Eliminated Bug in Galilean Orbits=> Spectacular Accuracy!

[t00fri]

Chris:

After systematically comparing the predictions for the Galilean moons
from your checkin yesterday, with Jean Meeus mutual event listing in
Dec. 2002 S&T, p. 100 ff, it was soon clear to me that something in
the 'customorbit.cpp' code for the /latitude/ calculation of the four
Galilean Jupiter moons was fishy. Soon I found it:

// Calculate periodic terms for the tangent of the latitude
....
B = degToRad(B);
B = atan(B);

Indeed it's tan(latitude) that is computed in ... and hence the
statement B = degToRad(B); makes no sense and is actually incorrect.

After commenting B = degToRad(B); out for all four Galilean moons, the
resulting predictions turn out to be /spectacularly accurate/ without
any fiddeling whatsoever!! Everything with your/Grant's parameters.

I attach just two amazing examples, but I have explicitly and
successfully tested most of J. Meeus event listings!:

1) The /annular/ eclipse shadow of Europa on Io (2e1A) on Dec. 27



2) The /total/ occultation of Europa by Callisto (4o2T) on Dec. 15


Neither of these examples works (correctly) with your CVS code as of yesterday.

Bye Fridger

[chris]

Chris:

After systematically comparing the predictions for the Galilean moons
from your checkin yesterday, with Jean Meeus mutual event listing in
Dec. 2002 S&T, p. 100 ff, it was soon clear to me that something in
the 'customorbit.cpp' code for the /latitude/ calculation of the four
Galilean Jupiter moons was fishy. Soon I found it:

// Calculate periodic terms for the tangent of the latitude
....
B = degToRad(B);
B = atan(B);

Indeed it's tan(latitude) that is computed in ... and hence the
statement B = degToRad(B); makes no sense and is actually incorrect.

Brilliant! I'm so glad you found this bug . . . I'd noticed that the latitude was off, but could not figure out the source of the error. I was looking at how to fix the reference plane, and didn't even think to look for a bug in the orbit computations themselves. Check this fix in immediately!



--Chris

[t00fri]

Chris:

After systematically comparing the predictions for the Galilean moons
from your checkin yesterday, with Jean Meeus mutual event listing in
Dec. 2002 S&T, p. 100 ff, it was soon clear to me that something in
the 'customorbit.cpp' code for the /latitude/ calculation of the four
Galilean Jupiter moons was fishy. Soon I found it:

// Calculate periodic terms for the tangent of the latitude
....
B = degToRad(B);
B = atan(B);

Indeed it's tan(latitude) that is computed in ... and hence the
statement B = degToRad(B); makes no sense and is actually incorrect.

Brilliant! I'm so glad you found this bug . . . I'd noticed that the latitude was off, but could not figure out the source of the error. I was looking at how to fix the reference plane, and didn't even think to look for a bug in the orbit computations themselves. Check this fix in immediately!

:D :D :D

--Chris

Sure, I will;-)

I am also very happy, after all this struggle, the accuracy is simply breathtaking!

Bye Fridger

[granthutchison]

This is great news, indeed.
On a similar topic, Chris, how are the mutual events for Saturn's moons looking now that you've switched from B1950 to J2000?

Grant

[t00fri]

This is great news, indeed.
On a similar topic, Chris, how are the mutual events for Saturn's moons looking now that you've switched from B1950 to J2000?

Grant

Chris wrote to me already yesterday that he was heading straight into mutual events for Saturn's moons;-).

Is there a reliable compilation, too? Where?

Bye Fridger

[Guest]

It seems that many of the interesting (i.e. precise) observations of mutual events during the 1995-1996 ring crossings may be in professional journals and not directly findable on the Web. They should be available if you have access to a library with online or hardcopy subscriptions. (I have not yet taken the time to search Cornell's holdings.)

However, I did find a few public images:

http://www.theastronomer.org/planets.html has several in its section on Saturn.
For example, http://www.theastronomer.org/images/saturn_950716.jpg was taken shortly after an "appulse" of Thetys and Dione, and is timed to the second.

Despite thw wording (it refers to the events as being in the future) JPL's Saturn site has links to several pages of images taken during the ring crossings. See http://www.jpl.nasa.gov/saturn/

For example, a composite Hubble picture shows several moons with precise times at http://www.jpl.nasa.gov/saturn/hst13.html

I hope this helps a little.

