The Traspena meteorite: heliocentric orbit, atmospheric trajectory, strewn field, and petrography of a new L5 ordinary chondrite

MNRAS 518, 3850–3876 (2023) https://doi.org/10.1093/mnras/stac2911 Advance Access publication 2022 October 17 The Traspena meteorite: heliocentric orbit, atmospheric trajectory, strewn field, and petrography of a new L5 ordinary chondrite Manuel Andrade , 1 , 2 ‹Jos ´ e ´ A. Docobo , 1 , 3 , 4 Javier Garc ´ ıa-Guinea , 5 Pedro P. Campo , 6 Mar Tapia , 7 , 8 Luis S ´ anchez-Mu ˜ noz , 5 V ´ ıctor Villasante-Marcos , 9 Eloy Pe ˜ na-Asensio , 10 , 11 Josep M. Trigo-Rodr ´ ıguez , 11 , 12 Jordi Ib ´ a ˜ nez-Insa , 13 Marc Campeny 14 and Jordi Llorca 15 1 CITMAga, E-15782 Santiago de Compostela, Galiza, Spain 2 Observatorio Astron ´ omico R. M. Aller (OARMA), Departamento de Matem ´ atica Aplicada, Escola Polit ´ ecnica Superior de Enxe ˜ nar ´ ıa (EPSE), Universidade de Santiago de Compostela (USC), Campus Terra, E-27002 Lugo, Galiza, Spain 3 Observatorio Astron ´ omico R. M. Aller (OARMA), Departamento de Matem ´ atica Aplicada, Facultade de Matem ´ aticas, Universidade de Santiago de Compostela (USC), Campus Vida, E-15782 Santiago de Compostela, Galiza, Spain 4 Real Academia de Ciencias de Zaragoza, Facultad de Ciencias, C/ Pedro Cerbuna 12, E-50009 Zaragoza, Spain 5 Museo Nacional de Ciencias Naturales, C/ Jos ´ e Guti ´ errez Abascal 2, E-28006 Madrid, Spain 6 Observatorio Astron ´ omico R. M. Aller (OARMA), Universidade de Santiago de Compostela (USC), Campus Vida, E-15782 Santiago de Compostela, Galiza, Spain 7 Laboratori d’Estudis Geof ´ ısics Eduard Fontser ` e, Institut d’Estudis Catalans (LEGEF-IEC), C/ Carme 47, E-08001 Barcelona, Catalonia, Spain 8 RISKNAT, Institut Geomodels, Universitat de Barcelona, Faculty of Earth Sciences, Department of Earth and Ocean Dynamics, Barcelona, Catalonia, Spain 9 Laboratorio de Magnetismo de Materiales y Magnetismo Ambiental, Instituto Geogr ´ afico Nacional, Real Observatorio de Madrid, C/ Alfonso XII 3, E-28014 Madrid, Spain 10 Departament de Qu ´ ımica, Universitat Aut ` onoma de Barcelona (UAB), E-08193 Bellaterra, Catalonia, Spain 11 Institut de Ci ` encies de l’Espai (CSIC), Campus UAB Bellaterra, Carrer de Can Magrans s/n, E-08193 Cerdanyola del Vall ´ es, Barcelona, Catalonia, Spain 12 Institut d’Estudis Espacials de Catalunya (IEEC), Gran Capit ` a 2-4, E-08034 Barcelona, Catalonia, Spain 13 Geosciences Barcelona (GEO3BCN-CSIC), C/ Llu ´ ıs Sol ´ e i Sabar ´ ıs s/n, E-08028 Barcelona, Catalonia, Spain 14 Departament de Mineralogia, Museu de Ci ` encies Naturals de Barcelona, Passeig Picasso s/n, E-08003 Barcelona, Spain 15 Institute of Energy Technologies, Department of Chemical Engineering and Barcelona Research Center in Multiscale Science and Engineering, Universitat Polit ` ecnica de Catalunya, EEBE, Eduard Maristany 10-14, E-08019 Barcelona, Spain Accepted 2022 September 29. Received 2022 September 27; in original form 2022 July 30 A B S T R A C T The Traspena meteorite fell on 2021 January 18 about 20 km south-east of the city of Lugo (Galiza, Spain), shortly after a huge and bright fireball crossed the sky for 4.84 s. Astrometric measurements obtained from the fireball cameras of the Universidade de Santiago de Compostela (USC) as well as from many casual videos were used to determine the atmospheric trajectory of this meteoroid which penetrated the atmosphere and generated sound waves that were detected at three seismic stations. The original meteoroid had a diameter of about 1.15 m and a mass around 2620 kg. It impacted the Earth’s atmosphere with a steep entry angle of about 76 . ◦7 from a height of 75.10 km until fading away at 15.75 km with a velocity of 2.38 km s −1 . Before the impact, this small asteroid was orbiting the Sun with a semimajor axis of 1.125 au, a moderate eccentricity of 0.386, and a low inclination of 4 . ◦55. A weak evidence of dynamic link with the PHA (Potential Hazardous Asteroid) Minos was investigated. During the atmospheric entry, two major fragmentation events occurred between heights of 35 and 29 km at aerodynamic pressures between 1 and 5 MPa. The strewn field was computed after calculating the individual dark flights of the main body along with two smaller fragments. Fortunately, 2 month after the superbolide, a 527-g meteorite was found. It was examined using several geochemical and petrographic analyses which allowed us to classify it as a moderately shocked (S3) L5 ordinary chondrite with a bulk density of 3.25 g cm −3 . Key words: methods: analytical –methods: data analysis –astrometry –Earth – meteorites, meteors, meteoroids – planets and satellites: atmospheres. 1 INTRODUCTION The fall of meteorites offers a natural way to obtain direct information about the formation and evolution of the minor bodies of the Solar system. Therefore, in addition to the petrochemical analysis, the  E-mail: [email protected] knowledge provided by the study of the body entering the atmosphere and suffering ablation before landing, as well as that derived from its heliocentric orbit, are essential to attain a full understanding of the meteoroids at the origin of these phenomena. Furthermore, this type of studies could help us to assess future impact hazards posed by relatively large meteoroids or small asteroids. Currently, there are a few tens of meteorites with well-computed trajectories and heliocentric orbits. In fact, to our knowledge, 43 © 2022 The Author(s). Published by Oxford University Press on behalf of Royal Astronomical Society. This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( http://cr eativecommons.or g/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3851 MNRAS 518, 3850–3876 (2023) instrumentally documented fireballs related to reco v ered meteorites with calculated pre-entry heliocentric orbits have been published up to now. Only one of them was reco v ered in Spain, the ordinary chondrite Villalbeto de la Pe ˜ na (Llorca et al. 2005 ; Trigo-Rodr ´ ıguez et al. 2006 ; Llorca, Gich & Molins 2007 ). There was a second fall, that of the Puerto L ´ apice eucrite (Llorca et al. 2009 ; Trigo-Rodr ´ ıguez et al. 2009 ), for which no measurements of the meteoroid velocity through the atmosphere were available and it was therefore not possible to calculate its heliocentric orbit. Hence, every instrumentally documented fireball that ends with the fall of a meteorite deserves to be carefully analysed. In this paper, we investigate the superbolide (with an estimated absolute magnitude of −17 ±2) observed at 00:18:56.54 UTC on Monday 2021 January 18 that produced the meteorite that fell at about 20 km south-east of the city of Lugo (Galiza, Spain). This would be the first documented fall of a meteorite in Galiza for which it has been possible, in addition, to calculate the atmospheric trajectory, the strewn field, and the heliocentric orbit previous to the impact. This huge and bright fireball was successfully recorded by the cameras of the two stations of the Universidade de Santiago de Compostela (USC) dedicated to the observation of fireballs and meteors which are located in Lugo and Santiago de Compostela. In addition, several casual video records of the superbolide were obtained by many pri v ate security cameras, one of which also made a sound recording of the sonic booms. Moreo v er, three seismic stations in the region detected direct sonic waves. In the course of this research, new mathematical models and an original software have been developed at the Escola Polit ´ ecnica Superior de Enxe ˜ nar ´ ıa (EPSE–USC) to accurately calculate the heliocentric orbit, the fireball atmospheric trajectory, the dark flight, and the strewn field using well-known methods and some no v el approaches which are described in detail. This software computes orbital and dynamical parameters, together with their uncertainties, using optimization techniques and Monte Carlo simulations. In fact, the latter have been intensively used in this research as a standard method to routinely obtain the uncertainties of all parameters. As a result, the expected meteorite strewn field has been derived and the calculated impact point of the main body compared with the actual position of the only reco v ered meteorite. This was found by chance by an inhabitant of the region 2 month after the fall and it has been classified as an L5 ordinary chondrite. The paper is organized as follows. Section 2 contains a description of the observational data and astrometric methods. In Section 3 , we provide a complete analysis of the superbolide trajectory and its photometric and dynamical characterization, including as well an acoustic and seismic analysis of the sonic booms and the shock wave coupled to the ground. The meteorite fall, including the dark flight, the computation of the strewn field, and the meteorite reco v ery, is described in Section 4 . In Section 5 , the heliocentric orbit of the parent asteroid of the Traspena meteorite is presented together with a discussion concerning potential dynamic associations with near-Earth objects (NEOs). Petrography and geochemical composition of the meteorite as well as their mineral chemistry and classification are given in Section 6 . Finally, main results and conclusions are outlined in Section 7 . 2 OBSERVA TIONAL DA TA The outstanding fireball associated to the Traspena meteorite fall, named after a small village (meaning behind the rock in Galizan Portuguese) located in the Neira ri ver v alley (Galiza, Spain) where the meteorite was found, occurred at 00:18:56.54 UTC (01:18:56.54 local time) on Monday 2021 January 18 and lasted for 4.84 s. Due to the bad weather conditions (see Fig. 1 ), only reliable astrometric measurements of the bolide were obtained at the beginning and at the end but scarcely mid-way. Ho we ver, it was possible to obtain an uncalibrated light intensity curve for most of the bolide, including the main flares. Preliminary results of the computed pre-atmospheric velocity and the slope of the trajectory indicated a high probability of a meteorite-dropping fireball. The same day of the fall, we estimated a nominal impact point close to the village of Lebruxo, in the municipality of Baralla (Lugo). Luckily for us, this location is only about 20 km from the campus of Lugo (USC). 2.1 Ey ewitness obser v ations As far as we know, and mainly due to the restrictions imposed by the emergency pandemic situation on account of COVID-19, there was only one direct eyewitness, a resident of the city of Lugo who was fortunate to see the fireball from the terrace of his penthouse. He described the phenomenon as a bluish-white ball that came out of the sky and briefly made it appear to be daytime. Moreo v er, man y other people inside their homes also reported huge and powerful flashes and countless people in an area of at least 15 000 km 2 heard the sonic booms and even w ok e up frightened in the middle of the night. Many of them reported that they thought it was a thunderclap or even an earthquake that was shaking the windows of their houses. 