TychoCam™

The Technology

Every observing condition is a direct physical measurement with its own confidence: stars counted against a catalog at night, the colour of each sky pixel by day, and an infrared sky sensor that learns from both. The measurements are combined by one rule, day or night, and the system fails safe when it cannot see the sky. This supports continuous, unattended ASCOM Alpaca observing-conditions reporting.

 Measuring the Alpaca Required Observing Conditions

The TychoCam combines several hardware sensors with image-processing software to determine the 13 ASCOM Observing Conditions and present them on the website and through the APIs. The APIs comply with the ASCOM standards, and every value is reported in its ASCOM units.

Hardware Gauges and Measurements

Sensors attached to the TychoCam measure conditions at the site. The TychoCam integrates the following hardware, with a considerable amount of supporting custom software:

Condition Sensor Units
Ambient Temperature BME280 °C
Relative Humidity BME280 %
Barometric Pressure BME280 (station pressure) hPa
Infrared Sky Temperature IR sky sensor °C
Rain Rate Optical (IR) rain sensor mm/hr
Wind Speed and Gust Anemometer m/s
Wind Direction Wind vane degrees
Sky Brightness AS7343 spectral sensor lux

Calculated Observed Conditions

Condition Derived from Units
Cloud Cover Camera (stars, sky colour) and IR sky sensor %
Image Quality Star profiles in the camera image pixels
Sky Quality Sky luminance, calibrated against the camera mag/arcsec²
Bortle Class Sky Quality 1–9
Dew Point Temperature and humidity °C

Seeing (ASCOM StarFWHM, in arc-seconds) cannot be measured at an all-sky plate scale. It can be supplied by the site from an external instrument such as a DIMM, and is then shown on the website and reported through the APIs, as can other external data such as MASS or CCD measurements.

As a backup to the local sensors, the TychoCam queries the local National Weather Service to fill in missing data.

Image Processing - Night

The TychoCam uses a colour astronomy camera with a fisheye lens covering the whole sky: a ZWO ASI178MC (2080 × 2080 pixel images) or an SVBONY SV705C (2180 × 2180 pixel images). One image is taken every 30 seconds, with exposures from 30 ms in daylight to 60 seconds at night. Whenever the Sun is below −6°, each image is processed as follows (simplified):

  • Convert the Hipparcos star catalog to altitude and azimuth for the site and the time of the image.
  • Map catalog stars into the image with the camera model (fisheye projection, rotation and centring), which is fitted to matched stars and checked against physical limits before use.
  • Detect likely stars in the image and match them to the catalog.
  • Apply the obstruction mask, and allow for moonlight and twilight when deciding which catalog stars should be visible.
  • Measure cloud cover from the stars found versus the stars expected, correcting for coincidental matches.
  • During astronomical night, measure Image Quality: the FWHM of fitted star profiles, as an SNR-weighted median over a sample of stars in clear sky.
  • Plot the image with expected, found, and not-found stars, plus optional constellation lines or range rings.
  • Add optional data to the image legend for convenience: cloud cover, temperatures, dew point, sky temperature, and more.
Cloud Cover — One Method, Day and Night

Cloud cover is the percentage of the sky covered by cloud, weighted by solid angle so that every part of the sky counts by its true area. There are no separate day, twilight, and night modes. Every image is offered to every measurement method; each method reports a value with its own confidence (0 to 1), and declines to vote when its physics does not apply. The methods are then combined, weighted by confidence.