[selden]

sigh. that was me.

[chris]

This page has a list of mutual events of the Saturnian satellites:
http://www.bdl.fr/ephem/ephesat/phenomenes/Saturne/phenSaturnMut1995_eng.html

I haven't had a chance to check these events in Celestia, however.

--Chris

[Guest]

Here's a very detailed list of Saturnian satellite mutual events between 1994 and 1997, from the Institut de M?canique C?leste at the Bureau des Longitudes in Paris:
http://www.bdl.fr/ephem/ephesat/phenomenes/Saturne/phenSaturnMut1995_eng.html

Grant

[Guest]

And that was me, coming in second to Chris, and unable to delete my redundant posting because I forgot to sign in ... Sigh.

Grant

[granthutchison]

And that was me, coming in second to Chris, and unable to delete my redundant posting because I forgot to sign in ... Sigh.

Sigh again. I may make a hash of logging in, but at least the link I posted does work!

Grant

[t00fri]

It seems that many of the interesting (i.e. precise) observations of mutual events during the 1995-1996 ring crossings may be in professional journals and not directly findable on the Web. They should be available if you have access to a library with online or hardcopy subscriptions. (I have not yet taken the time to search Cornell's holdings.)

However, I did find a few public images:

http://www.theastronomer.org/planets.html has several in its section on Saturn.
For example, http://www.theastronomer.org/images/saturn_950716.jpg was taken shortly after an "appulse" of Thetys and Dione, and is timed to the second.

Despite thw wording (it refers to the events as being in the future) JPL's Saturn site has links to several pages of images taken during the ring crossings. See http://www.jpl.nasa.gov/saturn/

For example, a composite Hubble picture shows several moons with precise times at http://www.jpl.nasa.gov/saturn/hst13.html

I hope this helps a little.

Thanks Selden, that was indeed very helpful! Here comes another "success story" displayed in the composite image below. The timing takes correctly into account the light travelling time to Saturn of 83.04'. In order to make the relevant moons more conspicuous, I deliberately increased their radius by a factor of 10 (Tethys, Dione, Rhea, Enceladus) and 100 ( Pandora, Janus), respectively. Below, you see the comparison of the Celestia prediction with the Hubble photo provided by Selden.

All the bigger moons are /perfect/, including the eclipse shadow on Saturn (bottom image). The only "bad guy" is Janus, a tiny piece of rock!

Note, the events are 7 3/4 years ago!!

Bye Fridger

[t00fri]

OK, here is the solution for the above problem with recently
discovered minor moon 'Janus' (and to some minor extent
'Pandora'). They are listed in the file extras/minormoons.ssc
that starts with the remark:
-----------------------------
# Note: The mean anomalies listed for the orbits are pure fiction--I just
# wanted something other than the default value of zero to avoid a weird
# conjunction at epoch.
-----------------------------

Hence after adaptation of Janus' MeanAnomaly: 136 => 259
and a small correction of 'Pandora's' MeanAnomaly: 308 => 315,

everything falls beautifully in place, as the composite image below
demonstrates!

Note that now I have used the correct radii of the moons, whence the
eclipse shadow in the bottom image at left is tiny exactly like in the
HST photo on the rhs...

Bye Fridger

[granthutchison]

Fridger:
I don't think Saturn is "perfect" yet - certainly nowhere as good as your fine result with Jupiter.
The position of Rhea at 08:02 22 May 1995 is noticeably farther from Saturn's disc in your Celestia image than in the real image. Also, the ring-plane in Celestia seems to be too open to be within just a day of the ring-plane crossing on 21 May.
What EquatorAscendingNode are you using for Saturn? I calculate 169.53, which not only has the rings pretty much edge-on on 22 May, it has Rhea in a position that better matches reality.
Unfortunately I can't go much further checking the eclipse and occultation events from the Bureau des Longitudes webpage, because in 1.2.5pre7 there is an anomaly in the plane of the satellite orbits, which Chris tells me is because they were calculated in B1950, not J2000. But I'm hoping that has been fixed in the updated code you have access to.

