PDA

View Full Version : An Ugly Little Tale of Falsified Asteroid Data and Software Bugs



Roger E. Moore
2019-Jan-03, 02:26 PM
Grit your teeth and read on.


https://arxiv.org/abs/1812.06516

Response to Wright et al. 2018: Even more serious problems with NEOWISE

Nathan P. Myhrvold (Submitted on 16 Dec 2018)

Wright et al. (2018, hereafter W2018) respond to some of the irregularities and criticisms of the NEOWISE results raised by Myhrvold (2018a, 2018b). Contradicting earlier statements by the group, the response now acknowledges the validity of some of the most serious issues identified by Myhrvold. Among these is the false presentation of diameters that had been copied from previously published analyses of radar, occultation, and spacecraft observations as diameters obtained by thermal modeling. While W2018 confirms that this misrepresentation occurred for 117 asteroids, it fails to provide the actual modeled diameters or to identify the asteroids for which copied diameters were substituted. W2018 confirms an error in the NEOWISE analysis that contributed to the issue, first identified by Myhrvold (2018b, hereafter M2018b), that many fit curves published by NEOWISE do not pass near the data points they claimed to fit. W2018 now documents a software bug that was found and corrected in 2011 but that the team failed to disclose until now. This bug apparently corrupted the vast majority of NEOWISE results, which have yet to be corrected. Results affected by the bug were knowingly included, without comment, in a compilation of NEOWISE results archived in 2016 to the Planetary Data System (PDS). Although W2018 argues that the bug had only rare and small impacts on results, W2018 elsewhere provides examples that contradict this claim, thus rebutting their own false claim.

Other arguments offered in W2018 to rebut claims of M2018b are shown here to be invalid and to rely on fundamental misconceptions, such as that the NEATM predicts visible absolute magnitude H. We learn from W2018 that H values used by NEOWISE were not taken from the literature, as previously claimed, but were in fact derived by fitting data. The combination of acknowledged errors, misrepresentations, and misconceptions presented in W2018 further undermines confidence in the NEOWISE project and demonstrates the urgent need for an independent effort to produce a full and correct error analysis of the NEOWISE results that have been published in the literature and the PDS to date.