Magnetic signatures of lunar impact craters

The hypothesis that the lunar core dynamo once generated a global magnetic field is widely accepted [1–2]. Paleomagnetic analysis of lunar samples shows that the lunar dynamo field operated from about 4.2 Ga to sometime between 1.92 and 0.8 Ga [3–5]. From the orbital observations, the strongest lunar magnetic crustal anomalies are found to concentrate on the lunar farside, and a prominent magnetic low correlates with the nearside Procellarum KREEP Terrane [6].

Impact cratering is a geological process that could either magnetize or demagnetize the lunar crust. Previous studies have investigated the magnetic signatures of impact basins hundreds of kilometers in size [7–9]. Among the largest multi-ring basins, only five Nectarian basins are unambiguously associated with central magnetic anomalies. For the smaller craters, previous studies have detected both magnetized [10] and demagnetized [11] signatures.

This study systematically analyzed the magnetic signatures of all lunar impact craters resolvable by the most recent magnetic field models. We used the locations and crater diameters from [12] which were based on optical images, and this database was supplemented by the peak-ring and multi-ring basins from [13] that were characterized using gravity data. We assigned the crater age using the updated version of the crater database of [14]. The surface magnetic field data from [15] was used as the primary model, and the model of [16] was then used to confirm the magnetic signature of the investigated crater.

The investigated craters in this study were classified into three classes: Magnetized (with central magnetic highs), demagnetized (with the magnetic low interior of the crater rim), and no signal (with no clear magnetic signal). The signal fidelity levels of the magnetized and demagnetized craters were further divided into three levels: Certain, probable, and possible. The two authors conducted the classification independently. To assess the likelihood of correct identification, we made use of two sets of synthetic magnetic field models generated by rotating the geographic coordinate system of the observed fields. We found more false identifications of magnetized craters than real observations when the crater diameter is smaller than 90 km. Therefore, we only report the results from the craters with diameters greater than 90 km.

In total, we analyzed 447 craters. Of these, only 26 and 42 were classified as magnetized and demagnetized craters by at least one of the analysts, respectively. If only considering the craters that both analysts agree upon, the numbers decrease to 10 and 30, respectively.

The debiased number of craters for a given classification was defined as the number of craters using the observed magnetic field models minus the number of craters using the synthetic maps in the given diameter or age class. For a few cases where the number of false identifications is larger than the number of using real detections, we simply set the debiased number as zero. The debiased percentage is simply the sum of the debiased numbers divided by the total number of craters in the given class.

In Figure 1, we plot the average debiased percentages of craters with impact-related magnetized and demagnetized signatures as a function of diameter and age. When only considering the certain and probable fidelity levels, about 1%, 3%, and 14% of the complex craters (90–206 km), peak-ring basins (206–582 km), and multi-ring basins (582–1321 km) are found to have magnetized and demagnetized signatures, respectively. When considering all three signal fidelity levels, the results show the same trend of increasing with increasing diameter but show a higher percentage.

In terms of crater age, when only considering the highest two fidelity levels, we find that about 2% and 3% of craters with pre-Nectarian and Nectarian ages show magnetized signatures, respectively. None of the younger Imbrian, Eratosthenian, or Copernican periods show evidence of craters with magnetized signatures. For the demagnetized class, we find that about 0.3%, 3%, and 16% of pre-Nectarian, Nectarian, and Imbrian aged craters show demagnetization signatures, respectively. These results are compatible with a lunar dynamo operating during at least portions of the pre-Nectarian and Nectarian periods, and then either weakening or ceasing at the beginning of the Imbrian period.

Fig 1. Average debiased percentages of the two analysts for craters with magnetized and demagnetized signatures as a function of (a) diameter and (b) age using the surface magnetic field models. The number above each bar shows the average number of craters in the interval from the two analysts after debiasing.

Lastly, we placed constraints on the mechanisms of generating impact-related magnetic signatures. The excavation of crustal materials and thermal demagnetization can account for the magnetic lows within the crater rim. Shock demagnetization might account for the magnetic lows that extend beyond the crater if an ambient core-generated field was absent when the crater formed. In contrast, the magnetized signatures are likely to be the result of the heated materials cooling in the presence of an ambient magnetic field. The reasons for only a small number of craters with magnetic signatures could potentially be a dynamo that was episodic, frequently reversing, or unstable in intensity and direction. Specific impact conditions (e.g., impact angle and the iron-metal content of the impacting projectile) could perhaps also be required for generating a magnetic signature.

Reference [1] Wieczorek M. et al. (2022) Lunar magnetism, [2] Weiss B. and Tikoo S.  (2014), Science, 346, [3] Garrick-Bethell, I. et al. (2009) Science 323, 356–359, [4] Garrick-Bethell, I. et al. (2017) JGR-Planets, 122, 76–93, [5] Mighani S. et al. (2020) Sci. Adv., 6, [6] Wieczorek M. (2018) JGR-Planets, 123, 291–316, [7] Halekas J. et al. (2003) Meteorit. Planet. Sci. 38, 565–578, [8] Hood L. (2011)  Icarus, 211, 1109–1128, [9] Oliveira J. et al. (2017) JGR-Planets, 122, 2429–2444, [10] Halekas J. et al. (2002) GRL, 29, 23-1, [11] Arkani-Hamed J. and Boutin D. (2014) Icarus, 237, 262–277, [12] Robbins S. J. (2019) JGR-Planets, 124, 871–892, [13] Neumann G. (2015) Sci. Adv. 1, e1500852, [14] Losiak, A. et al. (2015) LPI crater database, [15] Tsunakawa H. et al. (2015) JGR-Planets, 120, 1160–1185, [16] Ravat D. et al. (2020) JGR- Planets, 125.