Ground Truth: Verifying Stellarium's Model of The Solar System

Mick West

Staff member
ground truth, n.
1. A fundamental truth. Also: the real or underlying facts; information that has been checked or facts that have been collected at source.
2. a. In remote sensing: information obtained by direct measurement at ground level, rather than by interpretation of remotely obtained data (as aerial or satellite images, etc.), esp. as used to verify or calibrate remotely obtained data.
2. b. Information obtained by direct observation of a real system, as opposed to a model or simulation; a set of data that is considered to be accurate and reliable, and is used to calibrate a model, algorithm, procedure, etc.

The modern scientist's understanding of the universe is built largely on trust. They trust that the reference books and the scientific literature are largely correct. They trust that the mathematical models of gravity, gases and liquids, optics, the shape and rotation of the earth, and the sizes and motions of the planets are all accurate and reliable. They also trust that the information from their sensors, their thermometers, their spectrographs, their telescopes, other people's telescopes, and satellites is all generally correct.

This use of trust does not sit well with people who are habitually suspicious. Many people have a deep seated distrust of authority, of the government, and some of them see science as an extension of a somewhat malicious governmental authority. So when they see that science is based on a web of trust they naturally distrust it. For some people this distrust of science extends even as far as distrusting that the Earth is round.

The trust of science that scientist uses is not however a blind trust. It is a verifiable trust.

This verification can largely be described using the concepts of ground truth, as in the dictionary definition above. At its simplest ground truth is checking a source of data to see if it matches reality. For example if you buy a laser measuring device, you might (like me) not entirely trust that the numbers it is returning are accurate. So you'd probably do like I did and test it against a tape measure for a few distances.
A few of these ground truth verifications allowed me to form a verifiable trust of my laser measure. This then allowed me to use it in situations where a normal tape measure was hard to use. I now trust that this magical device gives accurate results, and I know I can verify it any time I like.

But how do we trust tape measures? That might sound like a silly question, but it's key to verifying the science behind the shape of the Earth and the solar system. We can trust tape measures because they all match.
There are literally billions of measuring devices out there, and they all match up. There's this massive web of verification that we can use to verify tape measures against each other. We can even verify them against publicly published standards, like this one at the Greenwich observatory where people could go and check their yardsticks.
Image source:

So I trust my laser measure because the ground truth matches my tape measures, which in turn can be verified against other more accurate measures. There's a self-correcting web of verification which allows me to trust a magical-seeming tool that measures distances with light.

What I want to do now is use this concept of ground truth to demonstrate how the individual can verify other seemingly magical evidence (like satellite images, or photographs of the sun) to form a self-correcting web of verification regarding the shape of the Earth and the nature of the solar system.

A quick word to the suspicious: this is not some attempt to brainwash you. This is very much a process that you personally should stop at every step and ask yourself "how do I know this to be true?" and "how can I verify this". Not only that, you should actually go and verify it yourself.

So let's start with one:

#1 Stellarium is accurate for any location, time, and magnification

Stellarium is a free program for your computer. It's also Open Source, meaning if you wanted to you could look at the underlying data and algorithms it uses. But the most important thing is that what you see in Stellarium is what you see in the sky. The ground truth here is simply what you see, and what you photograph.

To check your ground truth observations you need to set Stellarium to the time and location of that observation. For example I took this photo last night (May 7th 2017) at 9:50PM PDT
This is a ground truth observation. It's what I actually saw that that time. It's a photograph that anyone could take, an observation that anyone could make. If a program is accurate then it should match up with photos like these. Let's check this in Stellarium:
  1. Bring up the location window (F6 or use the compass icon on the toolbar)
  2. (optional) Select "Get location from network" (gives you more towns to choose from)
  3. Search for your location, a nearby town will do, I chose Folsom, about 10 miles away as my local town is too small to be included.
  4. Select the town from the list
  5. Check it shows up in the bottom left. You can close the location window now.
Now we have the location right we need the time:

  1. Bring up the Date and Time window (F5 or use the clock icon on the toolbar)
  2. Check it shows the correct local time. It should use the time set on your computer, but you should verify this and adjust if necessary.
  3. Pause it. Stellarium runs in real time, but we want to check a particular point in time.
  4. Set the date and time to match your photo
(If you are using Stellarium to verify a direct observation then skip steps 3&4)

We now need to change the type of "projection" that Stellarium uses to match what our camera is using. By default Stellarium is set to "Stereographic" which is the closest match for what we see with the naked eye with a wide field of view. However most cameras actually use "Perspective" (aka "Rectilinear"). This looks pretty much the same when you zoom in, just changes the angles slightly. The biggest difference is that the horizon remains straight. In the unlikely even that you have a fish-eye lens, you can select that.

