LUAstro Newsletter - February 2026
A Journey into the Heart and Soul Nebulae; gently interrupted by satellites
Welcome to LUAstro’s February newsletter where we focus on the data processing of astrophotography, with a piece about how we deal with the growing problem of satellites interfering with astronomy!
The Heart and Soul Nebula
This month the society enjoyed one exceptionally clear night, on the 13ᵗʰ-14ᵗʰ, Valentine’s Day, upon which we took ten hours of exposures on W3/W4/W5 complex, the Heart and Soul nebula. We recorded 120 five-minute broadband exposures, resulting in a total integration time of 10 hours and the final image does not disappoint.
The field of view of the RedCat51 telescope paired with an ASI071MC camera is 5.4 x 3.6 degrees, so this image spans more than ten times the apparent diameter of the full moon. At 7,000 light years away, this picture depicts a region of space almost 700 light years across.
All of the stars in the image can be removed with software, such as with StarNet, to expose the nebulosity of the image with greater clarity. Removing the stars is often an important stage in astrophotography processing, because it allows one to brighten the background nebula without overexposing the stars. To isolate nebulosity, StarNet uses a neural network where astrophotographers would historically use masks. The mottled blotchiness is an artefact of the removed stars.
Conversely, one can appreciate the stars alone, whose varying density highlight the structure of the dark absorption nebulae. It is this prevalence of optically-thick nebulae that can motivate infrared astronomy, which probes light wavelengths less attenuated by interstellar dust.
Satellite Mitigation
Artificial satellites have an increasing notoriety among astronomers. Indeed, one quarter of our data, amounting to 2.5 hours of exposures were ‘spoiled’ by satellites passing through the frame, as depicted below. Despite this, when stacking the full 10 hours of our data, we only omitted two sub-exposures, and their removal was not motivated by satellites. Stacking is the process of averaging all of the sub-exposures into one image with a higher signal-noise-ratio (SNR), enabling one to reveal fainter image details.
When stacking, we perform a technique called sigma clipping - If a pixel within a sub-exposure falls three standard deviations from its corresponding pixels in the other subs, the pixel is rejected and not stacked. This deals with the brighter satellites, although it does accompany a mounting loss of SNR if satellites repeatedly pass through the same part of the frame.
The fainter satellites, within three standard deviations of the data, are dealt with by the stacking process itself - they get averaged out where the majority of frames do not have any satellite passing through.
With an increasing number of satellites in the sky, these techniques may not be as viable in future. Then our only recourse might be to utilise narrowband filters, which are less susceptible to light pollution.
Analysing the satellites from our 10 hour data
Curious, we counted all satellites that passed through our sub-exposures, and plotted them against the timestamp stored in the exposures’ filenames in Figure 1. We can not guarantee that each bright streak was a satellite; some looked like aircraft as their brightness varied periodically across their transit.
Our theory is that the pause in satellites towards the middle of the night is not random, but significant, with the number of satellites in each sub-exposure n at any time t being described as a function of solar altitude α
where δ is the sun’s declination, -13° 10’ 15.4’’, ϕ the observer’s latitude and ω the hour angle in local time.1 While the sun is so far below the horizon that it only faintly illuminates the atmosphere during astronomical twilight, satellites orbit above the atmosphere; the altitude of Starlink satellites is around 500 km. At this height, they can still reflect sunlight unless the sun is very far below the horizon. Consequently, we expect this relation between solar elevation and observed number of satellites in each sub-exposure.
We have to correct our timestamps for the observer’s longitude of 2.78396° here; our local time is some 11.1 minutes behind Greenwich Mean Time, so the sun sets slightly later than it does in London. The hour angle ω is therefore
because the sun moves by 15° every hour, as a consequence of 24 hour days. As the number of satellite incidences per hour may not be negative, we need to add some vertical offset to n which is hence modelled by
The fit of this model is displayed in Figure 2, producing the numerical relationship
from which we might infer that when the sun is at zenith, α = 90°, an average of four satellites would pass through our field of view every five minutes. It seems reasonable, but we doubt its veracity.
If anyone is curious, the length of a day, as measured from the frequency of these satellite incursions, is 24.5 ± 2.3 hours!2 It is fascinating that this is a viable method of measuring the duration of a day, but I think we may be in need of more than 10 hours of data.
Hue-Stretching
When processing the Heart and Soul nebula data, Finlay experimented with using Python to stretch the colours of the image. In astrophotography, stretching typically describes the process of brightening the data; rescaling it such that all of the data is not confined to the darker parts of an image. In reality, stars are significantly brighter than nebulae, so we have to stretch photographs to see both within the same display dynamic range.
Stretching the hue of an image accentuates the subtle variations in colour across the image, and has the specific effect in astronomy of exaggerating the differences in elemental composition of the gas. This hue-stretch is probably not of any scientific use, but it does almost mimic narrowband imaging, where each colour channel is assigned to the light emitted by a different element.
This algorithm identifies the peak in the image’s histogram, the brightness level at which most of the pixels in the image share. It then considers a specified number of standard deviations either side of this peak, finding the minimum and maximum hue of the data in that region. It then stretches this hue to a greater range of colour, the effects of which can be seen below.
The colour of the original image is dominated by the orange-red of hydrogen-alpha, at 656 nanometre wavelength. Stretching the image colour in the 1σ region casts the nebula from green to purple, but these new colours do not seem to align with Hubble’s SHO palette suggesting that they are not a strong indicator of elemental composition. If time permits, we may experiment further with this - a subtle stretch has the potential to be quite pleasing.
Leo O’Hara’s View of Orion
With an impressively short integration time of only 20 minutes, Leo O’Hara took this wide-field photograph of the Orion molecular cloud complex, capturing many nebulae. The famous three stars Alnitak, Alnilam and Mintaka of Orion’s belt are prominent in the right of the image.
We are particularly grateful to Leo this month for helping us track down the source of some aggressive noise in our Heart and Soul nebula picture. We eventually tracked it down to some of our flat calibration frames that had defectively low quality.
Thank you to Leo, and let us hope for some more clear skies in March!
Ahmad S, Shafie S, Ab Kadir MZA, Ahmad NS. On the effectiveness of time and date-based sun positioning solar collector in tropical climate: A case study in Northern Peninsular Malaysia. Renew Sustain Energy Rev. 2013;28:635–642. doi:10.1016/j.rser.2013.07.044.
We found the duration of the day by changing ω to equal
where D is the duration of a day. We then fit a curve to the plot using SciPy, returning the value of parameter D alongside its covariance.