2.2 Video recordings 2.2.1 Dedicated CCD cameras The superbolide was recorded by four dedicated CCD cameras in the stations of Lugo, Santiago de Compostela, Madrid, and Estepa (Seville), all of them mounting wide-field video cameras (Madiedo et al. 2009 ). Ho we ver, only the first two (see Table 1 ) that belong to the USC allowed us to obtain sufficiently accurate measurements to be taken into account in the calculations after applying the appropriate astrometric reduction procedure. We note that these two cameras were installed after the daytime superbolide observed over Galiza some years ago for which, unfortunately, it was not possible to determine a reliable atmospheric trajectory which could have allowed finding the meteorites that supposedly could have fallen in the North Region of Portugal (Docobo et al. 2008 ). Although that night it was foggy in Lugo, the south-facing camera (the nearest camera to the bolide, only about 20 km from the end of that) was able to record the bolide from almost its beginning until just before its end. Since most of the meteor flares are perfectly visible on this camera, it was used to synchronize all casual videos (with ratios from 15 to 60 frames per second) with an accuracy well below onetenth of a second as well as to obtain an uncalibrated light intensity curve of the bolide. Unfortunately, weather conditions were not much better at the second nearest station in Santiago de Compostela, as the sky was almost overcast that night as well. Nevertheless, the initial part of the bolide (about 0.64 s) could be perfectly recorded on video thanks to a gap in the clouds and it was useful to measure the meteoroid velocity at the beginning of the atmospheric trajectory and to calculate its pre-atmospheric velocity. The astrometric reductions of all observations recorded at the USC stations were performed at the Escola Polit ´ ecnica Superior de Enxe ˜ nar ´ ıa (USC). The recordings from the stations located in Madrid and Estepa (Seville) are visually impressive, mostly the latter, considering that Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3852 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Figure 1. Location map of the Traspena superbolide. Annuli show the positions of the two USC fireball stations (in blue) and those of the casual cameras (in black) where the superbolide was recorded. Dashed grey lines indicate the lines of sight used in calculations. On the other hand, triangle symbols indicate the positions of the seismic stations where a seismic signal was detected. The ground projection of the atmospheric trajectory is represented by a black arrow in the middle of a map of the NW of the Iberian Peninsula (CRS: ETRS89–TM29). Inset shows the weather situation o v er Galiza (NW of Spain) on 2021 January 18 at 00:15 UTC. Fog and low clouds are shown in green in this RGB composite image (Source: Eumetsat. Projection: spherical Mercator). Table 1. Geodetic coordinates referenced to the GRS80 ( ≈WGS84) ellipsoid and orthometric heights of the USC fireball stations (dedicated to detect bolides and meteors) and the IGN seismic stations. Station Latitude ( ◦) Longitude ( ◦) Height (m) EPSE 42.99308 −7.54469 419 OARMA 42.87594 −8.55547 240 EAGO 42.80250 −8.08917 497 ECAL 41.94111 −6.73694 948 EPON 43.32694 −7.14917 541 Notes . EAGO: Agolada (Pontevedra); ECAL: Calabor (Zamora); EPON: A Pontenova (Lugo) they are 400- and 650-km away , respectively . Ho we ver, they do not allo w for suf ficiently accurate measurements and thus they were not considered at all in calculations. 2.2.2 Casual video r ecor dings In addition to the data provided by our stations, some local digital media published several casual recordings of the superbolide on the internet. Many of these were recorded by security cameras which are usually directed downwards, so the recordings mainly correspond to the final part of the superbolide. We selected seven of them: Caldas de Reis and Nigr ´ an in the province of Pontevedra; Corull ´ on, Foncebad ´ on, Le ´ on, Ponferrada, and Toral de los Vados/Toural dos Vaos in the province of Le ´ on. All of them have been particularly Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3853 MNRAS 518, 3850–3876 (2023) useful for tracking the last part of the bolide. Their positions are shown in Fig. 1 , where the ETRS89 Transverse Mercator Coordinate Reference System (ETRS-TM29) has been used as recommended for conformal pan-European mapping at scales larger than 1:500 000. Four of the wide-field video recordings did not have enough reference stars to obtain precise astrometry of the bolide. To solve this, they were calibrated by means of in situ measurements with the theodolite of distant landmarks, correcting for the difference in positions when the instrument could not be placed in the exact same spot as the camera. When possible, they were combined as well with astrometric calibrations implementing corrections of geometric lens deformation. In the case of Nigr ´ an, a no v el calibration procedure was also implemented by means of a laser that was used to target the less bright stars in the field and thus consider a larger number of reference stars. The best recording is that of Corull ´ on from where the last part of the bolide was clearly seen at altitudes between 20 ◦and 15 ◦abo v e the horizon. To calibrate this record, given the scarcity of calibration stars in the video, we used several different positions of Vega as a result of its diurnal motion throughout the year. The calibrations of all observations recorded in the casual videos were performed at the Observatorio Astron ´ omico R. M. Aller (OARMA–USC). Nevertheless, the calibration of some video recordings involves additional uncertainties due to an insufficient number of good reference points on the ground and/or calibration stars in the field of view. 2.3 Acoustic and seismic data An object at high-hypersonic velocity produces sonic booms due to the propagation of a shock wave produced by the meteoroid in the lower atmosphere. In fact, many witnesses close to the trajectory heard sonic booms shortly after the superbolide. Furthermore, strong detonations were recorded 3 min 21.5 s after the end of the superbolide by the microphone of a security camera located in Cueto, a small village close to Ponferrada and located at a geometric distance of 65 km of the mean point in the trajectory where the main flares were observ ed. Moreo v er, electrophonic sounds were reported by an eyewitness at the vertical of the municipality of Triacastela where the meteoroid entered the atmosphere. These can be caused by extra low and very low frequency radio emissions generated by plasma turbulence in the fireball w ak e being transduced into acoustic waves (Keay 1992 ). In addition, four seismic stations of the Instituto Geogr ´ afico Nacional (IGN) recorded tremors compatible in time with the impact of the meteoroid in the Earth’s atmosphere, namely A Pontenova (EPON) in Lugo, Agolada (EAGO) in Pontevedra, Calabor (ECAL) in Zamora, and Lobios (ELOB) in Ourense, although only the first three were found to actually correspond to the shock waves of the meteoroid. Table 1 shows a list of the USC fireball stations and IGN seismic stations involved in this work. 3 THE ATMOSPHERIC TRAJECTORY 3.1 The 3D geometric model of the superbolide The trajectories of bolides have traditionally been calculated using most commonly either the intersection of planes method (Ceplecha 1987 ) or the straight least squares (SLS hereafter) method (Boro vi ˇ cka 1990 ). Boro vi ˇ cka himself highlighted the loss of information about the uncertainty of individual observations when a plane is calculated for each station as the main disadvantage of the former. Moreo v er, he also pointed out its impossibility to compensate for the Earth’s rotation. In this work, we hav e dev eloped an approach that takes advantage of Borovi ˇ cka’s idea of considering the minimal distance to all observed lines of sight to find the trajectory. Therefore, every instantaneous measurement of the meteor position is used to obtain its line of sight in that moment which goes from the observer’s tridimensional position to the bolide. Since we are considering an Earth-centred inertial (ECI) reference frame, the method allows us to take into account the time change during the flight of the bolide and thus compensate every measurement of the Earth’s rotation, thereby removing the Coriolis force. In addition, this also a v oids the need to correct for diurnal aberration, another effect due to the Earth’s rotation caused by the change in the observer’s perspective. Thus, we compute the atmospheric trajectory as the 3D straight line which yields the minimal distance to all lines of sight. In the first instance, to detect potential outliers we consider the  1 -norm minimization problem (also called least-absolute minimization) rather than the least-squares minimization (i.e. the well-known  2 -norm or Euclidean norm). The former is more robust and, therefore, less sensitiv e to outliers. Moreo v er, it giv es equal weight to all observations, unlike  2 -norm which gives more emphasis to large residuals. In fact, least squares techniques are useful when the observations are subject to random and normally distributed errors. Ho we ver, if these assumptions are violated, e.g. in cases where observations are affected by systematic errors or outliers, robust estimation techniques produce better results (Rice & White 1964 ; Kiountouzis 1973 ). In a second step, when the measurement set is free of outliers, the minimization problem is solved by considering the  2 -norm. Consequently, the computation of the minimum distance straight line, T ⊂R 3 , has been addressed as a mathematical two-step optimization problem in which we want to minimize distances from every line of sight i to the 3D straight line of the trajectory D i = D i p i , s i , u , r , (1) where p i = ( x i , y i , z i ) ∈ R 3 is the position of each station in geocentric Cartesian coordinates and s i = ( ξi , ηi , ζi ) ∈ R 3 is a unit vector from each station in the direction of each line of sight for i = 1, 2, . . . , n (with n the total number of observations); u = ( x u , y u , z u ) ∈ R 3 is an arbitrarily chosen point on the trajectory; and r = ( ξr , ηr , ζr ) ∈ R 3 is a unit vector in the direction to the radiant. Ho we ver, considering that v ectors e xclusiv ely related to each station are known, the only unknowns will be u and r . Hence, we define the objective function || d|| η: R 3 ×R 3 −→ R ( u , r )  −→ || d ( u , r ) || η= n  i= 1 w i || d i ( u , r ) || η, (2) assigning an objective value to each solution, with η= 1 in the first step (  1 -norm) and η= 2 in the second one (  2 -norm), being w i the weight of each observation. Thus, we have to find the solution of the two unconstrained  ηnorm problems minimize ( u , r ) ∈ R 3 ×R 3 || d ( u , r ) || η, (3) firstly with η= 1 and then with η= 2. Finally, the solution will be the straight line T corresponding to the lo west v alue of the objecti ve function. Instead of using linearization and the gradient method, as in the case of solving the non-linear equations of the SLS method (Borovi ˇ cka 1990 ), we hav e solv ed our minimization problems carrying out parallel processing as a way to speed a large array of global (Nelder–Mead simplex, differential Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3854 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Figure 2. The ground projection of the Traspena atmospheric trajectory is shown in a map of the region (CRS: ETRS89–TM29) using a solid black line whereas the dashed line indicates the dark flight (without atmosphere). The discs show the positions on the trajectory where the superbolide was observed and their colour indicates the location from which it was observed. The black drop icon marks the actual meteorite fall site whereas the red one indicates the calculated site (we note that they differ by only 31 m so that in the figure they are practically superimposed). Inset on the left bottom sho ws de viation of each line of sight for each station from the average straight trajectory in an ECI reference frame. The sign is positive if the line of sight crosses vertical plane abo v e the trajectory. Note the different scales on the x and y axes. evolution, and simulated annealing) and local methods (gradient, conjugate gradient, Newton, BFGS, and principal axis) running in parallel with different sets of parameters. After many simulations, we have seen that methods using deri v ati ves generally fail to reach the minimum value of the objective function if the initial point is not close enough to the solution. Therefore, we have solved the two problems described by equation ( 3 ) using primarily global methods which do not require the calculation of deri v ati ves and which have provided us with an excellent initial approximation to the solution. Subsequently, the solution has been refined using local methods as well. In most cases, the method of principal axis of Brent ( 1973 ), a deri v ati ve-free algorithm with quadratic convergence, has proven to be the best choice together with the differential evolution algorithm (Storn & Price 1997 ). We assume that the observations (as well as the ECI Cartesian coordinates) are given in the epoch of date. Geodetic coordinates are referred to the GRS80 ellipsoid ( ≈WGS84) and orthometric heights (heights abo v e mean sea lev el) are calculated using the geoid-ellipsoid separation from the EGM2008 geoid model which yields a value of + 57.0 m o v er the entire region. Our calculations are based on 31 lines of sight from 9 different stations co v ering the full flight since the beginning of the bolide to the end. Measurements of the position of the bolide along its track in the recorded videos were carried out using accurate timing for each frame after a joint synchronization of all recordings. Deviations of individual lines of sight from the 3D straight trajectory fitted in an ECI reference frame are shown in Fig. 2 . The low density of data in the middle part of the trajectory is due to Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3855 MNRAS 518, 3850–3876 (2023) Table 2. The atmospheric trajectory of the Traspena superbolide. Geodetic coordinates and azimuth as well as projected length are referenced to the GRS80 ( ≈WGS84) ellipsoid. The apparent radiant is given at the beginning of the trajectory. Beginning End Time (UTC) t 00:18:56.54 ±0.04 00:19:01.38 ±0.04 Geodetic latitude ( ◦) ϕ 42.77488 ±0.00047 42.86900 ±0.00053 Geodetic longitude ( ◦) λ−7 . 