PDA

View Full Version : Future apparitions of Comet Halley



Ara Pacis
2017-Mar-21, 06:17 PM
Does anyone have a list, or know where I can get a list of future apparitions of Comet Halley for about the next 1000 years? I've looked online and only see apparitions listed for 2061 and 2134. The best info I can find says that it varies between 74-79 years due to perturbations and numerical integrations can't be done more than a couple thousand years either direction, and those are variable by on the scale of a week for each apparition due to unpredictable outgassing. I looked at some planetarium programs, but Stellarium won't run and Home Planet doesn't get the last few or future few predictions anywhere close.

Or maybe it's not predictable at all, if I read the conclusion in this paper (https://arxiv.org/pdf/1409.7762.pdf) correctly.


Finally, we have computed the Lyapunov exponent for the present day Halley’s orbit and a series of other orbits differing in a and e by an amount equivalent to the observational uncertainty. We have found that L is greater than zero with a value of approximately 10−2, indicating that the orbit is indeed chaotic. The corresponding timescale for the prediction of Halley’s orbit to within present day observational constraints is less than 100 years, suggesting that the orbit of Halley’s comet can not be accurately predicted for timescales much greater than this

Although this paper (https://arxiv.org/pdf/1606.07037.pdf) suggests Lyapunov is closer to 300 years.

We find the Lyapunov time scale of Halley’s orbit to be of order 300 years, which is significantly longer than previous estimates in the literature. This discrepancy could be due to the different methods used to measure the Lyapunov time scale. A surprising result is that next to Jupiter, also encounters with Venus contribute to the exponential growth in the next 3000 years.
They claim Venus is the main perturber for the next few thousand years until Jupiter takes over. Still, the paper doesn't list any dates for future apparitions. A video here (https://www.youtube.com/watch?v=0G6vnq1TcRk) by one of the authors might be useable if it wasn't moving so fast or if I could DL it to do frame by frame analysis. Putting it at .25 speed and using the slider at full screen reveals they have an orbit taking ~150 years instead of ~76. So... :wall:

I found another post (https://forum.cosmoquest.org/archive/index.php/t-159729.html) linking to http://orbitsimulator.com/gravitySimulatorCloud/simulations/1452654910254_halley.html, but how accurate is this? The dates it renders for perihelion (guessing by looking at the map) are:
2061 Jun 28
2134 Mar 23
2209 Jan 26
2284 May 19
2358 Jun 04
2430 Jul 02
2504 Aug 16
2579 Aug 10
2653 Nov 27
2726 Jan 16
2795 Nov 11
2863 Nov 9
2931 Sep 17
3000 Aug 22

No wonder writers give up on hard SF and just go for science fantasy or space opera.

Eclogite
2017-Mar-21, 06:48 PM
No wonder writers give up on hard SF and just go for science fantasy or space opera.I can appreciate your frustration, but isn't the hard science here that small objects, especially those that outgas randomly, are not going to follow a predictable orbit. That immediately suggests at least five plot lines for hard SF writing to me.

antoniseb
2017-Mar-21, 06:49 PM
You can still be hard science fiction if you are wrong about a date for Halley's comet's future perihelion. Pick a plausible date.

Ara Pacis
2017-Mar-21, 07:46 PM
I guess I just want to know how much faith I can put into orbitsimulator and my extrapolations of meaning of listed apparitions (brightest appearance or point of closest approach instead of perihelion). Afterall, Orbitsimulator shows the moon orbiting the sun well away from Earth. I'd have thought NASA would publish this sort of thing, but I can't find it. I don't need perfect accuracy, I just want to know within a year or so. I already have a timeline in mind. It's not a plot point or an anchor point, but a reference point in a story that may or may not be important depending on where it is.

antoniseb
2017-Mar-21, 08:27 PM
You've already identified someone that appears to have taken perturbations from Venus and Jupiter into account. I think you're covered within about a year, and probably within a few weeks out through 3000. Yes some unexpectedly large variation in outgassing acceleration could give it a bigger or smaller impact from Jupiter or Venus (or Earth) and we could be off by a few years, but odds favor your above date being closer than that.

CJSF
2017-Mar-21, 08:51 PM
I agree with antoniseb. If there are elements of uncertainty built into the far future apparitions (that is, the uncertainty isn't from our inability to calculate, but actual chaotic variations from the comet itself), then you're fine. Unless you're worried that in a few thousand years someone will read your stories and laugh at your missing the date?

CJSF

Noclevername
2017-Mar-21, 08:54 PM
No wonder writers give up on hard SF and just go for science fantasy or space opera.

