During the observations, both Jupiter and Earth have orbital motion.
The orbital motion not only changes the distance between the planets,
but also the angle of viewing. As the angle between Earth and Jupiter
changes the observer sees a different orientation of the orbit of the
moon of Jupiter. At the beginning of the observations, for a moon at its
maximum Eastward extent, let the position of the moon be 0°. The maximum
Westward extent would then be 180°.
As the observations continue the angle between Earth and Jupiter
changes by some amount, say 16°. Thus at the end of the observations the
maximum extents are now 16° and 196° (each extent progressed by 16°).
If an orbit period was equal to the observation time, then the moon
would have begun at 0° and ended at 196° (more than 180°). If the orbit
period were half the observation period, then the moon would have begun
at 0° and ended at 188°. Thus while each orbit appears to have 180°
between maximum extents, the true travel angle slightly different. The
observed data is modified to account for this effect by the EXCEL
spreadsheet.
For Project Jupiter the best estimate of the orbital period, and
other parameters, is determined mathematically from the observation
data. Mathematically there are several measures that may be used to
judge when the observed data is best reproduced by the computer. Prior
tests have shown that the true elliptical orbit variation from an
assumed circular orbit need not be considered for Project Jupiter.
Four measures of the Goodness of Fit used in the computer model are:
1. Standard Deviation
One measure of the fidelity of the data ( how well it is trending) is
the statistical measure called the Standard Deviation. The larger the
Standard Deviation, the larger is the variation in the data being
considered.
When the orbital period is being estimated, the observer’s moon
positions are subtracted from the position computed by the EXCEL
program. As the EXCEL program inputs begin to match the observer’s data
the Standard Deviation value begins to drop.
The value does not get to zero because of observational biases,
errors, transcription mistakes, difficulty in estimating small
separations, etc. The EXCEL program computes the Standard Deviation, so
the Quad-A observer need not be concerned with the mathematics.
The concept here is that the statistical measure called the Standard
Deviation is used as an indicator of when the orbital inputs best match
the observed data. When a good match is found the Standard Deviation
value reaches a minimum.
2. Correlation Coefficient
The Correlation Coefficient is a statistical comparison of two sets
of data. The Correlation Coefficient ranges from -1 to +1, with ±1
representing the strongest (best) similarity measure ("correlation").
The observer’s Project Jupiter separation data is compared with EXCEL
generated data for a sinusoidal curve. When a computed "correlation
coefficient" approaches 1.0 its is an indication that the computed
separations are strongly matching the observed separations.
One of the EXCEL inputs for the sinusoidal formula is the orbital
period, so a high correlation coefficient is one indicator that the
orbital period input is the best estimate possible. As the Correlation
Coefficient value approaches 1.0, it is also confirmation that the orbit
is not strongly elliptical, validating an earlier assumption.
3. Least Squares
When two data sets are compared, one measure of their similarity is
to examine the differences in the individual data values. By taking each
value and multiplying it by itself ("squaring it") negative values do
not cancel positive values, and the check becomes a sensitive check of
the similarity of data sets. When the sum of the squared values is a
minimum, the best agreement between the data sets has been determined.
The EXCEL generated data is automatically subtracted from the
observer’s data, squared, and then summed. The program automatically
displays the result so that the user can know when the minium value has
been found. This provides added assurance that the best determination of
the orbital period, based on the transmitted Quad-A observer’s data, has
been found.
4. Residuals
Residuals are the differences between what was observed and the
mathematical representation of the process observed. Usually the
residuals themselves do not have a pattern to then, they are randomly
distributed. The Project Jupiter analysis makes reasonable graphical
efforts to ensure that no residual biases impact the determined orbital
period.