Grant

[chris]

Fridger:
I don't think Saturn is "perfect" yet - certainly nowhere as good as your fine result with Jupiter.
The position of Rhea at 08:02 22 May 1995 is noticeably farther from Saturn's disc in your Celestia image than in the real image. Also, the ring-plane in Celestia seems to be too open to be within just a day of the ring-plane crossing on 21 May.
What EquatorAscendingNode are you using for Saturn? I calculate 169.53, which not only has the rings pretty much edge-on on 22 May, it has Rhea in a position that better matches reality.
Unfortunately I can't go much further checking the eclipse and occultation events from the Bureau des Longitudes webpage, because in 1.2.5pre7 there is an anomaly in the plane of the satellite orbits, which Chris tells me is because they were calculated in B1950, not J2000. But I'm hoping that has been fixed in the updated code you have access to.

Grant

I know that I haven't done anything to fix the B1950/J2000 problem yet. Some adjustments will need to be made; the custom orbits for the Saturnian moons use the constant 168.8112, which I believe is the equatorial ascending node of Saturn on the B1950 ecliptic.

--Chris

[t00fri]

Fridger:
I don't think Saturn is "perfect" yet - certainly nowhere as good as your fine result with Jupiter.
The position of Rhea at 08:02 22 May 1995 is noticeably farther from Saturn's disc in your Celestia image than in the real image. Also, the ring-plane in Celestia seems to be too open to be within just a day of the ring-plane crossing on 21 May.
What EquatorAscendingNode are you using for Saturn? I calculate 169.53, which not only has the rings pretty much edge-on on 22 May, it has Rhea in a position that better matches reality.
Unfortunately I can't go much further checking the eclipse and occultation events from the Bureau des Longitudes webpage, because in 1.2.5pre7 there is an anomaly in the plane of the satellite orbits, which Chris tells me is because they were calculated in B1950, not J2000. But I'm hoping that has been fixed in the updated code you have access to.

Grant

Grant:

I definitely agree with you once we apply a very quantitative attitude. I was merely happy that things looked very encouraging in a first go;-).

Indeed, as Chris explained above , in the present CVS solarsys.ssc he put EquatorAscendingNode 168.8112
rather than your value 169.53.

Despite the B1950/J2000 issue, I agree that with 169.53 the major moons, notably Rhea are really perfect and Saturn's ring is virtually invisible.

Bye Fridger

[granthutchison]

Some adjustments will need to be made; the custom orbits for the Saturnian moons use the constant 168.8112, which I believe is the equatorial ascending node of Saturn on the B1950 ecliptic.

That looks like it might be the EquatorAscendingNode in Fridger's views - it fits with the very slight anomalous openness of the rings.

Grant

Update: By the time I posted this, Fridger's confirmation above had come in!

[t00fri]

Some adjustments will need to be made; the custom orbits for the Saturnian moons use the constant 168.8112, which I believe is the equatorial ascending node of Saturn on the B1950 ecliptic.

That looks like it might be the EquatorAscendingNode in Fridger's views - it fits with the very slight anomalous openness of the rings.

Grant

Update: By the time I posted this, Fridger's confirmation above had come in!

To make things more precise, I overlayed the HST and Celestia images in GIMP, after scaling them precisely to identical sizes, making one semi-transparent and inverting the HST image for better viewing.

The result is that in the three comparisons, Tethis, Dione and Enceladus are sitting /spot on/!
Rhea is good in the middle HST image, but /still/ slightly too far away from Saturn by one spot diameter ("1 sigma") in the image above, all calculated with EquatorAscendingNode 169.53.

Janus and Pandora still need a little tuning of their MeanAnomaly in order to get them perfect.

Bye Fridger

[ou8poop2]

...while the rest of us sit back and wait for these updates to show up in the next release...
j/k guys, you're doing great! but as much as we all would like THE most realistic viewing possible, i prefer to wait until the bugs are worked out. my solar.sys looks like swiss cheese!

[selden]

Apparently the orbits of Pandora and Prometheus are chaotic. You can get them right at one epoch, but they'll be wrong at other times. For example, they both are about 20 degrees away in opposite directions from where they would be when using the orbital parameters determined from the Voyager flybys to predict their current locations. This seems to be due to their gravitational interactions in their close but eccentric orbits.

See http://www.jpl.nasa.gov/releases/2002/release_2002_190.cfm