VOL. 12, NO. 4 REVIEWS OF GEOPHYSICS AND SPACE PHYSICS NOVEMBER 1974 Structure of the Moon M. NAFI TOKSOZ, ANTON M. DAINTY, $EAN C. SOLOMON, AND KENNETHR. ANDERSON Department of Earth and PlanetarySciences, Massachusetts Instituteof Technology Cambridge, Massachusetts 02139 Seismic data from the four stations of the Apollo passive seismic rietworkhave beenanalyzed to obtain the velocitystructure of the moon. The long reverberating train of seismic energyobserved in lunar seismograms may be explained by scattering in a near-surface zone.The earliest partsof the seismogram correspond to bodywave phases and may be interpreted by conventional methods. Thereare no clearly identifiable dispersed surface wavetrainspresent on lunar seismograms. This absence canbe explained by scattering of surface waves by near-surface heterogeneities. Analysis of body wave phases from artificial impacts of known impact time and position yieldsa crustalsection. In the Mare Cognitum regionthe crustis about 60 km thick and is layered. In the 20-km-thickupper layer, velocity gradients are high and microcracks may play an important role. The 40-km-thicklower layer has a nearly constant 6.8-km/s velocity. There may be a thin high-velocity layer presentbeneaththe crust. The determinationof seismic velocities in the lunar mantle is attemptedby usingnatural impacts and deepmoonquakes. The simplest modelthat can be proposed for the mantle consists of a 'lithosphere' overlying an 'asthenosphere.' Shear waves are attenuated in the 'asthenosphere,' and this zone may be partially molten. No seismic data con- tradict this model. Thebest value forthecompressional wave velocity of thelithosphere is8.0--, 8.3km/s, but this may be an average over several differingregions. Density models are calculated for the moon by using the latest valuefor the moment of inertia(C/MR •' = 0.395 + 0.005), meandensity, and tem- perature and pressure dependence of density for likely lunar models. The meandensity in the lunar man- tle, corrected to standard (surface temperature, zero pressure) conditions, is 3.4-3.5 g/cma. Thesevalues togetherwith the mean compressional wave velocities are consistent with an olivine-pyroxene (olivine- rich) lunar mantle but do not excludeother compositions. On the basisof the density models,some limitations can be placedon the maximum allowable radiusof an iron-rich lunar core. If the lunar mantle is chemically and mineralogically homogeneous, the maximum radiusof such a core is about 700 km for an FeS composition and about.450 km for pure Fe composition. There are no geophysical data that in- dicate whether the moon does or does not have an iron-rich core. INTRODUCTION A four station seismic network has been established on the moon with the deployment and long term operation of seis- mometers by the Apollo 12, 14, 15, and 16 missions. A large number of moonquakes and meteoroid impactshave been re- cordedby one or more stations of the network. In addition to these, there have been nine large artificial impactsand natural events (Saturn third-stageboosterrocket (S4B) or ascent stage of the lunar module (LM)). Seismograms from theseartificial impacts and natural eventshave been analyzed to determine the velocitystructure and some physical properties of the lunar interior. Also, by using the seismic constraints on the shallow structureand the improved accuracy of the lunar momentsof inertia [Williams et al., 1973], density models can be cal- culatedfor the lunar interior. In this paper we treat the prob- lem of lunar structure using both the seismic data and the den- sity models. The latter play an especially important role in puttingsome limitson the maximum allowable size of an iron- rich core. We start with a review of the seismic data. First we describe brieflythe general characteristics of lunar seismograms and in- terpret thesein terms of a scattering model. Second,usingthe distinct phases and early portions of the artificial impact rec- ordswe determine a velocity model for the outer 100km of the moon, using both travel timesand synthetic seismograms. The properties of the deepinterior of the moon as can be specified from seismograms of distant natural eventsare discussed next and followed by a section on the compositional and physical implicationsof the velocities. The densitymodelsare covered in the last section; velocity structureand possible lunar tem- perature profiles are incorporated in the calculations. Copyright(D 1974 by the AmericanGeophysical Union. Although we attempt to present a comprehensivedis- cussion on the subject of lunar seismology and structure in this paper, we cannot cover many topicswith sufficient detail and must refer the readerto many specialized articles[Dainty et al., 1974a; Duennebierand Sutton, 1974a, b; Lammlein et al., 1974; Latham et al., 1971b, 1972b, 1973, 1974; Solomon, 1974; Toksbz and Solomon, 1973; Toksbz et al., 1972a, b, c]. THE SEISMICNETWORK, ARTIFICIAL IMPACTS, AND LUNAR SEISMOGRAMS Seismicnetwork. The Apollo passive seismic experiment (PSE) network consists of four stations, one eachdeployed at the Alsep 12, 14, 15, and 16 sites on the moon (the Apollo 11 seismometer, which was powered by solar cells, operated for one lunation and did not becomepart of the network). Co- ordinatesand deploymentdates of thesestationsare listed in Table 1 and the selenographic distributionis shownin Figure 1. Each seismic package consists of three long-period (LP) components and one short-period (SP) vertical sensor. Detailed descriptions of these instruments are givenby Latham et al. [1970b, 1971a, 1972a, b]. As shown by the response curves in Figure 2, the LP seismometers can operate either in a flat mode or a highly peakedmode. Sincethe beginning of the experiment, most data have been obtained in the peaked mode. All seismometers have functioned well except for the SP seismometer at station 12 (inoperative at deployment)and the LP vertical at station 14 (inoperative as of January 1972). Lunar seismograms. Seismograms from moonquakes, meteoroidimpacts, and nine artificial impacts are availablefor use. Among the most important data used in this study are seismograms resultingfrom seven controlled impacts.The im- pactingbodies are the Saturn third-stage booster (S4B) and the 539