1994 ±0.0055 −7 . 3119 ±0.0011 Ellipsoidal height (km) h 75.15 ±0.31 15.81 ±0.04 Orthometric height (km) H 75.10 ±0.31 15.75 ±0.04 Velocity (km s −1 ) v 15.07 ±0.11 2.38 ±0.32 Acceleration (km s −2 ) a –−7 . 28 ±0.71 Duration (s)  t 4.84 ±0.06 Length (km)  l 60.98 ±0.38 Projected length (km)  l p 14.02 ±0.40 Geodetic azimuth ( ◦) A 318.7 ±1.4 Slope ( ◦) γ76.7051 ±0.0018 Pre-atmospheric velocity (km s −1 ) v ∞ 16.42 ±0.29 Apparent radiant (RA) ( ◦) αapp R 125.49 ±0.51 Apparent radiant (Dec) ( ◦) δapp R 32.334 ±0.088 Radiant (constellation) Cancer a combination of three factors, namely, the bad weather conditions around some stations, the huge flares observed in that part of the trajectory which saturated the frames even for the most distant cameras and, as we already said, to the fact that the casual recordings obtained by the security cameras are mainly directed towards the ground. Anyho w, the relati vely lo w dispersion of the data would indicate that there are no large errors in the calibration of the videos. The asteroid hit the atmosphere from the southeast and began to be observed at a height of 75.10 km, with an average slope of 76 . ◦7051 to the horizon, at a velocity of 15.07 km s −1 . After 4.84 s, and a flight of 60.98 km, the fireball disappeared at a height of 15.75 km with a velocity of 2.38 km s −1 . We have developed a Monte Carlo approach for propagating trajectory uncertainties from the observed astrometric positions (altitudes and azimuths) of the bolide. As usual, we have considered 10 4 random samples of these positions assuming a Gaussian distribution for each coordinate with a standard deviation equal to three times the uncertainty of each of them. Then, we have calculated an optimal atmospheric trajectory for each sample using the abo v ementioned method. Accurate values of the parameters characterizing the atmospheric trajectory along with their uncertainties are shown in Table 2 . The median absolute deviation of the fit is only 70 m. On the other hand, considering that the trajectory is really slanted and that the duration of the bolide is quite short, the expected bending due to the Earth’s gravity is indeed very small, roughly 100 m at most. The atmospheric trajectory projected on the ground is plotted in Fig. 2 . It runs entirely through the foothills of the Os Ancares mountain range. The beginning is located at the vertical of the municipality of Triacastela. The fireball continued to fly o v er the municipality of L ´ ancara to finish in that of Baralla. 3.2 Velocity fit The length travelled from the beginning of the bolide as a function of time was obtained from the projection of each measurement on the calculated trajectory. This was counted from the first measurement on the video recorded at the OARMA station to the last measurement obtained at Corull ´ on. Lengths corresponding to the first 2.88 s, before main fragmentation events (see Section 3.7 ), have been calculated according to Whipple & Jacchia ( 1957 ) by fitting the exponential model L 1 = a 1 + b 1 t + c 1 e k 1 t , (4) where L 1 is the length of the fly, t is the time since the beginning, and a 1 , b 1 , c 1 , and k 1 are coefficients to be calculated. Nevertheless, considering that this bright bolide has survived until reaching a significant low height, even after having undergone a severe fragmentation and a strong deceleration, we have supposed that in the last part of its trajectory (11 observations during 1.51 s before the end) the meteoroid was being uniformly decelerated and thus we have fitted the quadratic model (McCrosky et al. 1971 ) L 2 = a 2 + b 2 t + c 2 t 2 , (5) where L 2 is the length of the last part of the fly, t is the time since the beginning of this part, and a 2 , b 2 , and c 2 are coefficients to be calculated. Continuity has been imposed as a condition on the length travelled by the meteoroid and on its first deri v ati ve with respect to time. However, we found that although imposing second-derivative continuity results in the velocity becoming a smooth function of time, the model does not fit well the dynamic behaviour of the bolide in the final part. In fact, we suppose that the acceleration may have changed abruptly in this final part as a result of the fragmentations suffered by the main body of the meteoroid until reaching a constant value. From the fittings given by equations ( 4 ) and ( 5 ), we can straightforwardly obtain velocities and accelerations. Length depending on time is shown in Fig. 3 . We observe that the bolide was clearly decelerated at lower heights until reached a final velocity of 2.38 km s −1 . On the other hand, the very first observation of the bolide corresponds to that recorded at the OARMA station between the clouds co v ering most of the field of view at that moment. Therefore, the pre-atmospheric velocity v ∞ in the geocentric inertial reference frame was obtained using a linear fit of the measurements derived from the first 17 frames of the video recorded at the OARMA station during the first 0.64 s. We assume that this was the velocity at which the meteoroid impacted the Earth’s atmosphere. This impact would have taken place at a height of about 120 km, as we can make the assumption that atmospheric effects are negligible above this height (Egal et al. 2017 ). Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3856 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Figure 3. Length modelled according to L 1 (in grey) and L 2 (in black) along the trajectory as a function of time. Confidence bands at 1 σare plotted as well. Residuals for each measurement are shown in the bottom part of the plot. Figure 4. Plot of observational data with velocity normalized to initial velocity ( v 0 = 15.07 km s −1 , in this case) and height normalized to the atmospheric scale height (usually, h 0 = 7.16 km). The solid black line shows the analytic solution y ( v) of the meteor physics equations given in Gritsevich ( 2007 ). Uncertainty bands derived from the 5-per cent trimmed variance for this fit are plotted as well. Velocities at the beginning and at the end of the bolide, as well as the pre-atmospheric velocity, are shown in Table 2 . 3.3 Dynamics of the atmospheric flight: the dimensionless coefficient method The atmospheric flight has been parametrized using the dimensionless coefficient method devised by Gritsevich ( 2007 ) from the equations describing the trajectory that Stulov ( 1997 ) introduced. Therefore, considering the heights and the velocities derived from the geometric trajectory, the dynamics of the bolide can be characterized by two parameters, namely the dimensionless ballistic αand the mass-loss βparameters. Moreo v er, since no reliable photometric information is available in this case, the application of this method derived from the meteoroid dynamic equations is the only way to accurately calculate the meteoroid mass along the entire trajectory. On the other hand, we assume that the fact that the meteoroid has undergone several fragmentations in the last part of the bolide has no noticeable implications for the description of the motion, since the application of a single body theory would still be valid to describe the motion of the largest surviving body. This method uses the theoretical height–velocity relation derived from the equations of the meteoroid motion in order to define a function that contains all the dynamic information about the motion through the atmosphere F i ( y i , w i , α, β) = 2 αe −y i − i e −β, with  i = Ei ( β) −Ei ( βw 2 i ) , i = 1 , 2 , . . . , n, (6) where Ei ( x) =  x −∞ e z z d z, is the exponential integral function. On the other hand, y i = h i / h 0 (with h 0 = 7.16 km, as usual, being the scale height of the homogeneous atmosphere) and w i = v i / v 0 (with v 0 being the initial velocity at the beginning of the luminous trajectory) are the normalized heights and the normalized v elocities, respectiv ely, and n is the total number of observations. Estimates for αand βare customary obtained by applying the least-squares method to the equation ( 6 ). The usual resolution algorithm (Gritsevich 2009 ) consists in first numerically estimating βfrom the extremum condition and then calculating αfrom that. This iterative procedure is continued until the solutions are close enough to the desired minimum. Nev ertheless, we hav e solv ed the problem by searching directly for the global minimum using the same minimization techniques described abo v e for obtaining the equation of the 3D atmospheric trajectory (see Section 3.1 ), thereby αand βhave been calculated simultaneously checking that the values obtained really correspond to the global minimum. Considering that heights and velocities for each observation are kno wn, the only unkno wns will be αand β. Hence, we define the objective function Q : R 2 −→ R ( α, β)  −→ Q ( α, β) = n  i= 1 [ F i ( y i , w i , α, β) ] 2 . (7) Thus, we have to find the solution of the constrained problem given by minimize ( α,β) ∈ R 2 Q ( α, β) subject to α> 0 , β> 0 . (8) Some measurements were discarded, not because they did not fit a hypothetical trajectory from a geometrical point of view, but because they did not verify the constraints imposed by the meteoroid dynamic equations, probably due to errors in heights or to time inconsistencies in the recordings. This means there is no solution for αand βcompatible with such a set of measures. Usually, αand βparameters are calculated without providing any estimate of their uncertainties. In order to assess these, we have carried out a no v el statistical analysis using the 5-per cent trimmed variance, a robust measure of dispersion. The α–βdiagram is shown in Fig. 4 and results are listed in Table 3 . From these parameters, we can obtain the initial mass of the meteoroid which depends on α M 0 = 1 2 c d A 0 ρ0 h 0 αρ2 / 3 m sin γ3 , (9) where γis the slope of the meteoroid flight to the horizontal, h 0 is the scale height of the homogeneous atmosphere, ρ0 the atmospheric Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3857 MNRAS 518, 3850–3876 (2023) Table 3. Parameters of the Traspena atmospheric flight: assumed atmospheric surface density, drag coefficient, initial shape coefficient, shape change coefficient, and actual measured value of the meteoroid bulk density are indicated in the first five rows. In addition, initial kinetic energy, maximum aerodynamic pressure (and time of that), ablation coefficient, ballistic and mass-loss parameters, as well as mass and size are also summarized in the next rows. Parameter Source ρ0 (kg m −3 ) 1.2250 ISA a c d 1.84 Ideal sphere A 0 1.209 Ideal sphere μ2/3 Uniform mass-loss ρm (g cm −3 ) 3.25 ±0.01 Actual measured value Initial kinetic energy, maximum aerodynamic pressure, and ablation E k (J) (3.53 ±0.86) ×10 11 ( = 0.0844 ±0.0206 kt TNT) P max (MPa) 7.84 ±0.23 t ( P max ) (s) 3.65 σ(s 2 km −2 ) 0.0652 ±0.0053 Ballistic and mass-loss parameters α3.32 ±0.26 β2.47 ±0.20 Mass and size Beginning End M (kg) (2.62 ±0.63) ×10 3 1.9 ±1.2 ∅ (m) 1.15 ±0.09 0.10 ±0.02 Note . a Acron ym of International Standard Atmosphere ( ISO 2533:1975) surface density, c d is the drag coefficient, A 0 the initial shape coefficient ( = S 0 / W 2/3 , being S 0 the initial cross-sectional area and W the body volume), and ρm the meteoroid bulk density. The instant mass of the meteoroid is obtained, as usual, after integrating the fundamental equations of meteoroid motion and ablation (Bronshten 1983 ). That is M( t) = M 0 e σ 2 [ v ( t) 2 −v 2 0 ] , (10) where v 0 is the initial velocity at the beginning of the luminous trajectory, v( t ) the instantaneous velocity in time t , and σthe ablation coefficient. Moreo v er, this instant mass is related to βsince σ= 2 β (1 −μ) v 2 0 , (11) where