This is about my
project, something I finished in August of last year. I’ve
been on a long journey with some other Java-related
projects more recently, but I’d simply like the chance to
go back to one of my most significant C projects I’ve ever
written. I think it would also be a good chance to revisit
what I learned in the process of writing
fate as well,
because Tim Dodd of Everyday Astronaut posted a video
recently that had some
discussion about inclination and azimuth, and despite
wrangling with that for weeks to understand how it worked,
I still had to pause the video and think through what he
The project is actually very simple on the surface, if you have the most recent TLE (more specifically 3LE) data, then you can input that into the command line and every second, some output will be printed to show you above what point the satellite is as well as where to point a telescope or your eyes to find the object in the sky.
What is TLE?
TLE stands for two-line element set. You’ll actually notice that I use 3LE, which stands for 3-line element set, and it is exactly the same, save for the fact that a 3LE set has an additional line at the top that states the name of the satellite. This is what a 3LE set looks like:
0 ISS (ZARYA) 1 25544U 98067A 19174.05036204 .00001525 00000-0 33398-4 0 9995 2 25544 51.6442 332.2030 0008185 74.7737 30.3832 15.51219884176168
A TLE set would just be the same thing but with the line
0 removed. For the purposes of
making things easier to understand, I’ll primarily refer
to both 3LE and TLE as just TLE.
The Wikipedia page for TLE sets is actually surprisingly helpful and informative. The main gist is that TLE is a way of communicating the position and motion of a satellite orbiting the Earth in as few numbers as possible. TLE sets are readily available from SpaceTrack.org’s TLE search page, and you can look for satellites in the satellite catalog. The U.S. Air Force 18th Space Control Squadron produces the TLE data sets by using a variety of sensing techniques and calculating the TLE set data for publication on the SpaceTrack website.
Here are some additional reading items you might find helpful:
It’s all well and good that we have the data in TLE format, but how can that data be translated into usable numbers that show us where to point a telescope for example?
The first step is to refer to SpaceTrack Report #3. I’ll let the author(s) summarize:
The NORAD element sets [TLE sets] are “mean” values obtained by removing periodic variations in a particular way. In order to obtain good predictions, these periodic variations must be reconstructed (by the prediction model) in exactly the same way they were removed by NORAD. Hence, inputting NORAD element sets into a different model (even though the model may be more accurate or even a numerical integrator) will result in degraded predictions. The NORAD element sets must be used with one of the models described in this report in order to retain maximum prediction accuracy
Essentially, the report lays out the mathematical models for reconstructing the orbit of a particular satellite from the data given by the TLE data set. Now there are 5 different models that the report lays out, SGP, SGP4/8 and SDP4/8. These are collectively known as “simplified perturbation models” and take into account atmospheric drag, gravitational drag caused by the Earth’s oblate shape, the Earth’s spin, and various other factors in order to predict the motion of the satellite over time.
Since my primary goal was to figure out the position of
the International Space Station, I selected SGP4. As far as
I am aware, SGP4 and SDP4 are the most commonly used
models, this can be checked with the
entry for each satellite entry’s OMM data.
Translating all of the formulae into working C code was not super challenging. There is even working FORTRAN that I referred to whenever I was having trouble figuring out what the intent of a formula was. However, there are quite a few variable values that are missing as well as this part:
Solve Kepler’s equation for (E + ω)
and the changes made to the model with a perigee at different distances, which really confused me. That being said, combined with looking at the FORTRAN listings on SpaceTrack Report #3 as well as with the LizardTail website source and the Revisiting Spacetrack Report #3 code listings in the appendix, I was able to reconstruct the entire mathematical model in C code with most of the constants updated to the modern values. I don’t think it is entirely perfect, but the numbers it produces look pretty correct to me nonetheless.
Now if you ask me, I’d say that the finer details of the model itself aren’t actually that important. I can’t say for sure what the purpose of each and every calculation is. Again, the model takes into account the many different variables that affect the gravitational pull and drag experienced by a satellite, but that is as far as the extent of my knowledge about the perturbation models goes.
