Model, assumptions & sources
The ideal sub makes sky-background shot noise dominate read noise (Dr Robin Glover, SharpCap):
t_sub = C · RN² / (sky + dark) C = 1 / ((1+p)² − 1)
p is the accepted rise in total noise over a zero-read-noise camera: 5% → C ≈ 9.76, 10% → 4.76, 2% → 24.75. At that point the stack keeps 1/(1+p) of the maximum SNR (95% at 5%), or 1/(1+p)² of the time efficiency (≈91%).
Sky rate, wavelength by wavelength (330–1100 nm):
sky = A_eff · Ω_px · Σ_λ N(λ) · F(λ) · QE(λ) · W(λ) · τ(λ) · τ_mul · Δλ + Σ_lines φᵢ · F(λᵢ) · QE(λᵢ) · τ(λᵢ)
N = sky photons s⁻¹ m⁻² nm⁻¹ arcsec⁻² on a 5 nm grid. F = filter transmission: measured or maker curves where they exist, 50%-edge top-hats or maker FWHM top-hats where they don't (the filter menu says which). Narrowband filters are integrated on their own fine sampling, and airglow lines between 470–520 nm and 640–690 nm are added one by one at the filter's transmission for each line. QE = the sensor's absolute sensitivity curve (colour sensors: red, green and blue pixels separately). W = camera window. τ = optics throughput by telescope type. A_eff = π/4·(D² − d²) minus spider vanes (count × thickness × (D − d)/2). Ω = (206.265·pixel·bin/FL)².
Sky: the natural sky shape is the ESO Cerro Paranal sky model (zenith, no moon, solar flux 130 sfu, which is mid solar cycle; at solar minimum the natural sky is ~0.3 mag darker). Through the SQM response it reads 21.83. Your SQM sets the total at the zenith. If your reading is brighter than that, the extra light is light pollution with the lamp spectrum you pick under Fine-tune (default 4000 K LED). Between 21.83 and 22.13 the natural sky moves toward the model's solar-minimum spectrum (70 sfu, SQM 22.13), which is redder per SQM photon; darker than 22.13 that spectrum is scaled. At your target's height the airglow (lines and airglow continuum) brightens as a thin layer at 87 km (the OH layer) (van Rhijn: X_vR = [1 − (R/(R+h))²·sin²Z]^−½) and is dimmed by 10^(−0.4·k·(X−1)); zodiacal light and starlight only get the extinction term (Noll et al. 2012). For light pollution there is no single sourced law, because its glow versus altitude depends on where the towns are: the Krisciunas & Schaefer (1991) thin-layer law 10^(−0.4·k·(X−1))·X, X = (1 − 0.96·sin²Z)^−½, is assumed. The target dims by 10^(−0.4·k·X) with the Kasten & Young (1989) airmass. k(λ) is the Kitt Peak extinction curve plus the Paranal molecular bands (O₂ 687/760 nm, H₂O).
Moon: Krisciunas & Schaefer (1991) brightness from illumination, moon height and moon–target distance (defaults 45° and 90°), with the colour of scattered moonlight from the ESO sky model, added to the sky as flux. The moonlight colour uses one scattering geometry (30° from the Moon), and the opposition surge near full moon is not modelled. Accuracy about ±30%.
Time to reach an SNR: the standard CCD equation per pixel at your image scale and binning, for N = T/t frames of length t:
SNR = S·T / √((S + B + D)·T + N·RN²) → T = X² · (S + B + D + RN²/t) / S²
S = target signal through the same spectral chain as the sky, after atmospheric extinction. Galaxies use a real galaxy spectrum normalised to the region's surface brightness in the band it was measured in (no fixed colour is assumed). Its emission lines (Hα, [NII] 654.8/658.4, [SII] 671.6/673.1, [OIII] 495.9/500.7, Hβ) were measured on the full-resolution spectra (Gaussian fits, doublet ratios tied) and are placed at their own wavelengths, redshifted by the galaxy's velocity (HyperLEDA) for named galaxies. These are whole-galaxy average lines: individual HII regions are many times brighter in Hα (use Emission nebula → Custom for them). Galaxy regions are the as-imaged values of Fisher & Drory (2008): bright core 17.6 V (16.6–18.2), main disk 21.3 V (20.5–22.0), faint outer arms 22.4 V (21.5–23.1), catalogue edge 25.0 B (RC3 D25), stellar halo and streams 27.5 r (26–29). Emission nebulae use Hα in rayleigh plus [NII], [OIII], [SII] and Hβ from published line ratios, each at the filter's transmission for that line. Values from the all-sky Hα map (6′ beam) underestimate compact cores. Nebula continuum is not included. Colour cameras: per red, green or blue pixel before debayering; the slowest channel that the filter is meant to feed is reported. SNR is per pixel: 2×2 binning roughly doubles it. Flats, darks, calibration noise, gradients and sky-subtraction errors are not included, so real results take somewhat longer.