μis the shape change coefficient representing the rotation of the meteoroid (0 < μ< 2/3). The ablation coefficient is nearly constant for homogeneously composed bodies that undergo progressive ablation during the atmospheric trajectory. Regarding the meaning of μ, if μ= 0, thus, there is no spin while μ= 2/3 indicates that fast rotation may allow equal ablation o v er the entire meteoroid surface so that no change in the shape of the object is expected, i.e. it would be self-similar. The latter is the usual assumption based on experience (Bouquet et al. 2014 ) as it represents a uniform mass-loss of the body o v er the entire surface area. The expression given in equation ( 10 ) provides the final mass at the terminal height of the bolide. Thus, neglecting ablation after this point, this would be, if it exists, also the final mass of the meteorites. The assumed values for all the unknown parameters during the atmospheric trajectory as well as the masses and diameters together with the ablation coefficient derived from them are summarized in Table 3 . In addition, maximum dynamic pressure and impact energy are also listed. These parameters can also be used to estimate how Table 4. Physical features of each detonation in the sonic booms: time since the beginning of the first detonation, duration, average frequency, and relative sound pressure level are listed. Detonation t (s)  t (s) f (Hz) L p (dB) 1 0.000 0.032 510 (C5) −30.6 2 0.041 0.026 579 (D5) −31.7 3 0.119 0.046 584 (D5) −32.1 4 0.365 0.039 566 (C  5) −35.4 5 0.589 0.069 350 (F4) −34.5 6 0.710 0.052 306 (D  4) −35.6 7 0.794 0.048 333 (E4) −36.7 likely a fireball is to be a meteorite-dropping candidate (Gritsevich, Stulov & Turchak 2012 ; Sansom et al. 2019 ). Our calculations show that the shape of the trajectory in normalized velocity–height variables (see Fig. 4 ) can be successfully fitted to the trajectory of a single body with ablation. This would indicate that the fragmentation, although clearly observable in the videos, does not significantly affect the shape of the trajectory in the velocity– height variables. This behaviour has been reported in the past (Stulov 2000 ) and it could be due, as Barri & Stulov ( 2003 ) suggest in the case of the Bene ˇ sov fall, to the fact that the initial masses of the separated fragments before the ultimate break-up of the body are small compared to the mass of the main body. In fact, according to Barri ( 2010 ), a single body model with ablation would be the best approximation of an observation trajectory when the mass lost due to ablation is greater than the total initial mass of fragments. Anyway, we must take into account that in both studies βis derived from the ablation coefficient σsetting the shape change coefficient to μ= 0, that is, considering a non-rotating meteoroid. 3.4 Acoustic analysis In the audio recording mentioned in Section 2.3 , the silence of the night can be heard to be broken by the sonic booms caused by the meteoroid entering the atmosphere immediately followed by the barking of dogs mixed with a long rumbling sound extending o v er sev eral seconds. This is a bizarre scenario v ery similar to that described by ¨ Opik ( 1970 ) commenting on the first audio recording of the sonic boom caused by a meteoroid penetrating the Earth’s atmosphere and which resulted in the Bo v eedy meteorite (Northern Ireland) on 1969 April 25. As far as we know, this would be one of the few historical records of the sonic booms of a meteoroid entering the Earth’s atmosphere and resulting in a meteorite. Sonic booms were recorded 3 min 21.5 s after the end of the bolide by a security video camera equipped with a microphone in Cueto (Le ´ on), a small village close to Ponferrada and located at a geometric distance of 65 km of the mean point in the trajectory where the main flares of the bolide were observed. The first detonation is the strongest as well as the most complex since it is composed of a powerful start immediately accompanied by tw o weak er detonations that seem to sound almost at once. Then another detonation is heard followed by three more rapid detonations. Typically, these sonic booms are caused by a cylindrical blast wave originating from the closest point on the trajectory. An acoustic analysis of the signal has been carried out in order to determine its most rele v ant physical features. Results are summarized in Table 4 whereas the waveform and the spectrogram of the acoustic signal for the first second are shown in Fig. 5 . Furthermore, taking into account the physical features of the acoustic signal, one of us (MA) has written down the approximate Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3858 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Figure 5. Waveform (top) and spectrogram (bottom) of the first second of the acoustic signal corresponding to the sonic booms caused by the entry of the meteoroid into the Earth’s atmosphere (recorded from Cueto about 3 min 21.5 s after the end of the bolide). music, changing only the tempo, that this small asteroid produced when it entered the Earth’s atmosphere. That is 3.5 Seismic analysis The entry of meteoroids in the atmosphere generates infrasound waves by means of two mechanisms: hypersonic flow and fragmentation. Infrasound frontwaves travel through the atmosphere being refracted and guided o v er long distances in a way highly dependent on the temperature and humidity gradient as well as on the wind profile (K ulichko v 2004 ; Silber et al. 2009 ). In addition, the sonic wavefront undergoes the decay of energy with distance in its propagation through the air and subsequent interaction with the ground, becoming seismic waves and making possible its detection by seismic stations in the neighbourhood (Edwards, Eaton & Brown 2008 ). To detect sonic waves coming from an atmospheric phenomenon, infrasound stations are needed, not necessarily nearby, as the networks of these infrasound stations are very extensive due to the fact that the sonic wave travels a long distance with little attenuation. On the other hand, to detect seismic waves, seismic stations are needed fairly close to the event, as seismic waves suffer more from energy attenuation with distance. No measurements have been found at the infrasound stations nearest to the Traspena meteoroid trajectory. On the other hand, there are nine seismic stations in the surroundings owned by the Spanish Digital Seismic Network operated by the Instituto Geogr ´ afico Na- Figure 6. Seismic record section. Raw seismic data (grey waveforms) from IGN seismic stations (EPON, EAGO, and ECAL) are plotted together with processed seismic data (red waveforms) and placed at the mean distance to the atmospheric trajectory. The vertical blue lines represent the width of the source–station distance for the entire trajectory. The dashed black lines indicate the values of the time model. The line labelled as ‘ t 0 , t f sound’ represents a wave travelling at a sound velocity of 330 m s −1 directly arriving at each seismic station from the initial and the end time of the trajectory (since for a wave travelling at the speed of sound, the differences between the arri v al times from the start and end of the trajectory are small and they are represented as a single line of 330 m s −1 ). The lines labelled as ‘ t 0-seismic ’ and ‘ t f-seismic ’ represent, for the initial and final time of the trajectory, a wave that starts propagating as a sound wave and upon impacting the ground continues its journey to each seismic station as a seismic wave at 6 km s −1 . The light orange region shows the region compatible with seismic detections. The dashed blue line is the weighted regression line which provides the sound speed (278 ±93 m s −1 ) considering the experimental seismic time arri v als and distance ranges. cional , Spain ( 1999 ). All these stations are equipped with broadband instruments with nearly flat velocity responses between 0.01 and 50.0 Hz and a sampling rate of 100 counts s –1 . The raw data from these nine seismic stations did not show at first glance any notable events compatible with the bolide. In fact, these had to be detrended and 3-Hz high-pass filtered using Obspy/ PYTHON (Krischer et al. 2015 ) to find events compatible with the bolide. Finally, only four stations presented events that could be compatible with the meteoroid flight, which are as well the nearest stations (EPON, EAGO, ECAL, and ELOB). The remaining stations did not record any events due to distance (beyond 130 km) or low signal-to- noise ratio. The raw and processed seismic data together with a synthetic model of the sonic and seismic wave coming from the meteoroid flight are shown in Fig. 6 . First, the threshold of a sonic wave emitted from the beginning to the end of the bolide and propagating homogeneously in the atmosphere with a velocity of 330 m s −1 is depicted. Because of the hypersonic velocity of the meteoroid, the emission of sound waves at the starting and final points of the trajectory, propagating both at the speed of sound, are practically indistinguishable ( t 0-sound , t f-sound ). The thresholds of the combined sonic and seismic wave propagation for the initial ( t 0-seismic ) and final ( t f-sound ) emission are represented by a high seismic velocity of 6 km s −1 as an upper limit. In addition, due to the hypersonic velocity of the meteoroid flight and the fast descent in height, the seismic waves emitted at the end point were detected at the seismic stations earlier than those emitted at the start point. According to this velocity model, EPON, EAGO, and ECAL stations (see Table 1 ) are Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3865 MNRAS 518, 3850–3876 (2023) Figure 15. Lateral shift of the Traspena meteorite (main body) due to the wind during the dark flight. The vertical line shows the vertical path of the body without lateral winds. Note that both axes are not in the same scale. 4.2 Computing the strewn field Each fragment of the original meteoroid that survived the ablation phase during the atmospheric trajectory would have impacted on the ground at different locations depending on its shape and, abo v e all, on the atmospheric winds blowing during its dark flight. In fact, parameters obtained in Section 3.7 for each fragment, all of them referred to the main-body atmospheric trajectory given in Section 3.1 , have allowed to model the dark flight of each individual fragment as completely independent bodies and, thus, computing the strewn field much more accurately. In this way, the dark flight model has been applied not only to the main body but also to fragments D and B produced in the two major fragmentation events (see Fig. 12 ). To account for the wind variability, we have used Monte Carlo simulations assuming that both the zonal and the meridional components of the wind velocity are normally distributed around their nominal values at each height step. Thus, the equations of the model described in Section 4.1 have been integrated for each fragment taking random values of the wind components within a 20 per cent of variability assuming a Gaussian distribution. Each integration has been repeated taking new values of the wind components until 10 4 simulations were completed. Results concerning the dark flight of each fragment are given in Table 8 and the strewn field is shown in Fig. 16 . We observe that, as expected, the effect of the winds has been somewhat different depending on the height at which each fragment was released. Thus, fragments D and B compared to the main body A have undergone a larger deceleration in the longitudinal direction but a smaller lateral shift. Note that the region of fall of potential meteorites is unusually small because of the steep trajectory in the atmosphere. We w ould lik e to emphasize that our model does not assume a priori that the main body corresponds to the reco v ered meteorite and that this conclusion is indeed a natural result of its application. As a matter of fact, the accurate atmospheric trajectory calculated in Section 3.1 and the dynamics provided by the fine tuning of velocities given in Section 3.2 are the two factors that constrain the initial values that determine the dark flight of each fragment. On the other hand, we have estimated the mass distribution of fragments using the power-law mass frequency distribution given by Brykina & Egorova ( 2021 ) to obtain the cumulative number of fragments N m = 1 −δ δl 1 −δ m f M t −δ −l −δ + n l , (25) being m f the fragment mass, M t the total mass of all fragments, n l the number of fragments with the largest mass m l , and l = m l / M t , the dimensionless fragment mass. The power index, 0 < δ< 1, takes typical values between 0.5 and 0.7 for rocks suffering a single impact. In this case, in agreement with the results shown in Table 8 , we have taken the largest mass as that of the main body A weighing 0.527 kg and have fitted equation ( 25 ) to the masses derived in Section 3.7 for each hypothetical fragment D and B. Thus, we have derived a power index δ= 0.5825 for the fragmentation of the Traspena meteoroid. According to that, we could expect to find at least two meteorites of 0.25 and 0.14 kg that we could identify with fragments D and B, respectively, in Fig. 11 . We cannot rule out, however, that there are still smaller meteorites than those (which would have been produced in some of the small fragmentations) but, in any case, we expect their masses to be distributed as shown in Fig. 17 . However, considering that these small fireballs have been extinguished at higher heights and that likely their masses were small, this does not seem the most likely outcome. 