Lunar and Planetary Science XXX 1259.pdf A REASSESSMENT OF THE APOLLO LUNAR SEISMIC DATA AND THE LUNAR INTERIOR. A. Khan, K. Mosegaard, Dept. of Geophysics, University of , Juliane Maries Vej 30, DK-2100 Copenhagen O, ([email protected]),,K. L. Rasmussen, Radiocarbon Laboratory, National Museum of Denmark, Ny 11, DK-1471 Copenhagen K., Denmark..

Introduction. This study represents a reanalysis of the Specifically, our findings indicate a steady increase in

entire P -wave travel time set available [1, 2, 3] from the velocity from the surface down to the base of the crust, which

four-station seismic array placed on the lunar surface during has been estimated at a depth of 60 5 km, where the ve-

the Apollo missions. In comparison to earlier investigations locity reaches a value of 8.0 0.6 km/s. The results suggest [e.g. 2, 4, 5] where arrival times were inverted using linearised a gradual transition from the base of the crust to the upper methods, we have employed an inverse Monte Carlo sampling mantle without any clear discontinuities. The velocity in the algorithm to deal with the inverse problem [6, 7]. This method upper mantle is shown to remain roughly constant at a value is superior in dealing with non-linear inverse problems in that of 8.0 km/s down to its base at about 500 km depth. This no linearisations have to be introduced and since this particular result is taken as evidence for a homogeneous upper mantle. lunar inverse problem is highly non-linear [8], the investiga- A reestimation of the shallow moonquake source parameters tion undertaken here represents a more adequate approach, reveals these to be located in the depth range from 50 km to with the added benefit of providing more realistic error limits 220 km, providing further evidence to earlier stipulations [12, to the results. 13] for the uppermost mantle as a zone of seismic activity. The earlier studies [e.g. 4, 5, 9, 10, 11] led to the conclusion The data seem to suggest a velocity discontinuity of the that the Moon is a differentiated body comprised of a crust and order of 1.0 km/s just below the base of the upper mantle at

a mantle whose lower parts were thought be partially molten, a depth of 505 10 km, dividing the upper part of the mantle with detailed features however, being rather perfunctory. For from the middle one. A transition to a high velocity middle example, the model obtained by [2, 3] for the lunar man- mantle, interspersed with a low velocity layer, is indicated tle incorporates constant velocity zones with discontinuous by the data. A second velocity increase is palpable in the boundaries. However, these were introduced for computa- lower part of the middle mantle, with the velocity ultimately

tional convenience only and as noted [3] did not necessarily attaining a value of 11.01.2 km/s at 800 km depth roughly represent anything real. coinciding with the deep moonquake source region. In this Method of analysis. Inherent to the method is the use of part, comprising the depth range from 800 km to 960 km, the probability densities to delineate states of information which velocity is seen to be constant. A further reestimation of the in our case includes prior information on data and model deep moonquake source parameters has shown the majority parameters as well as information concerning the physical of these to be located in the depth range 850-960 km, with correlation between model and data. The posterior probability an abrupt decrease in the number of sources below 960 km. density, abbreviated ppd, obtained by combining prior infor- This apparent decrease coincides with a transition at roughly mation on data and model parameters with the information 960 km depth. The sharp decrease in the number of events regarding the physical correlation via the Bayesian method, in the lower middle mantle might possibly reflect the onset contains all the information about our system. Due to the fact of partial melt, as mentioned in an earlier study [9], and it that this ppd contains several extrema, where the global max- could be speculated that the deep moonquakes are caused by imum represents the most likely solution and a large number shear movement due to tidal deformations of the rigid part of secondary maxima may represent other possible solutions, of the mantle where it is least supported. Now, given the we employ a Monte Carlo method to explore the ppd.Dueto results presented here of a homogeneous upper mantle and its random nature the Monte Carlo method does not have the a more complicated middle mantle velocity structure as well

tendency to get trapped in secondary maxima leading to an as a prominent transition at 50510 km depth, it could be enhanced examination of the model space, whereby the true suggested that the initial melting of the Moon extended down model variability is exposed. This being an iterative process to the dividing line between the upper and the middle man- we end up with many thousand models each one displaying a tle, leaving the lower parts primitive in composition. This possible lunar velocity structure. These are then analysed in however, does not rule out the existence of partial melt in the terms of probabilities. lower mantle, as contemplated by [9]. On the other hand the Results. Our findings indicate the partitioning of the nature of the high velocity middle mantle, notably the deep Moon into moonquake source region, could lead to the speculation that an initial differentiation extended to depths of at least 1100

 a crust roughly 60 km thick, km, in line with an earlier argument [3]. If this has been the case the boundary between the upper and the middle mantle

 an upper mantle comprising the depth range from 60 simply represents a transition to a high velocity phase. km to just below 500 km and The results presented in this study do to some degree corrobo- rate the previous findings, but most importantly, they provide  a middle mantle constituting the depth range from

below 500 km to about 1100 km 1 . a more detailed picture of the deep lunar interior.

1

The depth to which our analysis extends. A Reassessment of the Apollo Lunar Seismic Data and the Lunar Interior: A. Khan et al. Lunar and Planetary Science XXX 1259.pdf References. [1] Nakamura Y. et al., Structure of the Lunar A., A Selenological Enquiry - Elucidating the Interior of Mantle, J. Geophys. Res., 81, 4818 (1976). [2] Nakamura Y., the Moon using Non-Linear Seismic Tomography and an G. Latham & J. Dorman, Apollo Lunar Seismic Experiment- Inverse Monte Carlo Samping Method, M.S. Thesis, Odense Final Summary, Proc. Lunar Planet. Sci. Conf. 13th., University (1998). [9] Nakamura Y. et al.,NewSeismic 117 (1982). [3] Nakamura Y., Seismic velocity structure of Data on the State of the Deep Lunar Interior, Science, 181, the Lunar Mantle, J. Geophys. Res., 88, 677 (1983). [4] 49 (1973). [10] Dainty A., M. Toksoz & S. Stein, Seismic Toksoz M., A. Dainty, S. Solomon & K. Anderson, Structure investigations of the lunar interior, Proc. Lunar Planet. Sci. of the Moon, Rev. Geophys. Space Phys., 12, 539 (1974). Conf. 7th., 3057 (1976). [11] Goins N., M. Toksoz & A. [5] Goins N., A. Dainty & M. Toksoz, Lunar Seismology: Dainty, he lunar interior: a summary report, Proc. Lunar The Internal Structure of the Moon, J. Geophys. Res., 86, Planet. Sci. Conf. 10th., 2421 (1979). [12] Nakamura Y. 5061 (1981). [6] Mosegaard K. & A. Tarantola, Monte et al., High-frequency lunar teleseismic events, Proc. Lunar Carlo Sampling of Solutions to Inverse Problems, J. Geophys. Planet. Sci. Conf. 5th., 2883 (1974). [13] Nakamura Y., HFT Res., 100, 12431 (1995). [7] Mosegaard K., Resolution events: shallow moonquakes ?, Phys. Earth Planet. Inter., 14, analysis of general inverse problems through inverse Monte 217 (1977). Carlo sampling, Inverse Problems, 14, 405 (1998). [8] Khan