the interactive light curve viewer (if available) to see the raw data.
  5. Look for the transit signal. In the light curve, you should see periodic dips at the period listed in the TOI catalog. If the AI candidate has a period of, say, 3.5 days, look for a dip every 3.5 days. The dip depth should match the TOI’s transit depth (e.g., 0.5% brightness drop). Use the phase-folded plot if the viewer offers it – that shows all transits stacked, making the signal clearer.
  6. Check the host star’s properties. In the Exoplanet Archive, look up the host star’s effective temperature, radius, and distance. A real planet host is usually a main-sequence star (like our Sun). If the star is a giant or a white dwarf, the signal might be a false positive (e.g., an eclipsing binary). Also check the star’s Gaia magnitude – if it’s too faint (e.g., >16), TESS data may be noisy.
  7. Look for known false-positive indicators. In the TOI catalog, check the “Disposition” column. If it says “False Positive,” the AI candidate is likely not a planet. Also look for a “Contamination” flag – if a nearby star is bright, the transit might be diluted or caused by that star. Use the “TESS Input Catalog” to see if there are nearby stars within the photometric aperture.
  8. Compare with the AI’s confidence score. If your AI model gave a probability (e.g., 0.95), see if the TOI disposition aligns. A high AI score with a “Candidate” disposition is promising. If the AI says “planet” but TOI says “false positive,” trust the TOI – it’s based on human vetting.
  9. Document your findings. Note the TOI number, the period, the transit depth, and the star’s properties. If the candidate matches a TOI with a “Candidate” disposition, you can say it’s “validated as a TESS candidate.” If it’s not in the TOI catalog, it might be a new candidate – but that requires more detailed analysis (like using the DAVE or EXOFOP tools).

Common Mistakes

  • Using the wrong coordinates: A small error in RA/Dec can point you to a different star. Double-check the format (hours vs degrees).
  • Ignoring the host star’s type: A red giant can produce a transit-like signal due to pulsations. Always check the stellar parameters.
  • Confusing the period: The AI might give a period in days, but TESS data might be in hours. Convert units carefully.
  • Not checking for blending: If a nearby star is in the same pixel, the transit depth can be diluted, making the planet seem smaller than it is – or a false positive if the nearby star is an eclipsing binary.

Tips

  • Use the “Exoplanet Follow-up Observing Program” (ExoFOP) website – it has additional vetting information and sometimes even phase-folded light curves.
  • If the light curve is noisy, look at the “PDCSAP” flux (systematics-corrected) rather than the raw SAP flux.
  • For a quick sanity check, use the “Transit Least Squares” (TLS) periodogram on the light curve – but that requires coding. If you’re not comfortable, just rely on the TOI catalog.
  • If the candidate is not in the TOI catalog, it might be from a different mission (like K2). Search the “K2” catalog in the Exoplanet Archive.

Troubleshooting

  • No light curve available: If MAST shows no TESS data for your coordinates, the star might not have been observed by TESS. Try searching the K2 or Kepler data archives.
  • The transit is not visible: Check if the period is correct. If the AI gave a period that is an alias (e.g., 2x the true period), the folded light curve will show two dips. Try folding at half the period.
  • The TOI catalog shows “False Positive” but AI says planet: Look at the reason for the false positive (e.g., “eclipsing binary”). If it’s a binary, the transit depth will be too deep for a planet (e.g., >2%). If the depth is shallow, it might be a misclassification – but trust the TOI unless you have strong evidence.
  • Star is too faint: If the star has a TESS magnitude >14, the light curve will have high scatter. Try using the “SAP” flux and bin the data to 10-minute intervals to reduce noise.

Frequently Asked Questions

Do I need to know how to code?

No. All the steps above use web interfaces. You can download the light curve and use a simple spreadsheet to plot it if you prefer, but the MAST viewer is enough.

Can I verify a candidate from a different AI model?

Yes, as long as you have the coordinates or TIC ID. The process is the same regardless of which AI flagged it.

What if the candidate is not in the TOI catalog?

It might be a new detection or from a different survey. Check the K2 or Kepler catalogs. If it’s truly new, you would need to perform a full vetting, which is beyond this guide.

How long does the verification take?

For a single candidate, about 30 minutes if you’re familiar with the tools. First time might take an hour.

Can I trust the TOI disposition?

Yes, TOI dispositions are vetted by humans using multiple data products. They are the current gold standard for TESS candidates.

What if the transit depth is very shallow (e.g., 0.01%)?

That might be a real small planet, but it could also be noise. Check the signal-to-noise ratio in the TOI catalog. If it’s below 7, it’s likely a false positive.

Related

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.