4.3 Searching meteorites A search group coordinated from the Escola Polit ´ ecnica Superior de Enxe ˜ nar ´ ıa (EPSE–USC) at the city of Lugo, located about 20 km north-west of the predicted fall site and formed basically by members of the USC has been looking for possible meteorites in the region for a considerable period since late January 2021. Unfortunately for us, the terrain conditions and the orography of the region w ork ed against our objective. Indeed, the entire area of probable fall is located in the foothills of the Os Ancares mountains and most of the terrain is heavily v e getated, alternating with some Table 8. The dark flight of the Traspena meteorite (main body) along with the expected dark flights of two smaller fragments. Geodetic coordinates are referenced to the GRS80 ( ≈WGS84) ellipsoid. We note that the main body was reco v ered at only 31 m from the expected impact point on the ground. Actual location and mass are shown for the reco v ered meteorite. A (main body: reco v ered) D B Latitude ( ◦) ϕ m 42.87133 (actual: 42.87112) 42.86592 42.86606 Longitude ( ◦) λm −7.3228 (actual: −7.3230) −7.3113 −7.3120 Longitudinal shift (m)  l m + 782 + 469 + 445 Longitudinal shift (only wind) (m)  l m −40 −412 −389 Lateral shift (m)  x m −499 −192 −222 Time of flight (s)  t m 105 131 131 Impact velocity (m s −1 ) v m 134 132 131 Final mass (kg) M m 0.92 (actual: 0.527) 0.25 0.14 Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3866 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Figure 16. The Traspena strewn field with the predicted impact area for meteorites (map source: IGN; CRS: ETRS89–TM29). The actual position of the sole reco v ered meteorite A is indicated with a black drop icon whereas its calculated impact position (both separated only 31 m) is marked with a red drop icon. In addition, calculated impact positions of possible fragments D and B are marked with magenta and orange drop icons, respectively. Locations on the atmospheric trajectory where each fragment was released are shown with coloured squares using the same legends. The bottom bar shows the probability computed using 10 4 Monte Carlo simulations with a variability in the speed of winds of 20 per cent. The ground projection of the atmospheric trajectory is shown using a solid black line whereas the dashed line indicates the dark flight (without atmosphere). grazing areas for cattle. In addition, in the spring and summer months, we had to be very careful with ticks which are endemic transmitters of dangerous diseases in this region. Nevertheless, our main drawback in finding the meteorite was that we did not have reliable calibrations of some of the most rele v ant videos, necessary to calculate an accurate solution for the atmospheric trajectory until a few months after the fall. During this time we interviewed dozens of people and explained to them the scientific importance of reco v ering meteorites. In this regard, another of the most serious problems we had to face, which unfortunately arises every time a meteorite falls, was the arri v al of meteorite hunters who led local people to become wary of giving information and even of reporting possible reco v eries. Thus, in the space of a few weeks we could see that the initial scientific curiosity of many locals about this event had sadly been replaced by greed at the real possibility that one of them could find a meteorite and sell it for a considerable sum of money to any of the meteorite hunters. That is why we encourage our scientific colleagues to not collaborate with these meteorite hunters whose only aim is to plunder a geological heritage that, as in other countries, is also protected by law in Spain. We could thus prevent these fascinating objects of the cosmos from ending up in the hands of pri v ate collectors instead of in public scientific museums where we can all enjoy them. 4.4 The reco v ered meteorite The inhabitants of western Baralla and eastern L ´ ancara knew soon after the fall that several search groups were looking for meteorites in the surroundings. We ourselves were constantly talking to them to find out if they had seen any rare stone. Finally, a local resident, Jes ´ us ´ Angel Farelo, who already knew that there were searches in progress in the area, would find an ordinary chondrite by chance on 18 March close to a dirt track in a meadow of his family where he usually takes his cattle to graze. This property is called O Formigueiro and it is located in the small village of Traspena, in the parish of Covas (Baralla), close Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3867 MNRAS 518, 3850–3876 (2023) Figure 17. Cumulative number of fragments as a function of the normalized fragment mass with the power index in equation ( 25 ) equal to δ= 0.5825. The reco v ered meteorite is shown in black in the bottom part of the plot, while fragments that might eventually still be found, together with their masses in kilograms, are shown in grey. to the river Neira, at geodetic coordinates λ= −7 ◦19  22 . 94  and ϕ = 42 ◦52  16 .  05. Astonishingly, this same person had w alk ed this path on other occasions since the fall but it was not until that day that he found it. Ho we ver, because of the suspicions that the meteorite hunters had raised, this person decided not to make the find public and to discreetly look for a geologist he trusted who could analyse the meteorite, so he eventually passed it on to one of us (JG-G). The reco v ered meteorite e xhibited a rounded brick shape of approximately 9 ×7 ×5 cm and weighed 527.0 ±0.1 g. A bulk density of 3.25 ±0.01 g cm −3 was measured. After 2 month outdoors, the fusion crust showed some tiny rust marks. There were no traces of impact except for a small nick in one corner. It is quite possible that it had fallen several tens of metres further up the ground (to the north), as the terrain has a moderate slope, and that the rain and the passage of the cattle during the last 2 month of winter had dragged it down the hillside to where it finally appeared (see Fig. 18 ). In fact, inner sections exhibit brownish areas of iron oxides towards the periphery and a less oxidized core, i.e. a pattern of weathering traces scarcely observed in chondrites which may be due to its 2 month of terrestrial lying on a damp bed. The middle part of the only reco v ered fragment was cut into small sections for physical and geochemical study at the Museo Nacional de Ciencias Naturales (MNCN–CSIC, Madrid), the Institut de Ci ` encies de l’Espai (ICE–CSIC, Barcelona), and the Universitat Polit ` ecnica de Catalunya (Barcelona), where 29, 24, and 18 g, respectively, are currently deposited. The smaller of the two remaining pieces, weighing 182 g, was given to the local council of Baralla by the finder of the meteorite. The main piece, weighing 202 g, was acquired by the Museo de Historia Natural of the USC (MHN–USC) in order to ensure its preservation for future generations and to have it permanently on public display. 5 DYNAMICS OF THE PARENT ASTEROID 5.1 The heliocentric orbit At present, there are only a few tens of reco v ered meteorites with accurate heliocentric orbits computed from reliable observations of a fireball as shown in Table 9 . Figure 18. Background photography (by Jes ´ us ´ Angel Farelo): Traspena meteorite as found by Jes ´ us ´ Angel Farelo. It was close to a dirt track in a meadow of his family in O Formigueiro (Traspena, Covas, Baralla). Foreground photography: this ordinary chondrite exhibits several regmaglypts. The pre-entry heliocentric orbit was analytically computed from the known radiant, the pre-atmospheric velocity, and the bolide time using the method of Ceplecha ( 1987 ), therefore assuming an initial hyperbolic collision orbit with Earth. This is also the same method that has been implemented in other recent orbit calculators, which in fact basically reproduce the results obtained here, e.g. Pe ˜ na-Asensio et al. ( 2021 ), but with some particularities that we comment below that make the software developed at the EPSE–USC highly accurate. We note that the velocity before atmospheric entry is not assumed to be the velocity of the meteor at the beginning of its luminous phase (see Section 3.2 ) since we cannot neglect deceleration before the starting of the thermal ablation. The zenith attraction correction was applied using the Schiaparelli scheme with the Andreev ( 1991 ) approach at the beginning point of the trajectory. On the other hand, since the ECI coordinates of the stations were moving with time when the trajectory was calculated, the radiant does not need to be corrected for the Earth’s rotation. Ho we ver, since the ECI coordinates of the meteor are in the epoch of the date, they must be precessed to J2000. Regarding the Cartesian heliocentric ecliptic coordinates of the Earth, these were directly computed using the JPL DE441 ephemerides (Park et al. 2021 ) for the beginning of the bolide on the TDB time-scale. Uncertainties have been derived using the Monte Carlo approach described in Section 3.1 . The geocentric radiant and the geocentric velocity together with the orbital elements (referred to the J2000 equinox) are listed in Table 10 and the heliocentric orbit is shown in Fig. 19 . Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3868 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Table 9. List of meteorites (sorted by date of fall) whose reco v ery was accompanied by the necessary observational information to ensure a sufficiently accurate calculation of the heliocentric orbit. The official name of the meteorite, fall date, type, mass of the meteoroid before atmospheric entry and reco v ered mass, are given in the second to fifth columns, respectively. In addition, semimajor axis, eccentricity, inclination, and orbital type are shown in the sixth to ninth columns, respectively. The main reference for the published heliocentric orbit is given in the last column. N a Name Date (UTC) Type M e | M (kg) a (au) e i ( ◦) Orbit Reference 1 P ˇ r ´ ıbram 07/04/1959 H5 < 5000 | 5.56 2.424 0.6742 10.4244 Apollo Ceplecha ( 1961 ) 2 Lost City 04/01/1970 H5 165 | 17 1.66 0.417 12.0 Apollo McCrosky et al. ( 1971 ) 3 Innisfree 06/02/1977 L5 42 | 4.58 1.872 0.4732 12.27 Apollo Halliday, Blackwell & Griffin ( 1978 ) 4 Bene ˇ sov 07/05/1991 LL3.5,H5 4100 | 0.01 2.483 0.6274 23.981 Apollo Spurn ´ y et al. ( 2014 ) 5 Peekskill 09/10/1992 H6 < 10 000 | 12.57 1.49 0.41 4.9 Apollo Brown et al. ( 1994 ) 6 Tagish Lake 18/01/2000 C2 75 | 10 1.98 0.55 2.0 Apollo Hildebrand et al. ( 2006 ) 7 Mor ´ avka 06/05/2000 H5 1500 | 0.63 1.85 0.47 32.2 Apollo Borovi ˇ cka et al. ( 2003 ) 8 Neuschwanstein 06/04/2002 EL6 300 | 6.19 2.40 0.670 11.41 Apollo Spurn ´ y, Oberst & Heinlein ( 2003 ) 9 P ark F orest 27/03/2003 L5 11 000 | 18 2.53 0.680 3.2 Apollo Brown et al. ( 2004 ) 10 Villalbeto de la Pe ˜ na 04/01/2004 L6 760 | 3.5 2.3 0.63 0.0 Apollo Trigo-Rodr ´ ıguez et al. ( 2006 ) 11 Bunburra Rockhole 20/07/2007 Eucrite 22 | 0.32 0.8529 0.2427 8.95 Aten Spurn ´ y et al. ( 2012 ) 12 Almahata Sitta 07/10/2008 Ureilite 83 000 | 3.95 1.308201 0.312065 2.54220 Apollo Jenniskens et al. ( 2009 ) 13 Buzzard Coulee 21/11/2008 H4 15 000 | 41 1.25 0.23 25.0 Apollo Milley ( 2010 ) 14 Maribo 17/01/2009 