“It's tough to make predictions, especially about the future" --Yogi Berra

Ara Pacis
2017-Mar-22, 04:59 AM
Thanks everyone. I appreciate the help. I started the OP without any ideas and discovered the orbitsimulator link after I was ready to post. But I wasn't sure what to make of it because I thought it was off by 45 days in 2134, only afterwards realizing they were measuring different things. Sorry if I seemed impatient, I didn't sleep well and felt groggy all day.

Still confused about how John Walker's Home Planet could be so wrong and why that video from one of the researchers (Zwart) seems to suggest a period of 150 years. I also finally figured out how to find it on JPL Horizons -their search was picky on spelling- but they only predict out to AD 2500. I'll just roll with orbitsimulator.com for now. :)

ngc3314
2017-Mar-22, 01:11 PM
As I read the language in that first quotation,


orresponding timescale for the prediction of Halley’s orbit to within present day observational constraints is less than 100 years,

the 100-year timescale is for departures from its present extrapolated orbit to become large enough for current data to measure, not for major changes item orbit or perihelion date (although of course after enough orbits those become almost arbitrarily large if it's really chaotic). The precision of current data for a point on the orbit would be no worse than hundreds of km, so there's a lot of room for that to expand before making huge changes beyond what Newtonian modeling predicts.

Tom Mazanec
2017-Mar-23, 11:37 PM
“It's tough to make predictions, especially about the future" --Yogi Berra
I quoted this to a visitor who bought most of my sci-fi library about twelve years ago, went by the name Cubist (he was a wizz at Rubik's Cube).
He said "It is easy to make predictions about the future. It is hard to make ACCURATE predictions about the future."

tony873004
2017-Mar-24, 01:05 AM
I found another post (https://forum.cosmoquest.org/archive/index.php/t-159729.html) linking to http://orbitsimulator.com/gravitySimulatorCloud/simulations/1452654910254_halley.html, but how accurate is this?

That's my post. It's a numerical integration that takes perturbations of all the planets into account. The same simulator does a good job with the planets and major asteroids when comparing their propogated positions to known historical positions, or to credible future predictions. But you've already figured out that comets are different
due to unpredictable outgassing
There's not much I can do to account for unpredictable forces, so I would guess that the predictions from my simulation are as good as anyone's.

You can press [<<] on the Time Step interface to run backwards in time and compare the simulation to known past encounters.

Press "A" on your keyboard for more viewing options.
Under the "Objects" menu, choosing "Vectors" lets you see distance and velocity information.

For slightly better accuracy, you can include the largest asteroids. On your keyboard, press "I" for import. Then type (or copy & paste): ceres, vesta, pallas, hygiea, euphrosyne, interamnia, davida, herculina, eunomia, ASTNAM=Juno, psyche, ASTNAM = europa, ASTNAM = thisbe, ASTNAM = iris, ASTNAM = egeria, ASTNAM = diotima, ASTNAM = amphitrite, ASTNAM = sylvia, ASTNAM = doris
Then press the [Import] button. Wait about 20 seconds while the program contacts JPL Horizons for the data.

tony873004
2017-Mar-24, 01:22 AM
Afterall, Orbitsimulator shows the moon orbiting the sun well away from Earth.
The Moon is quite accurate. The simulator successfully predicts solar eclipses centuries into the past and future. In the View menu you can choose your city or lat,lon, then press "A" on your keyboard, and view "Sun" or "Moon" from "Earth".

If the Moon is orbiting the Sun solo, and away from Earth, you may have set the time step too high. Don't go past 16384 seconds. Numerical integration is subject to bigger errors when the time step is high.

Press "A" on your keyboard, then choose Earth from the top dropdown list. Then zoom in to see that the Moon is clearly orbiting Earth.

Ara Pacis
2017-Mar-24, 06:17 AM
I thought that site looked familiar. I forgot it was one of our own who made it.

So, the speed was the issue. I was running it at 65k or 130k because I was feeling impatient. Is it more accurate if I enter the date manually? I tried that but nothing changed.

I'll have to explore it more fully.

DonM435
2017-Mar-24, 12:30 PM
I was watching the original (1954 Japanese edit of) Godzilla just last night, and whatever the eminent professor said in his native language, the English captions had him claiming that the Jurassic age was ten million years ago. He still managed to fit the Cretaceous period into his timeline, so he's probably up to the Eisenhower era.

CJSF
2017-Mar-24, 01:44 PM
I was watching the original (1954 Japanese edit of) Godzilla just last night, and whatever the eminent professor said in his native language, the English captions had him claiming that the Jurassic age was ten million years ago. He still managed to fit the Cretaceous period into his timeline, so he's probably up to the Eisenhower era.

