[This post is part of a series of posts on archaeoastronomy using open source software]
My work focusses on using Google Earth, a downloadable free virtual globe and geographic information program, to obtain satellite imagery of aerially-visible archaeological sites world wide that are suspected to have been used for astronomical observations, for instance stone circles and medicine wheels. The uniqueness of my approach is that it does not require actually having to visit these sites in person to survey them.
In addition, the free-source nature of the programs I use means that anyone with a modicum of expertise and a computer can duplicate the results. Previous to this, double checking the claims of astronomical alignments of a site required actually having to visit it, and having the time, funds, and expertise needed to conduct a comprehensive survey… not to mention also having the time and expertise needed to do the complicated calculations needed to estimate the horizon rise/set positions of various celestial bodies at any arbitrary date. The free software I use to do my studies makes all of this much, much easier (but perhaps not necessarily something a complete novice would attempt without some basic knowledge of astronomy and a reasonable background in computing and statistics). For anyone wishing to try out my methods, I will provide all the code and files related to an analysis I’ve done of the “Merry Maidens” stone circle in the UK.
Using Google Earth, I take screen shots of aerial views of the site, and using the free and open-source vector graphics program Xfig, I then upload the screenshot and place datum points at all intersecting walls, posts, standing stones, other significant site features, etc. I output these points to a data file, then use the R free and open-source statistical programming language to fit straight lines to all possible combinations of points (usually with some quality criteria, such as a requirement that the points be far enough apart that the uncertainty on which angle of the line is sufficiently small to give a reasonable indication of where on the horizon it is pointing). I call these lines the “site lines” (they also may potentially be “sight lines” to celestial body rise/set points on the horizon).
To determine the rise and set azimuths (angle from North) of celestial objects for any given date, I use the free downloadable pyephem ephemeris calculation package, written in the Python programming language. I consider celestial objects like the Sun, the Moon, and the brightest stars. The true rise and set azimuths must take into account the horizon of the surrounding terrain, thus I calculate this using topographic information for the area in 1′ grids that are publicly available from the National Geophysical Data Center at the National Oceanic and Atmospheric Administration website. I call these the “proposed astronomical alignments”.
Once I have the proposed astronomical alignments and site lines, I use statistical methods to determine what fraction of astronomical alignments match site lines, and what fraction of site lines match proposed astronomical alignments. One hallmark of a site that truly was used as a comprehensive astronomical observatory, is that both of these fractions are high. Or that there were no alignments to either the rise or set of some stars, but an unusually large number of alignments to both the rise and set of others. I assess the statistical significance of the observed alignments by creating many “synthetic” sites (by randomly scrambling the points on the site) and “synthetic” skies by randomly picking points on the horizon where some hypothetical celestial body might rise. Using this synthetic data, I can then obtain a probability distribution for the matches of site lines to astronomical alignments, and the probability distribution for the matches of astronomical alignments to site lines; these probability distributions test the null hypothesis that the site does not contain any astronomical alignments.