CM2 1500 | 0.03 2.43 0.805 0.25 Apollo Boro vi ˇ cka, Popo va & Spurn ´ y ( 2019 ) 15 Jesenice 09/04/2009 L6 170 | 3.67 1.75 0.431 9.6 Apollo Spurn ´ y et al. ( 2010 ) 16 Grimsby 26/09/2009 H5 33 | 0.22 2.04 0.518 28.07 Apollo Brown et al. ( 2011 ) 17 Ko ˇ sice 28/02/2010 H5 3500 | 4.3 2.71 0.647 2.0 Apollo Borovi ˇ cka et al. ( 2013a ) 18 Mason Gully 13/04/2010 H5 40 | 0.02 2.47 0.6023 0.832 Apollo Spurn ´ y et al. ( 2011 ) 19 Kri ˇ zevci 04/02/2011 H6 62 | 0.29 1.544 0.521 0.64 Apollo Borovi ˇ cka et al. ( 2015 ) 20 Sutter’s Mill 22/04/2012 C 50 000 | 0.99 2.59 0.824 2.38 Apollo Jenniskens et al. ( 2012 ) 21 Novato 18/10/2012 L6 80 | 0.31 2.09 0.526 5.51 Apollo Jenniskens et al. ( 2014 ) 22 Chelyabinsk 15/02/2013 LL5 1.2 ×10 6 | 1000 1.72 0.571 4.98 Apollo Borovi ˇ cka et al. ( 2013b ) 23 Annama 18/04/2014 H5 472 | 0.17 1.99 0.69 14.65 Apollo Trigo-Rodr ´ ıguez et al. ( 2015 ) 24 ˇ Zd’ ´ ar nad S ´ aza v ou 09/12/2014 L3.9 170 | 0.05 2.093 0.6792 2.796 Apollo Spurn ´ y, Borovicka & Shrben ´ y ( 2020 ) 25 Porangaba 09/01/2015 L4 0.976 | –2.54 0.64 8.6 Apollo Ferus et al. ( 2020 ) 26 Sari c ¸i c ¸ek 02/09/2015 Howardite 13 000 | 15.24 1.44 0.301 22.6 Apollo Unsalan et al. ( 2019 ) 27 Creston 23/10/2015 L6 55 | 0.688 1.300 0.410 4.288 Apollo Jenniskens et al. ( 2019 ) 28 Murrili 27/11/2015 H5 –| 1.68 2.521 0.609 3.32 Apollo Sansom et al. ( 2020 ) 29 Osceola 24/01/2016 L6 > 1800 | 1.099 1.486 0.3406 13.20 Apollo Meier et al. ( 2020 ) 30 Ejby 06/02/2016 H5/6 250 | 8.94 2.81 0.655 0.96 Apollo Spurn ´ y et al. ( 2017 ) 31 Stubenberg 06/03/2016 LL6 600 | 1.47 1.525 0.395 2.07 Apollo Spurn ´ y et al. ( 2016 ) 32 Dishchii’bikoh 02/06/2016 LL7 9000 | 0.0795 1.129 0.205 21.24 Apollo Jenniskens et al. ( 2020 ) 33 Dingle Dell 31/10/2016 L/LL5 40 | 1.15 2.254 0.5904 4.051 Apollo Devillepoix et al. ( 2018 ) 34 Hamburg 17/01/2018 H4 142 | 1.0 2.73 0.661 0.604 Apollo Brown et al. ( 2019 ) 35 Motopi Pan 02/06/2018 Howardite 5500 | 0.214 1.37640 0.431861 4.29741 Apollo Jenniskens et al. ( 2021 ) 36 Ozerki 21/06/2018 L6 7 | 6.5 0.84 0.199 18.44 Aten Kartashova et al. ( 2020 ) 37 Vi ˜ nales 01/02/2019 L6 –| 50 1.217 0.391 11.47 Apollo Zuluaga et al. ( 2019 ) 38 Arpu Kuilpu 01/06/2019 H5 2.2 | 0.042 2.75 0.671 2.03 Apollo Shober et al. ( 2022 ) 39 Flensburg 12/09/2019 C1 15 000 | 0.0245 2.82 0.701 6.82 Apollo Borovi ˇ cka et al. ( 2021 ) 40 Cavezzo 01/01/2020 L5 3.5 | 0.055 1.82 0.460 4.0 Apollo Gardiol et al. ( 2021 ) 41 Novo Mesto 28/02/2020 L5 470 | 0.72 1.451 0.60866 8.755 Apollo Vida et al. ( 2021 ) 42 Madura Cave 19/06/2020 L5 64 | 1.072 0.889 0.327 0.12 Aten Devillepoix et al. ( 2022 ) 43 Traspena 18/01/2021 L5 2620 | 0.527 1.125 0.386 4.55 Apollo This paper Note . a 1: Central Bohemian Re gion (Czech Republic); 2: Oklahoma (USA); 3: Alberta (Canada); 4: Central Bohemian Region (Czech Republic); 5: New York (USA); 6: British Columbia (Canada); 7: Moravian–Silesian Region (Czech Republic); 8: Bavaria (Germany); 9: Illinois (USA); 10: Castile and Le ´ on (Spain); 11: South Australia (Australia); 12: Northern State (Sudan); 13: Saskatchewan (Canada); 14: Region Zealand (Denmark); 15: Upper Carniola (Slo v enia); 16: Ontario (Canada); 17: Ko ˇ sice Region (Slovakia); 18: Western Australia (Australia); 19: Koprivnica–Kri ˇ zevci County (Croatia); 20: California (USA); 21: California (USA); 22: Chelyabinsk Oblast (Russian Federation); 23: Murmansk Oblast (Russian Federation); 24: Vyso ˇ cina Region (Czech Republic); 25: S ˜ ao Paulo (Brazil); 26: Bing ¨ ol (Turkey); 27: California (USA); 28: South Australia (Australia); 29: Florida (USA); 30: Capital Region of Denmark (Denmark); 31: Bavaria (Germany); 32: Arizona (USA); 33: Western Australia (Australia); 34: Michigan (USA); 35: Ghanzi (Botswana); 36: Lipetsk Oblast (Russian Federation); 37: Pinar del R ´ ıo (Cuba); 38: South Australia (Australia); 39: Schleswig–Holstein (Germany); 40: Emilia–Romagna (Italy); 41: Lower Carniola (Slo v enia); 42: Western Australia (Australia); 43: Galiza (Spain). 5.2 Looking for potential dynamic associations with NEOs The orbital elements indicate that the parent asteroid of the Traspena meteorite was an object of the Apollo group of the near-Earth asteroids (NEAs) with moderate eccentricity, low inclination on the ecliptic, with the perihelion very close to the orbit of Venus, and with the aphelion close to the orbit of Mars. Moreo v er, we hav e deriv ed the Tisserand’s parameter relative to Jupiter as 5.480 which would certainly indicate an asteroidal origin. The parent asteroid of the Traspena meteorite is one of the six parent meteoroids and asteroids known so far that crossed both the orbits of Venus and Mars and, interestingly, did so with both the shortest semi-major axis and the lowest eccentricity of all of them as Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3869 MNRAS 518, 3850–3876 (2023) Table 10. The heliocentric orbit of the Traspena parent asteroid. Orbital elements are referred to the J2000 equinox. Tisserand’s parameter relative to Jupiter together with ratios between the orbital period of the parent asteroid and those of Jupiter, Mars, and Venus, respecti vely, are gi ven in the last four rows. Quantity Value Geocentric radiant (RA) ( ◦) αR 126.52 ±0.57 Geocentric radiant (Dec) ( ◦) δR 30.77 ±0.13 Geocentric velocity (km s −1 ) v g 12.11 ±0.40 Radiant (constellation) Cancer Orbital period (yr) P 1.193 ±0.025 Perihelion passage (JD) T 2 459 293.20 ±0.88 2021 March 19 at 16:48 UTC Eccentricity e 0.386 ±0.013 Semimajor axis (au) a 1.125 ±0.016 Inclination ( ◦) i 4.55 ±0.19 Longitude of the asc. node ( ◦) 297.8270 ±0.0003 Argument of the perihelion ( ◦) ω 273.93 ±0.98 Longitude of the perihelion ( ◦)  211.76 ±0.98 Perihelion distance (au) q 0.6914 ±0.0076 Aphelion distance (au) Q 1.559 ±0.035 True anomaly ( ◦) f 266.08 ±0.97 Tisserand’s parameter T Jupiter 5.480 ±0.062 P Jupiter / P 9.94 ±0.20 P Mars / P 1.576 ±0.032 P Venus / P 0.516 ±0.011 shown in Fig. 20 . In addition, it was as well the sixth parent meteoroid or asteroid with the shortest distance at the aphelion. Unlike the identifications between meteor showers and comets, which were clearly established from the second half of the 19th century with the notable examples of the Perseids and Leonids (Schiaparelli 1867 ), the relationship between meteoroid streams and asteroids is much more recent. Meteoroids (which can eventually produce meteorites) are generated in collisions between asteroids with resulting velocities much smaller than the orbital velocity, so the orbits of young meteoroids are very similar to those of their parent bodies (Davis et al. 1979 ). In this way, Drummond ( 1982 ) searched plausible associations of NEAs with current meteoroid streams using the function D D (Drummond 1981 ) to estimate distances between orbits. Later, Halliday, Blackwell & Griffin ( 1990 ) suggested the existence of streams of fragments that cross the Earth’s orbit and are responsible for a considerable fraction of the meteorites that fall to Earth. At present, dynamic associations between well-documented meteorites have been thoroughly studied (Spurn ´ y et al. 2003 ) and even impacts of asteroids have been predicted before falling as meteorites since the first case of the asteroid 2008 TC 3 (Jenniskens et al. 2009 ). In order to determine membership in known meteoroid streams, Southworth & Hawkins ( 1963 ) were the first to introduce a distance function, D SH ( e , a , i , , ω), with which to measure the dissimilarity of two orbits. A widely used modification of that, D D ( e , a , i , , ω), was given by Drummond ( 1981 ) and Jopek ( 1993 ) proposed a hybrid function, D J ( e , a , i , , ω), alternative to the first two. Moreover, Steel, Asher & Clube ( 1991 ) and Asher, Clube & Steel ( 1993 ) proposed other functions, D SAC ( e , a , i ) and D ACS ( e , a , i ), respectively. The latter two may be useful to study very broad streams having a wide arc of nodal longitudes. Usually, certain previously established threshold values of these functions are considered to determine whether or not there is a dynamic association but we believe that this is not justified in Figure 19. The heliocentric orbit for the parent asteroid of the Traspena meteorite as seen from the ecliptic north pole is shown in black. The actual orbital inclination is 4 . ◦55. The light grey region represents the likelihood computed from 10 4 Monte Carlo simulations based on the formal uncertainties of the radiant and the pre-atmospheric velocity. The orbits of Mercury, Venus, Earth, and Mars are shown in light grey, green, blue, and red, respectively, and the Earth’s position in its orbit is shown by a small blue disc. The small curved arrows indicate the motion of the Earth and that of the Traspena parent asteroid while the arrow at the bottom right shows the direction to the point of the vernal equinox. Figure 20. Distribution in the ( a , e ) diagram of the parent meteoroids and asteroids of the well-documented meteorites listed in Table 9 . Traspena is the Venus- and Mars-crosser with both the shortest semi-major axis and the lowest eccentricity. the case of asteroids and that these functions only allow the selection of parent asteroid candidates. Furthermore, all these functions violate the triangle inequality and cannot, therefore, be formally defined as metrics but rather as quasimetrics as they satisfy a relaxed triangle inequality (Milanov, Milanova & Kholshevnikov 2019 ). Ne vertheless, se veral actual metrics have been defined recently that allow the calculation of distances between Keplerian orbits in a rigorous way (Kholshevnikov 2008 ). Thus, in our analysis, we have primarily considered the metric  2 ( p , e , i , , ω) given by Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3870 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Table 11. Similarity between orbits of the asteroids with possible dynamic associations in the five-dimensional space H of all Keplerian orbits (asteroids are ordered according to the value of the metric  2 , so Minos would be the asteroid with the tenth lowest value), in the three-dimensional quotient space H 5 , and in the four-dimensional spaces H 3 and H 4 of all Keplerian orbits (for the shake of brevity, only Minos values are shown in the quotient spaces). Absolute magnitudes are listed in the second column. Five-dimensional space H Asteroid H a  2 D SH D D D J 2022 AZ 23.91 0.0544 0.0752 0.0654 0.0678 2015 BF 26.60 0.0684 0.0853 0.0618 0.0753 2014 PG51 20.70 0.0738 0.0811 0.0837 0.0783 2017 PZ26 26.90 0.0798 0.0773 0.0981 0.0767 2019 AX14 26.68 0.0894 0.1055 0.0837 0.1006 2014 SD224 22.36 0.0972 0.0860 0.0835 0.0823 2019 AV 24.30 0.1010 0.1498 0.0977 0.1074 2018 BD 30.10 0.1030 0.1220 0.1532 0.1139 2015 PA57 24.50 0.1053 0.1224 0.0687 0.1108 1989 QF (Minos) 18.57 0.1059 0.1099 0.0487 0.1094 Three-dimensional quotient space H 5  5 D SAC D ACS 1989 QF (Minos) 0.0277 0.0326 0.0300 Four-dimensional quotient spaces H 3 H 4  3  4 1989 QF (Minos) 0.0508 0.0636 Kholshevnikov et al. ( 2016 )  2 2 = (1 + e 2 1 ) p 1 + (1 + e 2 2 ) p 2 −2 √ p 1 p 2 ( cos I + e 1 e 2 cos P ) , (26) with cos I = cos i 1 cos i 2 + sin i 1 sin i 2 cos ( 1 −2 ) , cos P = sin i 1 sin i 2 sin ω 1 sin ω 2 + ( cos ω 1 cos ω 2 + cos i 1 cos i 2 sin ω 1 sin ω 2 ) cos ( 1 −2 ) + ( cos i 2 cos ω 1 sin ω 2 −cos i 1 sin ω 1 cos ω 2 ) ×sin ( 1 −2 ) , where p 1 and p 2 are the semilatus rectum of each orbit. This provides the actual distance between two Keplerian orbits ( ε 1 , ε 2 ) on the fivedimensional space H of all Keplerian orbits. In addition, metrics  3 ( p , e , i , ω),  4 ( p , e , i , ), and  5 ( p , e , i ) given in the same paper, provide the minimum distances between the orbits on three- and fourdimensional quotient spaces of all Keplerian orbits whose elements are the class of orbits with fixed p , e , and i values and all possible values of and/or ω. This type of metrics may be of interest when orbital nodes and/or periastra undergo secular perturbations whereas the remaining elements change slowly. We have accomplished a search for a possible asteroidal parent body of the Traspena meteorite among the NEOs taking their orbital elements from the Small-Body DataBase (SBDB) of the JPL Solar System Dynamics & Center for NEO Studies (CNEOS) and measuring the similarity between orbits by means of the mentioned -metrics. Anyho w, v alues of the most used D -functions were also calculated and compared with the former. Results derived from - metrics and D -functions