Hey! I was watching that too! They also noted that Gojira, as a Jurrasic monster, was 2 million years old. I don't know if that's down to (mis)translation or if it was in the actual script. Do any of our Japanese-speaking members who've watched Gojira know?

CJSF

tony873004
2017-Mar-24, 02:53 PM
s it more accurate if I enter the date manually? I tried that but nothing changed.

No. Entering the date manually is for telling the simulation the epoch of the position and velocity vectors. It doesn't propagate the system to that date.

If you want to speed it up without using the time step, you can open the Preferences menu and change the number in the "Do Events" box. By default it is set to 100. This means do 100 iterations of math for every time you update the screen graphics. Higher means more speed vs poorer graphics (round orbits look like octagons if the graphics are updated 8 times per orbit.), but the accuracy of the simulation is not affected.

You can also speed things up by closing all other browsers and tabs so you get more CPU cycles to the simulation.

DonM435
2017-Mar-24, 04:12 PM
Hey! I was watching that too! They also noted that Gojira, as a Jurrasic monster, was 2 million years old. I don't know if that's down to (mis)translation or if it was in the actual script. Do any of our Japanese-speaking members who've watched Gojira know?

CJSF

That would produce an ethical/artistic dilemma for the translator: do you correct the error, or respect the original text?

publiusr
2017-Mar-24, 09:36 PM
It makes a showing in the year 3000--nice.

Ara Pacis
2017-Mar-24, 10:49 PM
No. Entering the date manually is for telling the simulation the epoch of the position and velocity vectors. It doesn't propagate the system to that date.

If you want to speed it up without using the time step, you can open the Preferences menu and change the number in the "Do Events" box. By default it is set to 100. This means do 100 iterations of math for every time you update the screen graphics. Higher means more speed vs poorer graphics (round orbits look like octagons if the graphics are updated 8 times per orbit.), but the accuracy of the simulation is not affected.

You can also speed things up by closing all other browsers and tabs so you get more CPU cycles to the simulation.

Does the "Period =" refer to the time of the actual period or to the osculating orbit? I ask because I noticed it fluctuates a lot within orbit and between apparitions.

I changed "Do Events" to 10,000 and ran it back to ca. 23,000 BC and saw a few instances between ~13,000 BC and 9,600 BC where the Period of Comet Halley was close to 90 years. It took a couple hours at 16k with "Do Events" at 10,000. I was gonna run it forward to double check the perihelia to see if the period was actually 90 years between apparitions and set "Do Events" to 100,000 to see if it would go faster, but it kept locking up Firefox and I had to close that tab. Is there a limit for "Do Events," or might that be a browser or memory issue? I want to go back to that era as fast as I can without losing accuracy or crashing the script. Part of my story references ancient astronomical events, and if it really is 90 years during that period of time, that's an amazing coincidence with already existing plot points.

Is there a thread on Orbitsimulator here on Cosmoquest? I looked on that site for documentation and watched the videos but still have questions. I glanced at the gravitysimulator board, but not sure if that's the appropriate place.

Is it just JPL objects that can be included, or can I add other objects, such as labels for various LaGrange points (or objects orbiting them) and cyclers (ballistic or semi-ballistic) like the Aldrin Earth-Mars cycler?

What I'd like to do, for my Hard SF story, is construct a realistic space transit and distribution infrastructure for a specific period of time in the distant future. The story is structured, to some extent, on transit times between planets and stations in various sun-planet LaGrange points and how that relates to timeline anchoring events on various planets. Some of these are plot points, and some aren't, but some might be depending on orbital mechanics I'm still unsure of. Right now, I'm writing around these issues, leaving a margin for various future propulsion capabilities, but as I start filling in my timeline with detail and chapter drafts, I'm starting to need more specific information.

Part of me thinks that since I want to promote STEM, I should occasionally highlight real orbital mechanics, the other part says most people won't care. But since people worked out the dates of The Martian, maybe I should get everything as accurate as practicable so that those who want to have fun digging into the mechanics behind the story will be pleased to find it's not made up. I don't know if it's worth the time and effort to learn to patch conics myself, or use Kerbal or STK to model actual space vehicles and trajectories. Or if it is, should I learn to do it myself, find a partner to collaborate with, or contract it out. I don't know how much work it would be to calculate ballistic and powered trajectories close enough to be within a day.

