Petro Janse van Rensburg

Near-Earth Asteroids (NEAs) are asteroids in stable orbits around the Sun that bring them close to or cross the Earth’s orbit and therefore some have impacting trajectories and could pose a threat. Small NEAs (diameter between 30 m and 300 m) pose a large threat because they are big enough to cause significant damage on impact but also more abundant compared to larger NEAs (diameter > 300 m), which means the impact scenario is more likely. An example of a recent small NEA impact is the Chelyabinsk meteor of 2013; a 20 m asteroid that caused significant damage to buildings when it exploded in the atmosphere (Brown et al., 2013). However, the small NEA population has not been well studied because they can only be observed with ~1m-class telescopes when they pass close to the Earth and become bright enough.

For this project, 14 small and 6 large NEAs were observed and characterised with the South African Astronomical Observatory (SAAO) 40-inch telescope and the Sutherland High-Speed Optical Camera (SHOC; Coppejans et al., 2013). Most observations were performed remotely from the SAAO facilities in Cape Town. NEAs have high sky rates, therefore we used the non-sidereal tracking mode of the telescope to keep the target in the small field-of-view of SHOC during the observation period. If the non-sidereal tracking was successful, the target remained at the centre, with the field stars moving in and out of the field-of-view. The targets were observed in three filters: SDSS r′, g′ and i′.

Calibrated, multi-band photometry was extracted by making use of PHOTOMETRYPIPELINE, developed by Michael Mommert (2017) and plotted as lightcurves (magnitude vs time). A densely sampled lightcurve was formed by shifting the g′ and i′ data with the colours g′-r′ and r′-i′, respectively, to normalise them to the r′ data. A Lomb-Scargle (LS) periodogram (LS power vs period) of each lightcurve was generated to extract the rotation period of the asteroid. The rotation period is the time it takes for an asteroid to rotate once on its axis and is twice the lightcurve period. It was extracted from the period at the highest peak in the periodogram. A phased lightcurve (magnitude vs phase) was formed by folding the data at the rotation period so that the asteroid rotation was clearly visible in the lightcurve with troughs and peaks.

Asteroids are divided into different taxonomic types in the Bus-DeMeo taxonomic scheme (DeMeo et al., 2009) based on their spectra and therefore composition. S-, Q- and V-type asteroids are “stony” and mostly composed of silicates, C-type asteroids are mostly carbon-based, X-type asteroids have a whole range of compositions and D-type asteroids are possibly made of silicates, carbon and water ice. But photometry is easier to perform on faint, moving targets, and since each taxonomic type has a different spectral slope in the visible wavelength range, we can differentiate between the different types by using the colours g′-r′ and r′-i′ instead. A machine learning algorithm was trained on synthetic colours from observed spectra available at http://smass.mit.edu/minus.html and classified by S. Navarro-Meza to generate decision boundaries separating the S-, C-, X-, Q-, D- and V-type asteroids. The machine learning algorithm was trained using the k-nearest-neighbour method with k=5. The observed colours of the targets were used with the machine learning algorithm to determine the taxonomic class of the asteroid and thereby infer its most probable composition. Future work includes combining the results with other NEA surveys, thereby providing valuable insight into the most likely properties of potential future impactors.