Now you simply need to look in the direction you took the photo (South in my case) and see if it matches.

It's a match! Not only do the sizes and positions match, but also the phase and angle of the moon. The moon in my image is a little over exposed, but I took another a few minutes earlier at 9:41:32
Which we can check against what Stellarium has calculated the Moon should look like.

So that's one bit my own personal ground truth for Stellarium, and like my multitude of measuring devices this all matches. Not simply with this particular photo of the moon and Jupiter but with all my photos and direct observations of the night sky. Every single time I've checked it has been correct. And it's not just my photos, every time I've checked other photos (where the date, time and location are known) then what is shows in the photos (stars, planets, satellites, the Sun and Moon) matches what is show in Stellarium. So the verified web of trust extends ever outwards, built upon your personal ground truth.

So what does it mean, that Stellarium is accurate? Well Stellarium (and the other programs like that, including many mobile apps) is based upon the conventional model of the solar system and the stars. Stellarium is modeling the solar system as if the Sun is 93 million miles away from the Earth, the Earth and the other Planets rotate around the sun in slightly elliptical orbits, and the Moon rotates around the earth. It's also modeling your position and view angle from the Earth as if the Earth is a globe. So that means the conventional model of the Earth and the Solar System fits exactly with the ground truth observations.

Also, because we know that the Stellarium predictions match the view from any arbitrary time and place, we can use Stellarium to observe the sky. We don't need to travel, we don't need to go outside. We have extensively verified that Stellarium is correct using ground truth, so we can use it to make observations about how things look from various positions. For example we can use to to demonstrate that the pole star, polaris, has an angle in the sky equal to latitude, and that it's not visible from the southern hemisphere, and that the Earth does not appear to be flat.

So given this ground truth, given this useful tool for observing the sky from anywhere in the world at any time, we can go on to check some other things, for example.
  1. Solar system simulators (like Celestia) match Stellarium
  2. Latitude and Longitude are accurate for any location
  3. The angle of the noon sun at the equinox matches your latitude.
  4. The International Space Station tool accurately predicts where and when the ISS will be seen
  5. The height of the ISS can be triangulated.
  6. Satellite images match local cloud cover
  7. Several satellites give full-disk images of the Earth several times a day
  8. The ISS live feed matches satellite images.
  9. The apparent size of the sun and the moon do not vary through the day
In some cases the ground truth of one thing will depend on another. But ultimately it's a consistent web of verification that only works with a spherical Earth orbiting the sun, and the stars being trillions of miles away.
Last edited:
Excellent stuff. :)
Ultimately it's a consistent web of verification that only works with a spherical Earth.
I suppose we can say something similar for google maps. A great many "flat earth predictions" only work if something is seriously amiss with google maps - if distances are incorrect; if places aren't where they're presented as being; if countries are not the sizes we believe them to be - and yet we can certainly verify for ourselves whether it's accurate or not, to a very precise degree.

As far as I know, no one has ever shown even the tiniest discrepancy between what the software presents, and what we observe in reality.
Just a couple of ground truth pics that illustrate definition 2.a.

1) This is an approximate pixel size from an airborne spectroscopic image (hyperspectral imagery) survey in the Atacama Desert. It is important to visualize imagery resolution on the ground to know the scale that is being measured. The airborne data were collected with Specim AISA Eagle/Hawk "Dual" hyperspectral cameras from a fixed wing aircraft (Kingair as I recall) flying at approximately 3,000m AGL).


2) Taking ground based spectroscopic measurements to validate and calibrate airborne measurements. Scale change is an issue as the ground instrument measures an area a few mm in size vs the pixel sized area shown above. We take many measurements over a pixel size area and average the results to validate the airborne data. The spectral range for this survey is 450 nanometers (nm) to 2450 nm with a spectra resolution of 3nm to 7nm. The blue box is a Spectral Evolution Inc spectrometer. Pistol device is the measurement probe (right hand) and light source (connected to the spectrometer by a fibre optic cable). The small square box (left hand) contains a reflectance standard for instrument calibration (basically a white material that has a "flat" spectral response).


3) Ground truth results showing airborne measurement in red and average of field measurements in green. Only the shortwave IR is shown (~2100nm-2400nm). The spectral signature is the mineral alunite.