Should I separate these questions into different threads for Comet Halley, Orbitsimulator, and calculating interplanetary trajectories?

tony873004
2017-Mar-25, 02:07 AM
Does the "Period =" refer to the time of the actual period or to the osculating orbit? I ask because I noticed it fluctuates a lot within orbit and between apparitions.
I added the "P =" to specifically address the question in the thread you mentioned: "What is Halley's Comet's Period?" Off the shelf, the program isn't designed to do this, so I added code to the Autopilot (see menu Autopilot > Per Graphic Update) that tells it with each graphic update to recompute Halley's Keplerian orbital elements, and compute the period from semi-major axis. So every time it updates the graphics, it computes Halley's period as if only the Sun and Halley existed. So the fluxuations you see are the effect the planets have on Halley's orbit.


I changed "Do Events" to 10,000 and ran it back to ca. 23,000 BC and saw a few instances between ~13,000 BC and 9,600 BC where the Period of Comet Halley was close to 90 years.
You want to draw more of a statistical conclusion. Rather than conclude that in between those years Halley's period was 90 years, it would be better to conclude that the simulation demonstrates that perturbations by the planets are capable of changing Halley's period by decades. Due to chaos, the error in Halley's "along-track" position grows exponentially. A small error in the "along-track" position might be the difference between a close encounter with Jupiter and a far one.


...but it kept locking up Firefox and I had to close that tab. Is there a limit for "Do Events," or might that be a browser or memory issue?
Browsers don't like it when scripts keep control of them for more than a few seconds. You can try Chrome or Safari, or Edge, but they will all have their limit as to what they're willing to put up with. That's why I included "Do Events".


I want to go back to that era as fast as I can without losing accuracy or crashing the script.
That's a drawback to numerical simulations. The program has to do A LOT of math to take into account the perturbations of all the planets. But in a case like Halley, its really the only way it can be done. So just be patient and let the computer do its thing over a few hours.

You can always set pause points so you can leave the computer unattended without missing your target date. Add the following lines of code to menu: Autopilot > Single Events.


1/1/2000 UTC {pause()}
1/1/2001 UTC {pause()}

This will cause the program to pause on those dates. Sorry, BC isn't supported in the Autopilot.
You can also save the program when you reach your target date. Menu File > Save As.
This will save it on my server and provide you with a link. It will also generate and display the HTML code incase you want to copy it and paste it into a text editor to save on your own computer. Javascript doesn't let me directly save files to your computer for good reasons.


Is there a thread on Orbitsimulator here on Cosmoquest? I looked on that site for documentation and watched the videos but still have questions. I glanced at the gravitysimulator board, but not sure if that's the appropriate place.
I don't think there is one. At least I haven't started one. Feel free to start one here if you like. You're welcome to post to the Gravity Simulator board, but it doesn't get much traffic any more. The documentation isn't that good. Part of the reason is that I keep adding new features. Perhaps if enough people request it, I'll make some better documentation and Youtubes.



Is it just JPL objects that can be included, or can I add other objects, such as labels for various LaGrange points (or objects orbiting them) and cyclers (ballistic or semi-ballistic) like the Aldrin Earth-Mars cycler?
Yes, you can add anything you want. Under the Objects menu is "Create Objects". That lets you create objects based on orbital elements. Or you can edit an existing object and change its position and velocity vectors to what ever you want. Press "E" on your keyboard, or choose menu Objects > Edit Objects.

An Earth-Mars cycler needs to have adjustment burns. If you know exactly when and how much delta-v you need, you can use Autopilot to schedule these.


Part of me thinks that since I want to promote STEM, I should occasionally highlight real orbital mechanics, the other part says most people won't care.
I teach high school. During an open house I was assigned to answer questions in the Maker Space about our STEM program. Just so the room looked 'techy' and all the computers looked 'busy', I displayed this simulation : http://bit.ly/2nxncsO on one of the unused computers. I got as many questions on the simulation as I did on the 3d printers.


But since people worked out the dates of The Martian, maybe I should get everything as accurate as practicable so that those who want to have fun digging into the mechanics behind the story will be pleased to find it's not made up.
The Martian made an effort to get the Earth-Moon cycler accurate. Most people wouldn't know the difference, but many of us like it when the fine details aren't fudged.


I don't know if it's worth the time and effort to learn to patch conics myself, or use Kerbal or STK to model actual space vehicles and trajectories. Or if it is, should I learn to do it myself, find a partner to collaborate with, or contract it out. I don't know how much work it would be to calculate ballistic and powered trajectories close enough to be within a day.
I don't know STK. AFAIK, Kerbal doesn't do n-body (numerical integration), which is probably a must for this type of project. I think Kerbal just does Keplerian orbits.


Should I separate these questions into different threads for Comet Halley, Orbitsimulator, and calculating interplanetary trajectories?
You would probably get better responses if general Orbitsimulator or trajectory questions were in their own thread instead of buried deep inside a Halley thread.

