Posted by Steve Holmes2 at 23:35 on 2010 Nov 12
I suspect that the accuracy of "Jupiter’s moons" predictions is mainly down to the accuracy of the programming platform used to produce them rather than any basic differences in the theory used. All "non-commercial" predictor programs seem to use the same set of equations and methods originally devised by Jean Meeus and so, all other things being equal, should give the same answers. The fact that they don’t (to some small degree) may be down to the differences between the maths libraries for Java, JavaScript, C+, QBASIC etc. The orbital parameters of the moons themselves will probably not change greatly, not least because their periods are locked together by gravitational resonances.Additional factors influencing the relative accuracy of such things as the timing of transits would be the precision to which the answer is specified (a difference of just 1sec could change a "rounded" time by a whole minute); whether the the latitude and longitude of the observer are taken into account; the model used to represent the shape of the globe of Jupiter; and whether the timings were relative to the leading edge, centre or trailing edge of the moon: this alone can make a difference of +/-1 minute.As to which source is the most accurate, this can of course only be determined by comparison with "real life"! However, I would certainly expect the predictions to be accurate to +/- 1 minute. On the other hand, one is not going to begin observing precisely at the moment an event is predicted to start, and so the fact that different sources differ by a few minutes is surely not of any practical consequence. For truly accurate data one must consult sources deriving from internationally acknowledged organisations. For example, the BAA Handbook takes its timings from the French Institut de Mechanique Celeste et de Calcul des Ephemerides. These organisations use much higher precision calculations than "Meeus-type" programs but often supply their data only in tabular format, not usually online, and not interactively.There are many places on the Internet where predictions for the Galileans can be obtained (for example the ‘Sky & Telescope’ website) but the reason that sites giving ephemeris information for comets, asteroids etc. are perhaps easier to find is surely that the orbits for such bodies are not always well-known, being liable to change due to perturbations and as new observations come in. Reliable long-term models for them are thus not possible.I myself wanted to make some observations of shadow-transit events at the recent close opposition, so I also had to rely on timing predictions. My usual source is the (rather old!) program "The Planets" but I also use a number of other such programs as well as Starry Night. The predictions made by these programs were consistent to about +/- 1 minute but, as stated in the post, not all programs identified the occultation and eclipse events correctly – this is presumably down to the skill of the programmer though, rather than any output accuracy issues.Before composing this reply I investigated other programs on the Internet and found a couple that seemed quite good. "Galileo" and "Jupiter2" cover basically the same ground as each other, though each has its own "unique selling points". Their predictions are identical to each other (bearing in mind that Galileo does not use seconds whereas Jupiter2 does), and when I checked several transit events from the BAA Handbook for 2011 their predictions were exactly as listed there. They cannot, however, produce lists of satellite events (though Galileo will find the next one) so for that I return to "The Planets" – this will list not only all events for a given period but also only those actually visible from a given location.The Planets may be found at http://www.cpac.org.uk/download.aspGalileo may be found at http://astrosurf.com/durey/galileo.htmlJupiter2 may be found at http://www.astrosurf.com/rondi/programmes/index.htmHope this is helpful! Steve Holmes Laxfield, Suffolk UK