(Apologies to those readers who may have clicked on this to get an understanding of how the perturbation models work, that’s simply something I never even needed to know to implement the model in code. If anyone does understand, I’d love to learn)
Conversion Between Coordinate Systems
Now having implemented the SGP4 model, you might think that
we can now extract the data we need. Not so. The SGP4 model
produces 2 vectors, specifying the position and velocity
(meters per second) of the satellite.
provides an additional 2 vectors called “unit orientation
vectors.” These are used to derive the position and
The problem is that the reference frame for the position vector uses the ECI coordinate grid, which means that we get 3 values in X, Y, and Z. This doesn’t help me, because all I want to know is latitude and longitude to the ISS.
I won’t go into specifics, but these three articles are extremely informative and detailed, and even an idiot like myself was able to understand what is being discussed. I highly recommend reading the entirety of the following:
- Orbital Coordinate Systems, Part I
- Orbital Coordinate Systems, Part II
- Orbital Coordinate Systems, Part III
As far as the high-level overview goes, it is worth taking a look at what ECI really is. ECI stands for “Earth-centered inertial,” Earth-centered meaning that the origin is at the center of the Earth and inertial meaning that it doesn’t move with the rotation of the Earth itself.
Calculating the look angle in azimuth rotation from true north and inclination angle from the horizon is relatively complex because you need to turn your own latitude and longitude into ECI coordinates as well and utilize some trigonometry to determine the angle created between the coordinates. Not so hard, right? Unfortunately, the complication comes from the fact that the Earth spins, causing a number of issues. Firstly, this means that the Earth is not a perfect sphere; it is in fact an oblate spheroid that bulges slightly at the equator. Because the standard latitude and longitude account for the ellipsoidal shape of the Earth, they are said to be geodetic. On the other hand, ECI considers the Earth as a perfect sphere. If that wasn’t enough, the Earth’s spin also means that your position is dependent on time.
In order to calculate everything, the current time is taken, and then converted to a single Julian date. This is then converted to Greenwich Mean Sidereal Time (GMST), which allows one to determine the rotation of the Earth without the fluctuations in a solar day. This solves the second problem, locating the observer taking into account the Earth’s rotation. Now, we can use trigonometry to convert from the geodetic latitude and longitude to the geocentric ECI coordinates. As a matter of fact, the implementation of this part the implementation looks deceptively simple. Now that we’ve solved both problems, we then have the coordinates of both the observer and the coordinates of the satellite in the same reference frame. Since both are in ECI coordinates, we can just use simple trigonometry to determine the angle which to point an instrument or the angle to look at in order to locate the satellite.
Calculating the position of the satellite above the Earth (called the sub-point) is a bit more simple. Essentially, we need to reverse the process of finding look angle and convert ECI into geodetic coordinates. Because longitude runs parallel to the Earth’s oblateness, we only need to factor in the current time in GMST to determine the Earth’s rotation. Then, we can compute the longitude with trig. To calculate geodetic latitude, we first calculate the geocentric latitude, which is rather straightforward trig. Since latitude is affected by the Earth’s oblateness, we then run a transformation which moves the angle closer and closer to geodetic latitude, until the diffence in improvement to the value becomes smaller than is worth calculating. This value is then close enough to the geodetic latitude to accept.
This has only been a high-level overview of the calculations needed to convert the available data into a usable format, and then converting that format into something that is understandable, like geodetic latitude and longitude, and the look angles. I myself don’t even know all of the specifics. Working on this project was a fascinating insight into the work done by astrophysicists and mission planners to determine how to get satellites and rockets into the correct orbits, and not only that, but to track them and create models for the orbital mechanics that affect the motion of the satellites through space.
Not only do I not usually talk about the C language, but I didn’t really go into any specifics of it in this particular blog post. That being said, I did talk a little bit about astronomy and space, which are both topics that I’m very curious to learn more about. I’m sure that every one of us watched a rocket launch, watched Neil Armstrong take humanity’s first steps on another planet, or simply read the news about the Opportunity rover. I’m absolutely certain that others have been inspired by spaceflight and can relate to wanting to advance space exploration in the future, if not the present.
I, for one, certainly would like to.