Ara Pacis
2017-Nov-14, 04:37 PM
Hey, sorry for the long break and resurrection of this thread (family health stuff - good prognosis), but how do I import Comet Halley or other comets into OrbitSimulator.com? I've been able to import the asteroids listed above and others by by using their various reference numbers. However, whenever I try to import Halley, I get some sort of JPL error (or it pulls up the asteroid named Halley). I've tried ASTNAM=, COMNAM=, and NAME= with variations on 1P and Halley. I found a JPL Horizons instruction document online that lists Halley as an example, but it doesn't work for me.

Also, still thinking of starting a thread on OrbitSimulator, once I have my questions sorted.

tony873004
2017-Nov-15, 07:39 PM
JPL Horizons is hit-or-miss with comet names.
Try 900033 for Halley.
This gives you their 1994 solution. They have several solutions because of the non-gravitational forces.


Record # Epoch-yr Primary Desig >MATCH NAME<
-------- -------- ------------- -------------------------
2688 1982 HG1 Halley
900001 -239 1P Halley
900002 -163 1P Halley
900003 -86 1P Halley
900004 -11 1P Halley
900005 66 1P Halley
900006 141 1P Halley
900007 218 1P Halley
900008 295 1P Halley
900009 374 1P Halley
900010 451 1P Halley
900011 530 1P Halley
900012 607 1P Halley
900013 684 1P Halley
900014 760 1P Halley
900015 837 1P Halley
900016 912 1P Halley
900017 989 1P Halley
900018 1066 1P Halley
900019 1145 1P Halley
900020 1222 1P Halley
900021 1301 1P Halley
900022 1378 1P Halley
900023 1456 1P Halley
900024 1531 1P Halley
900025 1607 1P Halley
900026 1607 1P Halley
900027 1682 1P Halley
900028 1682 1P Halley
900029 1759 1P Halley
900030 1759 1P Halley
900031 1835 1P Halley
900032 1910 1P Halley
900033 1994 1P Halley

Ara Pacis
2017-Nov-16, 09:05 PM
Hi Tony, thanks for responding so fast. I have a few question on Halley, but might be for anything in OrbitalSimulator or orbital mechanics in general.

Do I need to propagate the sim to the date/epoch of the ephemeris before importing or does the code sort it out? For now, I just used your Halley sim (linked above), added some asteroids and ran it to the future date I wanted at time step 2048, then saved it (link (http://orbitsimulator.com/gravitySimulatorCloud/simulations/1510674362976_OS-2930%20Sol%20Sys%20Sim%201.html)). Now I'll try to setup interplanetary transit trajectories. But, if I decide to add something in, like more comets, the moons of Jupiter or the new Planet 9 they're looking for (if they find it), will I have to start all over? Or does JPL send ephemerides for that date? I'm still a newb.

Also, a general question about apparitions of Comet Halley (or comets in general). In the sim in the 2931 apparition, it passes through it's highest point north of the ecliptic, and between Earth and the Sun, at .44 AU closest approach on August 8th. Would it be visible in the daytime sky? It also reaches elongations (is that the right term for a comet?) of about 45° west then east as it gets close to the sun and Earth, making it visible for a couple hours before sunrise then after sunset. As far as apparitions go, would this be considered good (because of proximity) or bad (because of daylight and sun)? Actually, I'm not even sure how close to the sun it has to be to start growing a coma and tail (within 3-5 AU is what I've read, but that's not very precise).

Incidentally, it also comes fairly close to Venus (.192 AU) and Mercury (.254 AU) too. Two approaches near Mars are closer than not, first at .344 AU inbound, and then at 1.181 AU outbound.

Fun story coincidences
Some of the characters in my story will be in space around Earth during this timeframe (either LEO; SEL 1, 4 or 5; or on the moon - I'm not sure yet). I'm wondering how good it would look. The moon will be ~3 days past new so the comet won't be visible to most of the near side, but anyone near the lunar north pole shadowed craters would get a good look. Another possible interesting view might be from the central pacific, south of Hawaii, on July 6th, during a total eclipse (https://eclipse.gsfc.nasa.gov/5MCSEmap/2901-3000/2931-07-06.gif) when Halley is ~40° west of the sun. I just realized I have some other characters sailing across the pacific in that timeframe - maybe they'd take a detour to see it. The Lunar nearsiders might get a fun look on July 20th, during a partial lunar eclipse (https://eclipse.gsfc.nasa.gov/5MCLEmap/2901-3000/LE2931-07-20P.gif). There might be an interesting sight on December, 30th when an Annular Eclipse (https://eclipse.gsfc.nasa.gov/5MCSEmap/2901-3000/2931-12-30.gif) in southeast Asia occurs after Comet Halley has dipped below the ecliptic but is still only 3.508 AU from the sun (and ~60° west, and south as seen from Earth). A lot of coincidences with my characters and their story. I'm not sure how accurate these predictions are, but I think I'll run with them. :)

