Remarks on generating realistic synthetic meteoroid orbits

TL;DR

This paper compares five methods for generating synthetic meteoroid orbits, evaluating their realism and quality using various statistical tests and proposing new metrics to assess the similarity to observed data.

Contribution

It introduces two new methods for synthetic orbit generation and provides a comprehensive comparison framework for assessing their realism against observed meteoroid orbits.

Findings

Different tests yield varying assessments of orbit realism.

Most methods produce similar results when applied appropriately.

Assessing orbit quality depends on the test and observed sample used.

Abstract

Context. To identify the real associations of small bodies, we can use synthetic sets of orbits generated by various methods. These are not perfect methods, therefore the assessment of their quality is an essential task. Aims. In this study, we compared five methods for generating synthetic meteoroid orbits. Three of them (ME0, KD10, and KDns) had already been proposed in the literature, while two additional ones (ME1 and ME4) are new methods. Methods. As far as possible, the synthetic orbits were compared with the orbits of the observed meteoroids. For quantitative comparison, we applied a few tests: the chi-square distance and the nearest neighbor NN_N tests used in previous works, and one-dimensional chi-square and Kolmogorov-Smirnov (K-S) tests, as well as a two-dimensional K-S test implemented in this study. To estimate a general property of the orbital sample, we proposed the use of the entropy H_N of the data set based on the nearest neighbor distances. Finally, we did a cluster analysis of the synthetic orbits. We calculated and compared the values of the orbital similarity thresholds. Results. We showed that generating 'realistic' meteoroid orbits and testing their quality is a complex issue. An assessment of the quality of the generated orbits depends on the type of test applied, and it refers to the sample of the observed orbits used. Different tests give different assessments. However, in practice, the investigated methods produced similar results if they were applied correspondingly.