Unveiling the Extent of Lunar Subsurface KREEP Layer from Geodynamic Modeling

UNVEILING THE EXTENT OF LUNAR SUBSURFACE KREEP LAYER FROM GEODYNAMIC MODELING. S. Santangelo1, A.-C. Plesa1, A. Broquet1, D. Breuer1, M. Grott1, 1German Aerospace Centre (DLR), Berlin, Germany ([email protected]). Introduction: An asymmetric distribution of subsurface heat producing elements (HPE, i.e. Th, U, and K) has been suggested to explain the different thermochemical evolutions of the lunar nearside and farside [1]. However, lunar differentiation models predict a globally uniform KREEP layer (Potassium, Rare-Earth Elements, Phosphorus + Th and U) [2]. The timeline and nature of a potential redistribution of HPE, during or after the lunar magma ocean (LMO) solidification, remains highly debated [2]. Surface heat flux measurements are key indicators of a planet’s subsurface thermal state, serving as primary constraints for thermal evolution models. In-situ heat flux measurements by Apollo 15 and Apollo 17 crews provided values higher than the Moon’s proposed surface average [3]. Conversely, remote sensing estimates close to the south pole retrieved much lower heat fluxes [4]. Here, we use the most recent global crustal thickness map derived from gravity and topography data [5] and new constraints on the abundance of HPE from LMO crystallization studies [6] to model lunar evolution. We investigate the lateral distribution and concentration of HPE, testing various extents and enrichments of a putative KREEP-rich anomaly formed during or after LMO crystallization [2]. Predicted present-day surface heat flux is then compared to observations. Methods: We model the thermal evolution of the Moon in a 3D spherical shell geometry, using the mantle convection code Gaia [7]. The code solves the conservation equations of mass, linear momentum and thermal energy, assuming a homogeneous mantle with purely Newtonian rheology and negligible inertia. First, we use a simplified model setup, similar to [1], to assess whether a global, hemispherical, or regional KREEP anomaly is capable of producing surface heat flux values comparable to Apollo 15 and Apollo 17 measurements on the lunar nearside. This simplified model consists of three layers: a homogeneous mantle, a crust of 39 km thickness and a thin KREEP unit between crust and mantle. The lateral extent of the KREEP layer is varied between global, hemispherical (i.e. nearside), and regional (i.e., 1300 km radius assuming circular geometry). 1D magma ocean crystallization models, similar to [8] and including the results in [6], produce a 1.6 km thick KREEP layer with Th concentration of 18.6 ppm, from which we compute U and K concentrations with ratios Th/U=3.7 and K/U=2500. The resulting four parameters (layer thickness and Th, U, K concentration) are used to model a global KREEP layer, simulating the product of a symmetric crystallization of the magma ocean. Additionally, we redistribute the volume of KREEP material into different configurations to produce the models in Fig. 1. For a hemispherical and regional KREEP layer, the concentration of HPE in the layer is varied to simulate all possible scenarios from a complete overturn and KREEP remixing (100% of KREEP in the mantle, no KREEP layer) to a perfectly efficient KREEP concentration underneath the nearside (0% of KREEP in the mantle, Fig. 1). For all scenarios, we compare the predicted heat flux to the Apollo measurements and discard inconsistent models . Next, we use a more sophisticated model, similar to [9], to investigate the local heat flux variability at the Apollo 15 and 17 landing sites, and in regions with remote-sensing heat flux estimates near the south pole. This second model employs a laterally variable crustal thickness [3], and a laterally variable KREEP layer thickness that considers basin excavation. In this second setup, we vary the lateral extent of the KREEP anomaly relative to the measurements. First, we assume all three locations of interest (Apollo 15 and 17 landing sites, and nearside south polar region) to be underlain by a hemispherical KREEP layer (Fig. 2). Additional models include a KREEP layer only beneath the two Apollo sites (large KREEP size model), only beneath Apollo 15 (medium KREEP size), and far from all three locations (small KREEP size). For each of these models, we also vary the HPE concentration of the crust and KREEP layer, and the lunar bulk U abundance, and we select models that can reproduce currently available lunar heat flux values (Fig. 2). Figure 1: Summary of KREEP distributions following LMO crystallization and potential overturn/redistribution of heat sources. Each sketch represents a possible configuration for the subsurface structure of the Moon. Best-fit models that can reproduce Apollo 15 and 17 heat fluxes are shown on a green background. Figure 2: Summary of model simulations varying KREEP layer lateral extent versus bulk U concentration (a, 30 ppm Th) and KREEP Th concentration (b, 16 ppb bulk U), for a Th/U ratio of 3.7 and K/U ratio of 2500. Each pie chart represents an individual simulation, and each sector a specific surface location: Apollo 15 (Ap15), Apollo 17 (Ap17), South polar region/Region 5 (R5). The best-fit and preferred model is indicated by a dashed-line border. All heat fluxes are computed at present day. Results and discussion: Constant crustal thickness setup. Using the simpler model setup with constant crustal thickness we are able to exclude end-member KREEP distribution scenarios (Fig. 1). The scenario of complete KREEP overturn/remixing (blue tiles in Fig. 1) shows too low heat flux values compared to the highest value measured (Apollo 15). The same applies to the case of a pristine global KREEP layer (topmost sketch in Fig. 1). These two configurations require bulk U concentration in excess of 25 ppm to reach heat flux values as high as Apollo 15, which is inconsistent with the bulk composition proposed in [10]. For a regional KREEP layer (~1300 km radius), we find that at least 60% of KREEP material is required to be remixed in the mantle to match Apollo 15 and 17 heat flux. Conversely, a hemispherical KREEP layer requires <60% of KREEP to be mixed in the mantle, in order to reach heat fluxes as high as Apollo 15. Therefore, our results suggest that some degree of HPE sequestering on the nearside is likely, but it may not have been an efficient process. Up to 60-80% of KREEP material from magma ocean crystallisation could have remained well-mixed in the mantle after overturn. Variable crustal thickness setup. Using the more complex model setup, we find that surface heat flux at all locations increases with increasing bulk HPE (Fig. 2a). Increasing the KREEP HPE concentration increases the surface heat flux within the KREEP region and decreases elsewhere (Fig. 2b), with the global surface average remaining constant. If the remote-sensing estimate of a measurably lower heat flux in the south polar region is considered accurate [4], then our models with a variable crust and KREEP thickness show that only a large, low-HPE-enriched KREEP beneath the Apollo 15 & 17 landing sites is able to reproduce the observation (~1600 km radius, 30 ppm Th concentration and 16-20 ppm bulk U). This result implies KREEP extending beneath Mare Serenitatis. However, if the south polar estimate is considered inaccurate (order of 100% error or higher), we find consistent scenarios also for a hemispherical or medium KREEP (~1200 km radius), with south polar heat flux being comparable to the surface average (12-14 mW/m2). The upcoming measurement by the LISTER instrument onboard Blue Ghost lander [11] will provide a heat flux value located sufficiently far from Oceanus Procellarum. Including this value in our model will allow us to put strong constraints on KREEP size and enrichment, along with bulk U concentration. References: [1] Laneuville M. et al. (2013). [2] Moriarty D. P. et al. (2021). [3] Warren P. H. and Rasmussen K. L. (1997). [4] Wei G. et al. (2023). [5] Broquet A. and Andrews-Hanna J. C. (2023). [6] Haupt C. P. et al. (2024). [7] Hüttig C. et al. (2013). [8] Schwinger S. and Breuer D (2021). [9] Plesa A.-C. et al. (2016). [10] Taylor G. J. and Wieczorek M. A. (2014). [11] Nagihara, S. et al. (2023).