Context and Why This Matters
Modern astronomy runs on public catalogs (Gaia, SDSS, and similar surveys), where the scientifically interesting sample is defined by the cuts applied to it, not simply by what the archive returns. Two astronomers using different, undocumented cuts on the same catalog can reach different conclusions from identical raw data.
Learning Goals
- Explain why a catalog query is a scientific decision, not a neutral data pull.
- Apply a parallax signal-to-noise cut and justify the threshold.
- Compute distances from parallax for a small archive sample.
Opening Phenomenon
A three-row archive excerpt reports parallaxes with very different fractional uncertainties. Naively inverting every parallax to get a distance produces one wildly unreliable value hiding among two good ones.
Terms Developed in Context
These terms form the measurement chain for Survey Archives and Reproducible Queries: each connects an observable to a model choice, and each has a boundary that must be respected when making a scientific claim.
- parallax
The apparent annual angular displacement of a nearby source caused by observing from different positions in Earth’s orbit. It supplies a geometric distance indicator when the angular signal is measured with adequate precision.
For parallax \(\pi\) in arcseconds, \(d(pc)=1/\pi\); the archive example uses milliarcseconds, so \(d=1000/\pi_{mas}\). Naively inverting a low-SNR or negative measured parallax produces a biased or meaningless distance.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- fractional parallax error
The ratio of parallax uncertainty to measured parallax, \(\sigma_\pi/\pi\). It provides a quick quality measure for deciding whether reciprocal distance is trustworthy.
A parallax-to-error cut greater than five is equivalent to fractional error below 20 percent. A threshold is a selection choice, not a correction for all distance biases.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- selection function
The probability that an object with given properties enters the observed or queried sample. It explains how magnitude limits, quality cuts, sky coverage, and detection efficiency shape archive conclusions.
Completeness and contamination are evaluated relative to the defined selection function. A catalog subset is not automatically representative of the underlying population.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- query reproducibility
The ability to rerun an archive request and recover the same sample from documented service, table, fields, filters, and query text. It makes the data-selection step inspectable in the same way as a reduction script.
The query record is part of provenance and defines the sample before any physical model is fitted. A screenshot or final row count is insufficient if the exact cuts and data release are absent.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- catalog flag
A coded field describing data quality, processing status, duplication, saturation, or another condition attached to a catalog row. Flags let the analyst exclude or separately treat measurements known to violate assumptions.
Flag filters contribute directly to the selection function and should be preserved in the query record. A zero or “good” flag does not guarantee suitability for every scientific question.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- completeness
The fraction of the target population satisfying the scientific definition that is actually included in the sample. It determines whether missing faint, crowded, or low-quality sources bias population inferences.
Completeness is measured relative to source properties and the selection function, often using injection tests or deeper comparison data. A large catalog can still be incomplete in precisely the regime most important to the analysis.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- contamination
The fraction of selected objects that do not belong to the intended population or fail the intended quality standard. It measures the cost of permissive cuts and affects the credibility of trends inferred from the sample.
Tightening cuts often lowers contamination while also reducing completeness, creating a documented tradeoff. Contamination is not identical to random measurement error; it concerns unwanted sample membership.
Developed in this lecture: The term helps interpret this evidence: Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%). It also constrains the working model: A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
Evidence
- Source 1: parallax = 5.1 mas, uncertainty = 0.2 mas (fractional error = 4%).
- Source 2: parallax = 1.2 mas, uncertainty = 0.5 mas (fractional error = 42%).
- Source 3: parallax = 8.8 mas, uncertainty = 0.1 mas (fractional error = 1%).
Model
- A parallax measurement is only useful for a distance estimate when its fractional uncertainty is small (a common cut is requiring parallax/error > 5).
- Distance in parsecs is the reciprocal of parallax in arcseconds; a poorly measured parallax produces a wildly uncertain, potentially unphysical, distance.
- A reproducible query documents every cut applied (magnitude limits, parallax quality, flags) so another astronomer could recover the same sample.
Working Equation
\[ d\,({\rm pc}) = \dfrac{1}{\pi\,(\text{arcsec})} = \dfrac{1000}{\pi\,({\rm mas})} \]
Worked Example
- Source 1: parallax/error = 25.5 (passes a >5 cut); distance = 1000/5.1 = 196.1 pc.
- Source 2: parallax/error = 2.4 (fails a >5 cut); a naive distance of 833.3 pc should not be trusted and this source should be flagged or excluded.
- Source 3: parallax/error = 88.0 (passes easily); distance = 1000/8.8 = 113.6 pc.
- Applying a uniform parallax/error > 5 cut removes Source 2 from any distance-based analysis while keeping Sources 1 and 3 — exactly the kind of documented decision a reproducible query record must state explicitly.
Common Misconception
Reporting "the archive returned N objects" without stating the magnitude limits, quality flags, and parallax-quality cuts that produced that number — the sample cannot be reproduced or critiqued without them.
Active-Learning Guidance
Given two additional archive rows, decide which pass a parallax/error > 5 cut and compute distances only for those that do.
Lab / Observing Connection
Lab 07 (Archive Query Reproducibility) requires writing out the exact cuts applied to a larger sample and reporting how many sources each cut removes.
Synthesis Questions
- What was measured directly in this lecture’s worked example, and what was inferred from the model?
- Which uncertainty or systematic would most change the interpretation?
- How does this technique connect to the technical observing report due at the end of the term?
References
- OpenStax Astronomy 2e, Chapter 19.2 (surveying the stars, parallax) and general survey-data practice.
- Local reference copy:
references/openstax-astronomy-2e.pdf.