Hornblower
2017-Nov-16, 10:37 PM
snip--

Also, a general question about apparitions of Comet Halley (or comets in general). In the sim in the 2931 apparition, it passes through it's highest point north of the ecliptic, and between Earth and the Sun, at .44 AU closest approach on August 8th. Would it be visible in the daytime sky? --snip
No way. At similar geometry in prior returns it has been about 1st magnitude at best.

tony873004
2017-Nov-17, 04:14 PM
Do I need to propagate the sim to the date/epoch of the ephemeris before importing or does the code sort it out?

Adding objects using menu Objects > Create Objects, you have to propagate to the epoch of the elements you are using.
Adding objects using the Import interface: you don't have to propagate. JPL Horizons will do that for you.



But, if I decide to add something in, like more comets, the moons of Jupiter or the new Planet 9 they're looking for (if they find it), will I have to start all over? Or does JPL send ephemerides for that date?

If you add them using the Import interface, JPL will send you the ephemerides for that date. If you are going to go decades or centuries into the future, you might want to delete and re-import Jupiter before adding moons to it, just incase my integrator has Jupiter a few thousand km away from where JPL is expecting it.

tony873004
2017-Nov-17, 08:51 PM
If you are going to go decades or centuries into the future, you might want to delete and re-import Jupiter before adding moons to it, just incase my integrator has Jupiter a few thousand km away from where JPL is expecting it.
Most of the default simulations contain Jupiter Barycenter (5) rather than Jupiter (599). Jupiter barycenter contains the mass of Jupiter its moons, and their combined momentum. So if you want to add moons to Jupiter, you should delete Jupiter Barycenter and then import Jupiter (599).

Ara Pacis
2017-Nov-21, 04:31 AM
Thanks for the tips, Tony. I tested some imports in my future timeframe, and as I expected (from the JPL guide), they failed. They don't have ephemerides for planets and moons and other small objects beyond a certain horizon. So, I'll have to reconstruct at a near present epoch and then propagate from there. Though, the Moons of Mars and Jupiter move fast enough, I may not need that level of accuracy.

You probably know this, but in case anyone else comes across this thread:


3. OBJECT SELECTION
SELECTING MAJOR BODIES:
System barycenters are available over longer time-spans than planet-centers because planet-centers are
defined relative to the barycenter by satellite solutions. These satellite solutions are based on shorter data arcs than the
entire system and can therefore be extrapolated only over shorter time-spans. For example, the planet Jupiter (599)
might be available over the interval 1600-2500, while the Jupiter system barycenter (5) is available over 3000 B.C. to
A.D. 3000.
...
20. LONG-TERM EPHEMERIS
SOLAR SYSTEM MODEL:
...
Satellites and outer solar-system planet-centers each have various shorter intervals, as warranted by their
observational data arc. Comets and asteroids are available only over the A.D. 1599 to A.D. 2200 interval of the DE-
405 ephemeris they are integrated against. (Only a few dozen small-bodies have sufficiently well-known orbits to
justify rigorous integration over time-spans of hundreds of years.)

tony873004
2017-Nov-21, 04:48 AM
There's a reason JPL Horizons doesn't have them beyond a certain date. The solar system is chaotic. Small inaccuracies in starting conditions lead to large errors later, especially in along-track (where the object is in its orbit).

My simulator doesn't do well with the moons of the planets except Earth. The reason is my simulator considers every planet to be a sphere. In reality, they're oblate spheroids. So the closer a satellite is, or the more oblate a planet is, the greater the error.
Earth is pretty spherical, and the Moon is relatively far from Earth compared to Earth's size. So my program models it well. You can propagate 100+ years in either direction and view total solar eclipses +- 2 minutes from their actual timings.

Mars is pretty spherical, but its moons orbit very close. So don't trust it more than a few weeks. The size, shapes, and orientations of its moons will be accurate for a long time, but the position of its moons on their orbits will not.

The outer plants are quite oblate, and some of their moons orbit relatively close compared to their sizes. So again, past a few weeks, don't trust their positions of their moons on their orbits. JPL Horizons accounts for all this, and even they don't want to give a prediction past certain dates.