for the most probable dynamic associations of a total of 29 430 NEOs (SBDB accessed on 2022 July 11) are shown in Table 11 . Table 12. -metrics and D -functions ordered according to the pvalues obtained with the Monte Carlo procedure used to assess the probability of attaining these small orbital distances as regards the hypothetical Traspena– Minos dynamic association. -metrics and D -functions Draws p -value  2 53 743 ±539 0.0319 D D 19 395 ±194 0.0884 D SH 14 472 ±143 0.1185 D J 8546 ±85 0.2007  3 30 818 ±315 0.0556  4 12 300 ±123 0.1394  5 7344 ±72 0.2335 D SAC 1983 ±19 0.8649 D ACS 1731 ±17 0.9908 The most prominent asteroid with which Traspena could be associated is Minos, also known as 6239 Minos and 1989 QF (Shoemaker et al. 1989 ). This is a bright ( H a = 18.57 mag) NEA, classified as a potentially hazardous asteroid (PHA) of the Apollo group, with a size of about 0.5 km and a rotation period of 3.6 h. To assess the probability of attaining this small orbital distance, we have followed a Monte Carlo procedure considering not only the distance between orbits but the absolute magnitude as well (Borovi ˇ cka et al. 2013b ). We have used a debiased orbit and absolute–magnitude distribution model containing 802 000 synthetic NEOs (with 17 < H a < 25) parametrized by their orbits and absolute magnitudes (Granvik et al. 2018 ). Therefore, to calculate the expected probability of such small distance, we have carried out a null hypothesis significance testing considering the null hypothesis H 0 : the asteroid Minos and Traspena are not dynamically associated , where we understand that two bodies are dynamically associated if they reside in nearly identical heliocentric orbits with this common motion suggesting a joint formation scenario caused by asteroid collisions or by the rotational fission of a rubble pile. Thus, we have extracted random pairs of synthetic NEOs from the database until one case with a shorter distance than that between Traspena and Minos was obtained. We repeated this procedure 10 4 times, recording the number of draws needed each time to find a match. We obtained an average of 53 743 ±539 draws using the metric  2 . Considering that there are 1715 NEOs of the size of Minos or larger, this gives a p -value of 0.0319. All p -values obtained for the whole set of -metrics and D -functions are listed in Table 12 and, in addition, a bar chart plot is shown in Fig. 21 . The only metric with a p -value below the typical threshold of 0.05 for the significance level is  2 . Of the remaining metrics, the best performing is  3 , that is defined in the metric space H 3 in which e , a , i , and ω are fixed and takes all possible values. This could be interpreted as that the hypothetical dynamic association may not depend on the node angles. In this case, the dynamical link would persist even if there were a long-term precession of the orbits. With respect to the D -functions, none of them allow in any way to reject the null hypothesis, so that if, as usual, we were to use these functions as criteria to determine a possible association between Traspena and Minos, we should definitely rule out such a possibility. Nevertheless, based on the p -value obtained for the primary metric  2 , we reject the null hypothesis, H 0 , at a significance level of 0.05 (corresponding approximately to 2 σ), in fa v our of the alternative hypothesis. In this way, we can say that there is some evidence at this level to conclude that Traspena and Minos are dynamically Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3871 MNRAS 518, 3850–3876 (2023) Figure 21. Bar chart plot showing p -values obtained for the hypothetical Traspena–Minos dynamic association considering the whole set of -metrics and D -functions ordered according to the metric space in which they are defined. The red dashed line indicates the typical threshold of 0.05 for the significance level. associated, possibly the former being a fragment produced during a mass-shedding event in the past. Ho we ver, this e vidence is weak and we would prefer to have at least a 3 σ-level association as customary in physics, in order to be convinced more conclusively of the evidence for such an association. Therefore, we cannot conclusively state that there is a dynamic link between Traspena and the asteroid Minos. As a matter of fact, we must take into account that the orbits of meteorites originating from asteroids have probably been chaotically modified under the influence of the Yarko vsk y effect, orbital resonances, and close encounters with the inner planets during millions of years (Vokrouhlick ´ y & Farinella 2000 ). Hence, we acknowledge that more studies are needed assessing the likelihood of past orbital convergence of the Traspena orbit and that of Minos. In this way, a long-term backward integration could be interesting in order to investigate their dynamics as well as hypothetical resonances with the inner planets. Regarding the latter, we should consider that resonances with Venus, Earth, and Mars in the region 1 < a < 2.5 au would be at least as strong as resonances with Jupiter (Gallardo 2006 ). 6 ANALYSIS OF THE METEORITE 6.1 Methodology The bulk meteorite was first examined by X-ray computed tomography (micro-CT) with a Nikon- XTH 160 system equipped with a Varex 2520DX flat-panel detector at the Museo Nacional de Ciencias Naturales (MNCN-CSIC). The bulk density of the meteorite was measured with a hydrostatic Westphal balance using distilled water at 4 ◦C. Then, the meteorite was sliced into two parallel cuttings in order to obtain several fresh fragments and thin sections to characterize the meteorite with different physical and chemical methods. Scanning electron microscopy (SEM) images and mineral chemistry analyses were carried out by Electron probe microanalysis (EPMA) on a JEOL JXA-8230 instrument equipped with five wav elength-dispersiv e spectrometers (WDS) at the Scientific and Technological Centres of the University of Barcelona (CCiTUB). Spot analyses had an accelerating voltage of 20 kV, a beam current of 15 nA, and a spot size of 1–2 micrometers. Analytical standards used were: hematite (Fe, LIF, K α), rutile (Ti, PET, K α), periclase (Mg, TAP, K α), rhodonite (Mn, LIF, K α), Al 2 O 3 (Al, TAP, K α), Cr 2 O 3 (Cr, PET, K α), NiO (Ni, LIF, K α), metallic cobalt (Co, LIF, K α), diopside (Si, TAP, K α), fluorapatite (P, PET, K α), pyrite (S, PET, K α), wollastonite (Ca, PET, K α), orthoclase (K, PET, K α), and albite (Na, TAP, K α). Magnetic susceptibility measurements were performed on five meteorite fragments (28 ×48 mm) using an Agico KLY4 instrument as well as a Bartington MS3-Magnetic Susceptibility System, and recording 10 measurements for each fragment, by the Instituto Geogr ´ afico Nacional (IGN). Magnetic hysteresis and remanence curves were analysed with a Coercivity Spectrometer developed by Kazan University (maximum applied field of 500 mT, nine measurements per point, increments of 0.5 mT). The Natural Remanent Magnetization (NRM) of the meteorite fragments was measured with an Agico JR5A spinner magnetometer, shielded from external field by self-compensated Helmholtz coils. About 1 g of the meteorite was carefully grinded and homogenized to obtain a powder for X-ray diffraction (XRD) analyses which were performed on a Bruker D8-A25 instrument equipped with a Cu X-ray source (Cu K αradiation) and a LynxEye position sensitive detector at Geosciences Barcelona . The XRD scans were acquired between 4 ◦and 70 ◦in 2 θwith a 0 . ◦016 step size and equi v alent counting times of 346 s. Phase identification was carried out by using the DIFFRAC.EVA software together with the Powder Diffraction File (PDF-2) data base. The Rietveld method (Young 1993 ) was employed in order to obtain information about mineral abundances in the meteorite and also about the Fe/Mg content of the major silicate phases. F or the Rietv eld analyses, the TOPAS 4.2 softw are from Bruk er w as used. Preferential orientation effects were considered for some of the phases by using the March–Dollase algorithm (Young 1993 ). 6.2 Description, composition, and petrography The Traspena meteorite weighted 527.0 g and was totally co v ered by a black and well-developed fusion crust sho wing se veral thumb-sized regmaglypts (Fig. 18 ). The bulk density is 3.25 ±0.01 g cm −3 and the magnetic susceptibility is log χ( ×10 −9 m 3 kg −1 ) = 4.89. These values compare well with the values reported for L-type ordinary chondrites (Rochette et al. 2003 ; Consolmagno et al. 2006 ). Fig. 22 shows micro-CT images of the bulk meteorite corresponding to the 3D distributions of mineral sizes and grains with higher densities (FeNi metal and FeS), respectively. X-ray tomography results indicate that metal grains are grouped in certain areas but there is no agglomeration of metal phases in shock veins or cracks. In addition, no porosity was observed. The petrographic and mineralogical study of the meteorite carried out by optical microscopy of transmitted light and SEM (BSE) reveals the occurrence of chondrules set in a recrystallized silicate matrix, with an apparent diameter ranging from ca. 0.4 to 2 mm. These chondrules exhibit porphyritic olivine pyroxene (POP), barred olivine (BO), and granular olivine pyroxene (GOP) textures (Figs 23 and 24 ). FeNi-metal, troilite (Tro), and chromite (Chr) are observed throughout. Most plagioclase (Pl) grains exhibit sizes of 5–50 μm. FeNi metal grains are often surrounded by brown metal oxides (Fig. 23 b). Micro-XRD measurements indicate that they correspond to trevorite-type ferroan phases, (Ni,Fe)Fe 2 O 4 , which is also in accordance with a mineral phase showing high coercivity. Olivine (Ol) grains contain planar fractures and exhibit undulatory extinction, indicating shock stage S3 (St ¨ offler, Keil & Scott 1991 ). Powder XRD measurements allowed us to perform a semiquantitative bulk analysis of the major minerals in the meteorite. The XRD pattern is dominated by reflections from Ol (forsterite) and low- Ca pyroxene (orthoenstatite). In addition, weaker features arising from Pl feldspar, Tro, and Kam can be identified. Semiquantitative Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3872 M. Andrade et al. MNRAS 518, 3850–3876 (2023) Figure 22. Micro-CT-Scan Images of the Traspena meteorite: (a) 3D projections of mineral sizes and (b) distribution of mineral grains with higher densities. Figure 23. Optical microscope images obtained with transmitted light (1) and cross-polarized light (2). Figures a1 and a2 show an elongated olivinerich (Ol) chondrule with interstitial orthopyroxene (Opx). Figures b1 and b2 show a barred chondrule comprised of alternating Ol and Opx surrounded by subhedral grains of the typical mineral association: Ol, Opx, Pl, and accessory sulfides, FeNi metal, and oxides. Figure 24. SEM (BSE) images: (a) Subhedral to anhedral Tro and kamacite (Kam) grains associated with a silicate matrix formed by Ol, Opx, and Pl with accessory Chr; (b) Ol-rich chondrule with interstitial Pl and accessory subhedral grains of Tro, and Kam; (c) Typical elongated barred texture comprised of alternating bands of Ol and combined Opx, and Pl; (d) Euhedral Opx crystal with alternating bands of Pl, surrounded by Ol and accessory Tro, and Kam; (e), (f) Subhedral to anhedral Tro and Kam grains associated with Chr in a typical Ol granular matrix with interstitial Pl. Table 13. Lattice parameters and cationic compositions for the main silicate minerals (forsterite and orthoenstatite) as obtained from Rietveld analysis of powder XRD. Forsterite Orthoenstatite a ( Å) 4.774 (1) 18.286 (1) b ( Å) 10.271 (1) 8.866 (1) c ( Å) 6.013 (1) 5.205 (1) Mg (mol per cent) 74 75 Fe (mol per cent) 26 20 Ca (mol per cent) – 5 mineralogical composition as obtained from Rietveld analysis yields ca. 44 per cent Ol, 43 per cent low-Ca pyroxene, 8 per cent Pl feldspar, 4 per cent Tro, and 1 per cent FeNi metal. The modal abundances found for Ol and low-Ca pyroxene are close to those typically found in ordinary chondrites (Dunn et al. 2010a , b ). Unit-cell refinement of forsterite and enstatite, in combination with the data of Morrison et al. ( 2018 ) regarding the compositional dependence of unit-cell parameters of minerals in rocks and extraterrestrial bodies, has also allowed us to evaluate the average cationic composition of the two major minerals in the meteorite. Table 13 shows the refined lattice parameters of forsterite and orthoenstatite, together with their corresponding average Mg/Fe and Mg/Fe/Ca contents, which correspond to Fa = 26 per cent in Ol and Fs = 20 per cent in low-Ca pyroxene. These values suggest that Traspena is a L-type chondrite. The fact that no monoclinic pyroxene phase is detected by XRD, and the observation of well-defined reflections for the Ol and Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 The Traspena meteorite 3873 MNRAS 518, 3850–3876 (2023) Table 14. Mean compositions (wt. per cent) of the Traspena meteorite metal and sulfide phases as obtained by EPMA. Kamacite Taenite Troilite Number of grains 10 11 24 Si < 0.1 < 0.1 < 0.1 Mn < 0.1 –< 0.1 Fe 86.0 ±8.7 67.2 ±3.3 64.1 ±0.4 Co 0.8 ±0.2 0.4 ±0.2 < 0.1 Ni 13.2 ±9.2 32.6 ±3.0 – P ––– S – – 36.0 ±0.4 Total 100.1 ±0.8 100.2 ±0.8 100.2 ±0.4 Table 15. Mean compositions (wt. per cent) of the Traspena meteorite silicates and chromite as obtained by EPMA. Olivine Low-Ca Plagioclase Chromite Pyroxene Number of grains 10 11 24 21 SiO 2 38.0 ±0.6 51.1 ±0.7 65.1 ±0.8 < 0.1 TiO 2 < 0.1 0.2 ±0.1 –2.6 ±0.1 Al 2 O 3 –< 0.1 21.2 ±0.3 5.7 ±0.4 Cr 2 O 3 < 0.1 < 0.1 – 55.8 ±1.1 FeO 23.2 ±0.8 14.1 ±0.4 0.5 ±0.2 30.8 ±0.6 MnO 0.5 ±0.1 0.5 ±0.1 < 0.1 0.8 ±0.1 MgO 38.1 ±0.7 28.7 ±0.5 < 0.1 2.3 ±0.4 CaO –0.8 ±0.3 2.2 ±0.2 < 0.1 Na 2 O < 0.1 < 0.1 9.6 ±0.9 – K 2 O –< 0.1 1.0 ±1.0 – NiO < 0.1 < 0.1 –< 0.1 CoO –––< 0.1 Total 99.8 ±0.6 99.7 ±0.8 99.5 ±1.0 98.2 ±1.0 End member Fa 25.5 ±0.6 Fs 21.3 ±0.5 Ab 84.1 ±3.2 Wo 1.5 ±0.7 Or 5.4 ±2.0 pyroxene phases implying high compositional homogeneity, indicate high petrological grade. 