Method Runs when What it measures Confidence falls when
Star count Sun below −6° Catalog stars that should be visible compared with stars actually detected. The number expected allows for sky brightness from twilight and moonlight, and a limiting magnitude fainter than the sky brightness explains is treated as extinction by thin cloud. Few stars are bright enough to see (twilight, bright Moon), or many matches could be coincidence (fog, noisy frames). Coincidence is measured on every frame using decoy positions where no real star exists.
Red/blue ratio Sun above +3° Each sky pixel is classified by its red/blue colour ratio: clear sky is blue, cloud is grey or white. The threshold varies with zenith distance and distance from the Sun. Sky below 10° altitude and fixed obstructions are excluded, and the area within 8° of the Sun is judged by whether the solar disk is visible. The Sun is low (full confidence above +10°), or too little usable sky remains after obstructions and over-exposed areas are removed.
Infrared sky temperature Any time, once trained Ambient minus sky temperature. Clear sky reads cold; cloud reads close to its own temperature. The sensor learns separate clear and overcast responses for day and night from the camera measurements, and votes only after it has proved it agrees with them. Until it has learned both clear and overcast skies, or while its recent agreement with the camera is poor. It never outranks a full-sky camera measurement.
IR-tracked carry Gaps, mainly twilight (Sun −6° to +3°) The last confident camera reading, carried forward and moved toward clear or overcast by the change in infrared sky temperature since then. The camera reading is more than an hour old, or there is no current IR reading.
Why Colour Rather Than a Clear-Sky Model

Earlier versions synthesized a clear-sky radiance image and thresholded the difference from the observed image. Because that difference was judged against its own median, it measured cloud texture and could not see a uniform overcast. The red/blue ratio is a direct physical test, used widely in all-sky cloud instruments. It is independent of exposure because a ratio cancels brightness, and it reads a uniform overcast correctly.

Two checks guard the colour test:

  • Clear-sky gradient: clear sky always reddens toward the horizon. A sky that reads clear but has a flat colour field, no visible Sun, and a single colour population is a blue-lit overcast, and is counted as cloud.
  • Sun check: if the Sun is above the horizon but its disk is hidden while the sky reads mostly clear, the colour reading contradicts itself and is not used.
Combining the Methods
  • Cloud cover is the confidence-weighted mean of the methods that voted. Overall confidence is 1 − Π(1 − ci), so a single strong method is enough.
  • When two confident methods disagree by more than 40 points, the disagreement is flagged and confidence is halved.
  • The published value is lightly smoothed with the previous one, so it does not chatter around a safety limit.
  • If no method can measure the sky, the last value is held for up to 5 minutes. After that, 100% (unsafe) is reported until a measurement returns.

Cloud cover, its confidence, and the method that dominated are published every 30 seconds. The infrared sensor is never calibrated against the system's own combined output: it learns only from star counts at night and colour by day.

Known Limitations
  • Thin cirrus and haze raise the red/blue ratio of clear sky and can read low in daylight.
  • Rain or water drops on the dome read as cloud.
  • Between about −6° and +3° Sun altitude, cloud cover depends on the IR-tracked carry and is reported with reduced confidence.
Image Quality and Seeing

Image Quality (IQ) is TychoCam's measurement of star sharpness, in pixels: the FWHM of a 2D profile fitted to each of a sample of stars spread over the central part of the sky, combined as an SNR-weighted median. Stars dimmed by cloud, stars near the lens edge, and elongated or implausible fits are excluded, so IQ describes star sharpness where the sky is clear. At an all-sky plate scale it reflects focus, optics, and pixel sampling; atmospheric seeing contributes only a small part.

Seeing (StarFWHM), in arc-seconds, is not measured by TychoCam. It is supplied by the site from an external instrument and reported to ASCOM/Alpaca clients as StarFWHM.

Sky Brightness, Sky Quality, and Bortle Class
  • Sky Brightness is the illuminance from the night sky in lux, as defined by ASCOM. The AS7343 spectral sensor measures the sky luminance L (cd/m²) from a weighted sum of its visible channels, and E = π × L is reported: the illuminance a uniform sky of that luminance puts on a horizontal surface. Direct light from the Moon itself is not included.
  • Sky Quality (mag/arcsec²) is derived from the same luminance measurement, and is self-calibrated on clear nights against a sky-background reference measured from the camera's star images.
  • Bortle class (1–9) is estimated from Sky Quality using fixed thresholds. There is no standard conversion, so it is approximate.

All three are reported only during astronomical night (Sun below −18°).

Optional Overlays and Diagnostics

For visualization and diagnostic purposes, optional overlays may include:

  • Solar and lunar positions
  • Sky coordinate grids
  • Constellation outlines
  • Environmental sensor values
  • Observing-condition metadata

During the day a cloud-mask image is also produced, showing each pixel's classification: cloud, clear, and the circumsolar zone.

These overlays do not affect cloud detection calculations.