Reconstruction of ejecta distribution of DART spacecraft impact on asteroid Dimorphos using data from LICIACube/LUKE

Introduction:  The NASA Double Asteroid Redirection Test (DART) mission tested planetary defense by performing a kinetic impact on Dimorphos, the smaller asteroid in the Didymos binary system, on September 26, 2022. The experiment aimed to validate kinetic impact as a method for asteroid deflection. The impact successfully shortened Dimorphos' orbital period by 33 ± 1 minutes [1], exceeding the mission's primary objective.

The momentum enhancement factor (β), measuring the momentum transferred via ejecta recoil, ranged between 2.2 and 4.9 [2], indicating that the momentum transfer was significantly greater than in a perfectly inelastic collision (β = 1). Without the ejecta effect, the orbital period reduction would have been approximately 7 minutes. This result aligns with pre-impact simulations.

Analysis of the ejecta distribution, modeled using data from the LICIACube spacecraft [3,4], revealed an ejecta cone with an elliptical base defined by half-angles of 69° and 51° [5]. This analysis highlights the critical role of ejecta in enhancing momentum transfer and confirms the effectiveness of kinetic impact for asteroid deflection.

Although the ejecta curtain can be approximated by a cone, the real physical structure is more complex than a cone, as evident from LUKE image data returned by LICIACube. As seen from Fig. 1, on both face-on and side-on views of the ejecta obtained during the flyby, it is clear that the ejecta curtain is a debris field  that is characterized by extended features originating from and around the impact location on Dimorphos. A critical understanding of the spatial distribution of these features in three dimensional space is required (a) to better understand how the momentum was distributed among escaping ejecta; (b) to constrain the ejection velocities of particles that are part of these structures; (3) to better estimate the mass of the ejecta, following radiative transfer calculations coupled with observed optical depths of features. As such, in this work we attempt to reconstruct the distribution of these features using LUKE data.

Figure 1: (a) A face-on view of the ejecta field during the approach phase of the flyby,  149 seconds after the impact, from a distance of 127.7 km from Dimorphos (b) an oblique view of the ejecta field during the final phase of the flyby, 189 seconds after the impact, from a distance of 144.3 km from Dimorphos.

Methods and Results: In order to accurately identify the extended features in both face-on and side-on/oblique images of the ejecta field, we first enhanced the images to highlight the details of the ejecta field using python scikit-image library [6]. Then we cropped the images to 400x400 px frames, centred on Dimorphos. These were then compiled to make short video clips thus enabling us to track the features in consecutive images.

Figure 2: Identified 12 features in both front, side and oblique viewing geometries.

Upon successful identification of 12 features (Fig. 2) , using the 3D animation software Blender (version 3.5), we trace out delimiters around each of the features in both face-on and side-on images. Next, we populate homogenous and evenly spaced cubic particles around Dimorphos inside a cube of 2km a side, and capture the particles that are inside the aforementioned delimiters of a given feature (Fig. 3) in face-on image (we use cubic particles of 50m per a side, comparable to the spatial resolutions of several LUKE pixels during the flyby). Next we switch to the simulation of the side-on view/ oblique view of the ejecta and populate the restricted particles from the previous step. Then, using the delimiters of the same feature in this different observing geometry we further restrict the volume in space that corresponds to the feature in question. Similarly we iterate over other features and obtain the 3D orientations and span of all the identified features. Transformations among different LUKE camera orientations are performed using NAIF/SPICE[7] package offered by Python3 package spiceypy[8].  

Figure 3: An example of an identified feature and how its delimitation is used to constrain the particles that are along the line of sight of this feature, in this case of a face-on image, superposed over a simulation of the observation. The origin of the coordinate system is at the centre of Dimorphos.

In Fig. 4, can be visualized the simulation of the ejecta field observation along with a LUKE image.

Figure 4. Left – LUKE image featuring the ejecta field observed with an oblique geometry as LICIACube was leaving the Didymos system following the closest approach. Right – Simulation of the ejecta field at the same time and geometry using the ejecta features retrieved using our method.

We then used images with longer exposure times to estimate the extends of ejecta features and thereby derive lower limits for ejecta velocities. Out of the features we used for this estimation, we found velocity ranges in the range of 30-70 m/s. We will present the final results and discuss the implications at the conference.

Acknowledgments: The LICIACube team acknowledges financial support from Agenzia Spaziale Italiana (ASI, contract No. 2019-31-HH.0 CUP F84I190012600).

References:

• [1] Thomas et al. (2023), Nature, 616, 448
• [2] Cheng et al. (2023), Nature, 616, 457
• [3] Dotto et al., (2011), PSJ, 199, 105185
• [4] Impresario et al., (2025), Acta Astronautica, 231, 223-234
• [5] Deshapriya et al. (2023), PSJ, 4:231
• [6] van der Walt et al. (2014), PeerJ 2:e453
• [7] Acton, C. H. (1996), P&SS, 44, 65
• [8] Annex, A., Pearson, B., Seignovert, B., et al. (2020), JOSS, 5, 2050