6.3 Mineral chemistry and classification Table 14 shows the mean compositions of metal and sulfide phases as obtained by electron microprobe analysis (EPMA). The composition of Kam is 86.0 ±8.7 wt. per cent Fe, 13.2 ±9.2 wt. per cent Ni, and 0.8 ±0.2 wt. per cent Co, and that of taenite is 67.2 ±3.3 wt. per cent Fe, 32.6 ±3.3 wt. per cent Ni, and 0.4 ±0.2 wt. per cent Co. The composition of Tro is very homogeneous, 64.1 ±0.4 wt. per cent Fe and 36.0 ±0.4 wt. per cent S. Table 15 shows the mean compositions of silicate phases and chromite obtained by EPMA. The Ol composition is Fa 25.5 ±0.6 mol per cent (range of 24.5–26.1 mol per cent), the low-Ca pyroxene composition is Fs 21.3 ±0.5 mol per cent (range of 20.9–21.8 mol per cent), Wo 1.5 ±0.7 mol per cent (range of 0.97–2.0 mol per cent), and the Pl composition is Ab 83.9 ±3.8 mol per cent (range of 80.1– 85.4 mol per cent), Or 5.4 ±2.7 mol per cent (range of 8.1– 4.3 mol per cent). The good agreement between the data obtained for the major silicate phases, Ol and pyroxene, by EPMA (see Table 15 ) and XRD (see Table 13 ), should be remarked. Fig. 25 shows the chromite composition of the Traspena meteorite along with mean values for H, L, and LL ordinary chondrites (Bunch, Figure 25. Plot of the chromite composition of the Traspena meteorite along with mean values for H, L, and LL ordinary chondrites (Boschi et al. 2017 ). Figure 26. Plot of (a) the mean Ol and pyroxene compositions and (b) the Co content in Kam of the Traspena meteorite along with mean values for H, L, and LL ordinary chondrites (Rubin 1990 ; Grossman & Rubin 2006 ). Keil & Snetsinger 1967 ; Wlotzka 2005 ; Boschi et al. 2017 ). Fig. 26 shows the mean Ol and pyroxene compositions (a) and Co content in Kam (b) of the Traspena meteorite along with mean values for H, L, and LL ordinary chondrites (Rubin 1990 ; Grossman & Rubin 2006 ). Based on the mineral composition values, absence of monoclinic pyroxene, presence of chondrules, size of Pl grains, modal distrib ution, b ulk density, and magnetic susceptibility values, the Traspena meteorite is classified as a L5 ordinary chondrite. 7 CONCLUSIONS The approximate cumulative frequency of impacts from objects larger than D metres across was estimated by Brown et al. ( 2002b ) to be N = 10 c 0 D −d 0 , being c 0 = 1 . 568 ±0 . 030 and d 0 = 2 . 70 ± 0 . 08. Thus, about 25 objects like the Traspena asteroid collide with the Earth every year which means that, in a region the size of Galiza, we would have to wait nearly 700 yr to see such a meteorite fall (and about 35 yr for a similar event to take place in the whole of the Iberian Peninsula). This would explain why Traspena is the first well-documented meteorite to fall in Galiza and to be reco v ered. This fall took place on 2021 January 18 at 00:18:56.54 UTC and although clouds and fog considerably affected some of the Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023 3874 M. Andrade et al. MNRAS 518, 3850–3876 (2023) observations, it was possible to calculate an accurate atmospheric trajectory, as well as the expected strewn field, and even a very precise pre-entry heliocentric orbit. The original asteroid, with a mass of 2620 kg and a diameter of 1.15 m, had a pre-atmospheric velocity of 16.42 km s −1 . It entered the atmosphere at 75.10 km of height and finished its luminous flight at 15.75 km of height with a velocity of 2.38 km s −1 . The slope of the atmospheric trajectory was 76 . ◦7051 and the kinetic energy was estimated to be 0.0844 kt TNT. We have detected the coupling at the seismic stations of the direct sound wave with an average sound propagation speed of 278 ±93 m s −1 which is compatible with the direct sound wave at the seismic station. On the other hand, seismic waves corresponding to the emission of sound waves by the hypersonic flight (cylindrical, surf ace w aves) and sound w aves corresponding to ablations (spherical waves) were identified, and their frequency content was found to peak at around 15 Hz. The three component (NS, EW, Z) characteristics of the event signals at the three stations show a clear NS directivity at the ECAL station and a similarity of morphology for the EPON and EAGO stations which are transverse to the trajectory . Finally , each of the main ablations were calibrated as a source emitting energy between 2 ×10 7 and 1.3 ×10 10 J. The mass-loss of the meteoroid through the Earth’s atmosphere was derived using a new four-parameter mathematical model (based on the hyperbolic tangent function) which provides an explicitly time-dependent expression. The physical parameters needed to fit the model were obtained from a non-linear least-squares minimization using the differential evolution method. This model not only fitted the experimental data very well, but also provided the time interval at which the meteoroid experienced the maximum ablation regime as well as the time at which the mass-loss was maximum together with its value, i.e. 1.81 ×10 3 kg s −1 at 2.75 s from the beginning of the bolide. Consequently, the meteoroid would have lost the half of its mass in less than a second. The first large fragmentation occurred at a height of 35 km under an aerodynamic pressure of 1.52 MPa and the sole reco v ered meteorite survived intact at maximum pressure of 7.84 MPa. This weighed 527 g and was reco v ered 2 month after the fall at a distance of 31 m from the nominal position of the impact point calculated after the dark flight. We estimate that some meteorites may still be out there, in particular, two fragments of masses around 0.25 and 0.14 kg distributed as shown in Fig. 16 . The Traspena meteorite is a L5 S3 ordinary chondrite. It has a bulk density of 3.25 ±0.01 g cm −3 and a magnetic susceptibility of log χ( ×10 −9 m 3 kg −1 ) = 4.89. It contains POP, BO, and GOP chondrules with an apparent diameter of 0.4–2 mm. The Ol composition is Fa 25.5 ±0.6 mol per cent and the low-Ca pyroxene composition is Fs 21.3 ±0.5 mol per cent and Wo 1.5 ±0.7 mol per cent. There is no monoclinic pyroxene. The size of the Pl grains is 5–50 μm and their composition is Ab 83.9 ±3.8 mol per cent and Or 5.4 ±2.7 mol per cent. Kam contains 13.2 ±9.2 wt. per cent Ni and 0.8 ±0.2 wt. per cent Co. The parent asteroid was an Apollo-type object orbiting the Sun with an orbital period of 1.193 yr on a moderately eccentric orbit, 0.386, with a low inclination, 4 . ◦55. Its perihelion was inside the Venus orbit and the aphelion just beyond the orbit of Mars. With a Tisserand’s parameter of 5.480, this is an orbit with a clear asteroidal origin. It is worth mentioning that the parent asteroid of the Traspena meteorite is the Venus- and Mars-crosser with both the shortest semimajor axis and the lowest eccentricity. On the other hand, research on the hypothetical dynamic link of Traspena with the PHA Minos is inconclusive because even though new metrics in the space of all Keplerian orbits provide some evidence in its fa v our, this shows weak statistical significance and therefore further studies are needed. During this research, new mathematical models and an original software were developed at the Escola Polit ´ ecnica Superior de Enxe ˜ nar ´ ıa (EPSE–USC) to determine both the atmospheric trajectory of the meteoroid through the Earth’s atmosphere and the dark flight, taking into account as many variables as possible. In addition, the application of these techniques to a few individual fragments for which astrometric measurements were available allowed to obtain a strewn field with the most probable fall locations for each of them. In parallel, the development of new computing tools was necessary to carry out these tasks. Therefore, procedures based on optimization algorithms and on Monte Carlo techniques were routinely incorporated into the new software in order to automate the complete calculation workflow from the astrometric measurements of the Traspena superbolide position to the determination of the heliocentric orbit and the search for potential dynamic associations with NEOs. Ho we ver, we w ould lik e to point out that no deterministic model, no matter how sophisticated, can account for absolutely all parameters due to lack of knowledge about them, e.g. the initial shape of the meteoroid, its strength, the actual winds during the dark flight, and so on. In our opinion, the most decisive factors, usually unknown, are the actual details of the fragmentations during the atmospheric trajectory and, abo v e all, the actual winds acting in the dark flight. Regarding the former, we must accept, as Spurn ´ y et al. ( 2020 ) have already observ ed, that an y fragmentation model is not able to reveal all details of the fragmentation process and that actually this is probably more complicated. On the other hand, concerning the influence of the wind field on the determination of dark flight, Ceplecha ( 1987 ) pointed out that poor knowledge was one of the main complications in the computations. We agree, as recent studies using Monte Carlo simulations have shown, that the main factor contributing to the shape, size, and location of the strewn field is the atmospheric wind profile (Moilanen, Gritsevich & Lyytinen 2021 ). Thus, drifts in the computed position of each fragment would be expected as well in the case of Traspena. These could be up to 300 metres according to estimates by Andrei ´ c ( 2012 ). Nevertheless, in this case, we have practically found the exact impact location of the main body, as the difference between the calculated and the actual locations where the meteorite was found is only of 31 m. This would be indeed the best proof of the validity of the mathematical models and methods developed and the assumptions adopted. ACKNOWLEDGEMENTS The authors would like to start by expressing their deepest gratitude to their colleagues from the Department of Applied Mathematics in the Campus Terra of Lugo (USC) for their perseverance and tireless work in the search for possible meteorites (in alphabetical order of surnames): Gerardo Casal, Juan B. Ferreiro, Duarte Santamarina, Miguel E. V ´ azquez-M ´ endez, and Miguel A. Vilar. The same grateful acknowledgement goes to the other members of the search team who occasionally participated in the searches: Chelo Astorgano, Ana Cabana, Pedro Ferreiro, Marolas Leis, Cidre Santamarina, Sarela Santamarina, and Paula V ´ azquez. We are also grateful to Jes ´ us ´ A. Farelo (Lebruxo, Baralla) for his description of the meteorite reco v ery and to Marcos A. Gonz ´ alez for his efforts as director of the Museo de Historia Natural (MHN– USC) to ensure that this outstanding piece of the Galizan geological heritage became part of the collection of our most important museum and that it did not end up on the black market for meteorites. Downloaded from https://academic.oup.com/mnras/article/518/3/3850/6762217 by Universidade de Santiago de Compostela user on 03 January 2023