Ara Pacis
2017-Nov-21, 04:13 PM
How large are the along-track errors? Over days, weeks, etc, up to a 10 year timeframe? I figure that if JPL can't predict that far out, I'm safe with an arbitrary position for the moons of Mars and the gas giants for my story's accuracy. But how internally consistent will they be after that point? I don't yet know what level of precision I want or need. If the error is big enough, I guess I could use a near-present simulation as a stand-in, re-importing from JPL as needed. Unless there's a better solution.

I like your program. It's better than my previous workflow, which was to take screenshots of an orrery program, draw ellipses in Paint, then fit curves in another image manipulation program, hoping my along-track guesses fit. I have some more questions about programming in it, but I should probably put those in a new thread.

tony873004
2017-Nov-21, 10:02 PM
How large are the along-track errors? Over days, weeks, etc, up to a 10 year timeframe? I figure that if JPL can't predict that far out, I'm safe with an arbitrary position for the moons of Mars and the gas giants for my story's accuracy. But how internally consistent will they be after that point? I don't yet know what level of precision I want or need. If the error is big enough, I guess I could use a near-present simulation as a stand-in, re-importing from JPL as needed. Unless there's a better solution.

I like your program. It's better than my previous workflow, which was to take screenshots of an orrery program, draw ellipses in Paint, then fit curves in another image manipulation program, hoping my along-track guesses fit. I have some more questions about programming in it, but I should probably put those in a new thread.

JPL Horizons wont give any data on planets prior to 9999 BC or past 9999 AD. So it would be reasonable to assume that they trust the results in this range.

Here is a link to my program, started with vectors provided by JPL Horizons in 2017, and propagated to January 1, 9000.
http://orbitsimulator.com/gravitySimulatorCloud/simulations/1511294847455_ss9000.html

Try importing the planets from JPL by pressing "I" on your keyboard. Then import:
199,299,399,4, 5, 6, 7, 8, 9
Then press "k" on your keyboard to re-plot.

It plots the newly-imported planets are on the same pixels as the propagated planets. If you zoom in, you'll see that it actually differs by a few planet radii. So if there is any along-track error during this time, we are getting roughly the same error. Not surprising since I'm using their numbers.
(Don't press Play unless you want to watch the solar system fly apart. Each planet will have a twin within a few radii of itself).

The reason along-track is so sensitive to error is that it depends on period. And period depends on semi-major axis. And semi-major axes get perturbed by objects whose positions, velocities, and masses aren't precisely known. A small error in semi-major axis = a small error in period. Errors in period build orbit-after-error. And the error grows exponentially.

Jens
2017-Nov-21, 11:17 PM
JPL Horizons wont give any data on planets prior to 9999 BC or past 9999 AD. So it would be reasonable to assume that they trust the results in this range.


I would think the range is just a programming issue, and I don’t think it follows that they trust the simulations in that range. Nobody is going to plan a space mission 100 years in the future, and rockets can (and do) adjust their trajectories en route, so I don’t see why it would be important for long-term predictions to be very accurate.

tony873004
2017-Nov-22, 01:40 AM
I would think the range is just a programming issue, and I don’t think it follows that they trust the simulations in that range. Nobody is going to plan a space mission 100 years in the future, and rockets can (and do) adjust their trajectories en route, so I don’t see why it would be important for long-term predictions to be very accurate.
The quotes provided by Ara Pacis, especially
Satellites and outer solar-system planet-centers each have various shorter intervals, as warranted by their
observational data arc. seem to suggest that they are only willing to provide what they trust. Horizons is used for more than just space missions. A few years ago, the range for planets was smaller. Around the time they increased it, Nasa also extended the dates for which it was willing to make solar eclipse predictions to 3000 years into the future. In any case, I've never seen 9999 as any kind of a limit for a computer. They don't count in base 10 anyway. If perhaps they limited the range at 16383, or 65535, or some other power of 2, I might be convinced it was a programming limitation. But they could always go double precision. My program wouldn't have any trouble blasting past the year 9999. The question is whether or not I'd trust the propagated data past that range. For size, shape, and orientation of the orbit, I'd trust it much longer than that, but not for along-track.

Ara Pacis
2017-Nov-22, 04:54 AM
My favorite part of the JPL Horizons guide is this.
19: STATEMENT OF EPHEMERIS LIMITATIONS
After some disclaimers about observation arcs of various types and uncertainties, they close with this:

IF YOUR CAREER OR SPACECRAFT DEPENDS ON A NON-LUNAR NATURAL SATELLITE OR
SMALL-BODY EPHEMERIS, CONTACT JPL BEFORE USING IT. YOU >MUST< HAVE ADDITIONAL
INFORMATION TO CORRECTLY UNDERSTAND EPHEMERIS LIMITATIONS AND
UNCERTAINTIES.
(emphasis theirs)

In other words, Horizons is good enough for day to day earthworm needs, but for life and death plotting, let them wrangle that beast for ya.

