科研之友
微信
新浪微博
Facebook
分享链接
摘要
With the advent of wide-field imagers, it has become possible to conduct a photometric lightcurve survey of many asteroids simultaneously, either for that single purpose (e.g., Dermawan, B., Nakamura, T., Yoshida, F. [2011]. Publ. Astron. Soc. Japan 63, S555-S576; Masiero, J., Jedicke, R., Durech, J., Gwyn, S., Denneau, L, Larsen, J. [2009]. Icarus 204, 145-171), or as a part of a multipurpose survey (e.g., Pan-STARRS, LSST). Such surveys promise to yield photometric data for many thousands of asteroids, but these data sets will be "sparse" compared to most of those taken in a "targeted" mode directed to one asteroid at a time.
We consider the potential limitations of sparse data sets using different sampling rates with respect to specific research questions that might be addressed with lightcurve data. For our study we created synthetic sparse data sets similar to those from wide-field surveys by generating more than 380,000 individual lightcurves that were combined into more than 47,000 composite lightcurves. The variables in generating the data included the number of observations per night, number of nights, noise, and the intervals between observations and nights, in addition to periods ranging from 0.1 to 400 h and amplitudes ranging from 0.1 to 2.0 mag.
A Fourier analysis pipeline was used to find the period for each composite lightcurve and then review the derived period and period spectrum to gauge how well an automated analysis of sparse data sets would perform in finding the true period. For this part of the analysis, a normally distributed noise level of 0.03 mag was added to the data, regardless of amplitude, thus simulating a relatively high SNR for the observations. For the second part of the analysis, a smaller set of composite curves was generated with fixed core parameters of eight observations per night, 8 nights within a 14-day span, periods ranging from 2 to 6 h, and an amplitude of either 0.3 mag or 0.4 mag. Individual data sets using these fixed parameters added normally-distributed noise of 0.05, 0.1, or 0.2 mag. The analysis examined the success rates for finding the true period as the noise increased towards levels simulating data for objects close to sky background levels.
After applying a filter to remove highly-ambiguous solution sets, the best chance for success was found to be when the true period was in the range of P approximate to 2-5 h and amplitudes were A >= 0.5 mag. The solution sets for lightcurves with low amplitude, long periods, and/or those that were sampled too sparsely in comparison to the period were often too ambiguous to be considered reliable for statistical rotation studies. Analysis of slow rotators (P > 24 h) found that somewhat reasonable solutions of P < 6h could be found for about 15-20% of those objects, even at higher amplitudes, indicating that the Fourier analysis had locked onto the noise in the data.
Efforts to produce an automated pipeline to help determine an unambiguous (or nearly so) solution based on the period spectrum from the Fourier analysis were made. These proved unsuccessful because of the number of parameters that must be considered and the difficulties in assigning an objective weight to each one in finding a final result. Despite this initial failure, further attempts will be made to quantify the U rating system.
Comparison of the synthetic data analysis results to those from two actual surveys shows a reasonable agreement between the two. A review of the pros and cons of sparse versus dense data sets shows that each has a significant role in future studies and that it will be critical to establish open lines of communications and data exchange between the deep wide-field sparse data surveys and dense data programs.
- 出版日期2011-12
