S3U

Science / Astronomy / Graduate / sae01

Propagate Correlated Astronomical Measurement Errors

Derive uncertainty in luminosity and distinguish it from model error.

All worksheets
5 questions0m 0s

Prerequisites

Differentials, variances, covariance, logarithms, and inverse-square flux.

Learn the skill

For isotropic emission without attenuation, L = 4 pi d^2 F. Linearizing gives delta L/L approximately 2 delta d/d + delta F/F. Squaring and averaging introduces a covariance term; independence is an assumption, not a default fact.

Worked example

Independent small fractional errors of 3% in distance and 8% in flux give approximately sqrt((2*0.03)^2 + 0.08^2) = 10% fractional standard deviation in luminosity.

Model and Assumptions

An invented nearby-source model uses isotropic luminosity L = 4 pi d^2 F. Fractional standard deviations are 0.05 for d and 0.10 for F. Their correlation is rho. First-order relative variance is 4(0.05)^2 + (0.10)^2 + 4 rho(0.05)(0.10).

1. What is the relative variance when rho = 0?
2. What is the approximate relative standard deviation when rho = 0?
3. What is the relative variance when rho = 0.5?
4. What happens when errors are large or highly non-Gaussian?
5. Does this formula cover unmodeled attenuation?

Further inquiry

Derive the variance formula from the differential, including the covariance factor. Explore rho from -1 to 1. Explain why a zero first-order variance under perfect anticorrelation would not guarantee an exact nonlinear luminosity.

Review criteria

  • Carry the cross term with its correct factor.
  • Distinguish first-order cancellation from exact cancellation.
  • State the physical assumptions required by the flux relation.