The failure I mentioned in a previous comment was an attempt to import certain planets and moons in epoch 2930 instead of their barycenters - to visualize the Jovian moons, etc. without having to spend hours propagating from a more modern epoch. I received an error except for Mercury, Venus, and Earth. I think it was an error - nothing showed on the screen and checking the object list showed something with a lot of asterisks followed by some JPL notation. Since the request was outside the parameters I quoted in the JPL Horizons guide, I figured it was an error. I received slightly different errors when trying to import Comet Halley by name (in the present epoch), until Tony listed solution numbers that worked.

Tony, what was the time step you used for propagating from 2017 to 9999? How much error does one get using 2048? I think I saw a 2 day difference in a conjunction with the Moon and Comet Halley (as seen from Earth) on a second glance, when looking at my saved sim for something else, but I'm not sure. But I digress, back to your sim, ss9000, I got an error when importing Pluto, asterisks plus "REVISED" and a date. Also, I did some quick arithmetic and the difference in Mean Anomaly between Your Earth and JPL's Earth is ~4.12°, or a bit under 11 Million km - if I did that calculation right:
149,597,870.7 km *2 = 299,195,741.4 km
* pi = 939,951,143.2 km circumference
/360 = 2,610,975.4 km/1°
* 4.119571° = 10,756,098.1 km

Jens
2017-Nov-22, 06:55 AM
If perhaps they limited the range at 16383, or 65535, or some other power of 2, I might be convinced it was a programming limitation. But they could always go double precision. My program wouldn't have any trouble blasting past the year 9999. The question is whether or not I'd trust the propagated data past that range. For size, shape, and orientation of the orbit, I'd trust it much longer than that, but not for along-track.

I wasn't thinking so much that it was a limitation of the computer's calculating power, but rather something much simpler, that perhaps there are only four fields in the interface (input screen), and that you would have to add another field in the interface to allow you to go to five digits. And I work for a government organization like NASA and I can't begin to tell you how much of a nightmare it can be to do something like that. "Write me a proposal explaining why this is necessary," "Write me a detailed analysis of how much this will cost"... But I could be wrong. I don't know the specifics of the project.

Ara Pacis
2017-Nov-23, 04:23 PM
I think I saw a 2 day difference in a conjunction with the Moon and Comet Halley (as seen from Earth) on a second glance, when looking at my saved sim for something else, but I'm not sure.

Well, after double checking, I think it was a notation error on my part. A previous conjunction was on 9/6 and the next was on 10/4 not 10/6. I ran the sim again from 1995 to 2930 at 256 instead of 2048 and the results appear to be the same, so far as I can tell. It took ~48 instead of 4 hours, but I'm confident in the result. Not that I ever doubted you or your program, Tony. :D

Hornblower
2017-Nov-23, 05:43 PM
What I don't get is why would-be sci-fi writers are concerned about being off a bit on the date of a return a millenium from now. To me that is a trifle compared with taking real liberties with physics for the sake of a story.

Ara Pacis
2017-Nov-24, 04:03 AM
What I don't get is why would-be sci-fi writers are concerned about being off a bit on the date of a return a millenium from now. To me that is a trifle compared with taking real liberties with physics for the sake of a story.

Dunno, probably depends on the writer. My story is hard science. Since it's hard science and relatively non-advanced (no magical physics, most of the liberties I take are with economics and politics, allowing humans to travel between planets). So, the apparition of Comet Halley would be a big deal to my characters, who are students learning about space and interplanetary travel in that year. It's also important to some characters who have religious beliefs related to astronomical phenomena. Running the sim and discovering that Comet Halley appeared during my story was an amazing coincidence!

My main concern was that the current period of Comet Halley is around 75 years, but according to the sim, in the future it's reducing, and by 2931 it's at ~69 years. but looking at historic apparitions, it had been 77-79 years. So, perhaps it's just part of a general trend. When I ran a sim back several millennia, it was around 90 years.

It's just part of my general goal of precision and accuracy in this story. Since it's hard physics, I'm also trying to get the orbital mechanics correct for planetary rendezvous. A one off trip between planets might be written simply without explanation. But I'm building a cyclic transportation network with many characters on different trips, modeled after a transportation schema I think is possible with currently plausible engineering. This means looking at how different synodic planetary alignments line up in relation to other synodic planetary alignments. I'll have to analyze realistic interplanetary trajectories no matter which epoch I use, so I may as well do it in the story timeframe. It's not much more work, but it is more satisfying - to me at least, and perhaps future readers will appreciate it.