What is still uncertain: SQM meters read ±0.1 mag, and wide-field SQMs read 0–0.3 mag brighter than the true zenith. Airglow varies 2× or more from night to night, so narrowband sky is uncertain by about ×2: 40–46% of the sky in a 3 nm Hα filter is airglow lines (40% through the measured Chroma 3 nm curve and 45% through a 3 nm top-hat here; 46% in the ESO model), and the Condor Array's measured narrowband rates are 2–4× this model while its own magnitudes agree within 30% (Lanzetta et al. 2023, Table 5). The sky shape is Paranal's and is assumed for other sites. Lamp spectra are used as the skyglow colour; scattering makes real skyglow bluer, and no sourced correction exists, so none is applied. Lamp spectra stop at 900 nm. Optics are modelled with fresh coatings (real ones are 2–10% lower: use the multiplier). Several QE curves are maker curves or proxies from related sensors (shown in the camera details), and even the lab-measured IMX455 curve differs ±7% between labs. The DSLR curve's absolute level is assumed (±30%). Filters marked spec only or edges only are top-hats. Astrodon I-series R and G use the E-series curve: their widths are unverified. Digitized maker charts below 2% transmission are set to 0. Sky-Watcher Newtonians (PDS and Quattro) use a protected-aluminium mirror model for their RAQ coating (aluminium with a quartz overcoat), about 90% per mirror; Sky-Watcher quotes 94%, which would make these tubes about 8% faster. Telescope obstructions marked approximate are not published by the maker and are estimated. The PlaneWave CDK17 is a corrected Dall-Kirkham, not a Ritchey-Chrétien. The RC14 preset's mirror coating (enhanced aluminium) is assumed. A retailer lists a 94% coating for the GSO 14″ RC (TS-Optics GSO 14″ RC Pro: “Coating 94 %, quartz mirrors”); the coating model gives 95.8% per mirror (400–700 nm) and 94.8% at Hα, while protected aluminium would cut Hα by ~13%. Its secondary holder (≈165 mm) and spider vanes (4 × 2.34 mm) were measured by the owner. Its Antlia 3 nm Hα Pro filter is a top-hat from the maker spec; the OIII/SII rows use the same spec as a stand-in. Sky picture → SQM values are approximate (Bortle 1–2 21.7, 3–4 21.2, 5–6 19.4, 7 18.4, 8–9 17.7). Bortle's 2001 scale has no SQM values and no peer-reviewed Bortle–SQM table was found, so the Bortle labels follow Wikipedia (approximate). The SQM values themselves sit inside the measured ranges of Hänel et al. 2018 (JQSRT 205, 278), Table 2: natural starlit sky 21.4–21.9, rural 20.3–21.6, suburban 17.2–20.4, urban 16.8–19.2. Read noise depends on gain and mode. Star clipping is not modelled. Planner gaps assume ~3 s per frame.
Sources:
· Swamp factor and method: SharpCap Smart Histogram, SharpCap forum
· Night-sky spectrum: ESO SkyCalc (Noll et al. 2012, Jones et al. 2013) · measured comparison: Patat 2008
· SQM response and zero point: Cinzano 2005 · Bessell V: SVO Filter Profile Service (Bessell 1990)
· Lamp spectra: LICA-UCM lamp spectral database (Tapia, Sánchez de Miguel & Zamorano)
· Extinction: IRAF KPNO table, Patat et al. 2011 · Airmass: Kasten & Young 1989, Appl. Opt. 28, 4735 · Moon and sky gradient: Krisciunas & Schaefer 1991, PASP 103, 1033
· QE: IMX455 measured by Ofek et al. 2023 (LAST), Alarcon et al. 2023, Gill et al. 2022; maker curves ZWO, Gpixel; DSLR shapes rawtoaces-data
· Read noise: INDI camera-data (QHY600 modes), QHY, Diffraction Limited, photonstophotos.net (DSLR, approximate)
· Filters: Chroma spectral data, Astrodon E-series lab scans (LCO), Optolong and ZWO charts; other makers' specs (Baader, Antlia, Astronomik, SVBONY, Radian, STC, Askar)
· Optics coating model: Rakić 1995 aluminium (refractiveindex.info), Malitson 1965 silica, Schott/Ohara glass data
· Galaxies: Fisher & Drory 2008, RC3 (de Vaucouleurs et al. 1991), HyperLEDA, spectra Kinney et al. 1996 and Brown et al. 2014, halos Martínez-Delgado et al. 2010
· Nebulae: Finkbeiner 2003 Hα map, Esteban et al. 2004, 2005, Madsen, Reynolds & Haffner 2006, rayleigh conversion Haffner et al. 2003
· Sky brightness ranges: Hänel et al. 2018 · Bortle labels (approximate): Bortle scale (Wikipedia) · Velocities: HyperLEDA · Telescopes: maker and dealer spec pages, each cited in the source code
Built for Stellar Nomads · no data leaves your browser.