Meta Research Bulletin (ISSN 1086-6590, USA)
December 15, 1999
Vol. 8, No. 4, pp.49-54.

Alexey V.Arkhipov
Institute of Radio Astronomy, 4 Krasnoznamennaya, Kharkov, 310002, Ukraine


The Moon is an indicator of possible alien visits to the Earth during past ~4 billion years. New computer algorithms are proposed and tested for the archaeological reconnaissance of our satellite. About  20,000 Clementine lunar orbital lunar images have been processed, and  a few ruin-like formations were found. According to a fractal analysis, some of these finds are different from the lunar surface on which they reside, and formally resemble terrestrial archaeological objects. At the least, the catalogued formations should be interesting as geological anomalies.


          As it is argued [1,2] the Moon could be used as an indicator of  extraterrestrial intelligence visits to the Solar System. Therefore, it is necessary to ascertain  the indicator's condition REGARDLESS OF THE RESULT. Unfortunately, such studies are outside of the professional activity of selenologists (because of their orientation only to natural formations and processes) as well as archaeologists (because archaeology adheres to a pre-Copernican geocentric position). That is why the first archaeological reconnaissance of the Moon is realized only as a private project, SAAM: Search for Alien Artefacts on the Moon.
          The SAAM Project  is developed since 1992. The justifications of lunar SETI, the wording of specific principles of lunar archaeology and the search for promising areas on  the  Moon were the first stage of the project (1992-95). Already  obtained results of lunar exploration (e.g. [1]) show that the search for alien artefacts on the Moon is a  promising  SETI-strategy, especially in the context of  lunar colonization plans. The aim of the second SAAM  stage  is  the search  for promising targets of lunar archaeological reconnaissance. The most probable to detect would be very ancient (age ~1-4 Gyr) analogies of proposed modern  lunar  bases.  Such long-term buildings should be under the lunar surface  for  protection from ionizing radiation and meteorites. Such super-ancient constructions could be eroded and detected as systems of low ridges and depressions, covered by regolith and craters.


          It is more reasonable to use for SAAM the archaeological method (e.g. preliminary assumption of artifact existence) than the planetological "presumption of naturality". According to Dr. B.V.Andrianov (the Russian authority in aerial archaeology): "The main demasking sign of objects, whose origin on the terrain is due to human activity, is their geometric regular configuration (at rare exclusions)" [3]. Terrestrial  buildings,  as  a  rule,  have  rectangular outlines. Hence, it is reasonable  to  search  on  the lunar images for unusual  patterns of  rectangular  shape. The status of  such finds can not be higher than that artifact candidates. The true nature of the finds cannot be ascertained by the remote sensing only. According to archaeological practice, direct exploration (e.g. excavations) is an obligatory element of  the search. Hence, the finds of the SAAM Project cannot be discoveries; but SAAM is a precursor for inevitable archaeological reconnaissance of the Moon.
          The lunar Clementine EDR Image Archive on CDs [4] was used for SAAM. The following  tests were proposed and used for the analysis of high-resolution camera (HIRES) data of the Clementine spacecraft.


          As a rule, natural landscapes consist of self-similar details on various dimension scales. For example, lunar craters are similar at diameter of 0.1m to 104 m. By contrast, artificial constructions have some typical dimensions caused by size of their constructors. Hence, the artifacts might be recognized as details of unusually frequent size. The search for such dimension anomalies is an essence of the fractal method proposed by M.J.Carlotto and M.C.Stein [5]. Unfortunately that test is too slow for the express analysis of ~80,000  HIRES images because of so many calculations. That is why the simpler and faster algorithm is proposed here.
          For this purpose, the probability distribution (M) of distances ( r ) between minima of  brightness along the lines of an image is constructed. In fact, M(r) is the distribution of the image detail's size. At long scales, this function could be approximated  by the fractal power law: M(r) ~ rn. As constructions have some typical size, the artifacts should increase the squared residuals of linear regression: log M(r)= n log r + C; where C is a constant. According to empirical results, M(r) of the HIRES images could be approximated by the power law  at rò5 pixels. That is why the regression is calculated in the interval of 5 ó ri ó 30 pixels (50m ó ri ó 900m) or up to 26 points.
          In each of twelve test squares of 96x96 pixels of the image, the computer calculates the best n and C by least squares technique and the average of  the squared residuals:

nmax sk2 = (gk /nmax) S [ log M(ri) - n log ri - C]2, i=1

where: k is the number of test square; gk is the apparatus factor or average (s*/sk)2 from a lot of HIRES images (s* is sk in the center of image at gk=1); nmax is the number of used scales up to M(ri)=0. Then the average dispersion <s> is estimated from these regional squared residuals.
          The analysis of 733  HIRES images (0.75mm-filter; the polar zones up to 75o-latitudes; 112-115 orbits) shows that  s distribution is  the classical Gaussian function. According to the Student's criterion for 12 estimations, if the inequality (sk -<s>) > 1.796 ( S(sk -<s>)2/11)1/2 is true in any test square, this area could be considered as anomalous with a probability of 0.95.


          The rectangular test reveals the rectangular patterns of lineaments on the image of  lunar surface. For each pixel of the image, the probe point at  the distance of 6 pixels and position angle j is selected. Let N be the total number of such pairs, and n is the number of  pixel pairs, where bright accounts are equal. The function W(j)=n/N characterizes the anisotropy of the image. For the correction of the camera aberration, W(j) was divided by its average quantity, which is calculated for many images at same j. The computer finds maxima of the smoothed W(j) and the corresponding  jm angles. Obviously  jm  describes the orientations of lineament groups. If there is (90oñ10o)-differences between jm , the image is classified as interesting.


          For the false alarm selection, the SAAM-transformation of the image was used for revealing indiscernible details of the lunar surface. This algorithm is very simple: The image is smoothed by the sliding window in a kind of circle with radius R, then the result of this procedure is subtracted from the initial image. Thus the pixels, which are brighter than the smoothed level, are considered as "white", and others are considered "black". This clipping permits us to see details of extremely low contrast as well as the high contrast features. Moreover, the large details (>R) of the image appear damped, and they do not interfere with  small-sized objects.


          J. Fiebag [6] supposed that the parallelism of the formation with lineaments of its surroundings is the criterion for naturality of the object. Although the human activity correlates with geological lineaments (e.g. rivers), the conservative Fiebag test was applied to the lunar finds.
          The lineament orientation of surroundings was estimated by the described rectangular test technique for the corresponding large-scale image from the ultraviolet-visible (UVVIS) camera. The UVVIS image cover 196 times the HIRES images' area with the same 0.75mm-filter. Only W(j)-peaks with statistical significance of  >0.9 were taken into account. If one of  the two directions of  the rectangular formation on a HIRES-image coincides (±10o) with any significant UVVIS direction, the object is not considered as interesting. This test rejects about 60% of finds.


          The polar zones of ñ75o to ñ90o latitudes are most suitable for the SAAM because of oblique lighting. For the preliminary archaeological search of those zones, 20 CDs  were selected randomly from the Clementine EDR Image Archive [4]. About 20,000 files or ~25% of the polar HIRES data were analyzed. Only 32 images were selected as interesting after geological test.
          There are three types of the finds:
          (a)  Quasi-rectangular lattices of  leneaments;
          (b)  Quasi-simmetrical, quasirectangular patterns of depressions;
          (c) narrow and shallow depressions with smoothed bottom of quasi-simmetrical and quasi-rectangular outlines;



Figure 1
Arrowed rectangular 800x800m pattern on the hill is an example of lunar ruin-like formations (long.=301.11 deg.; lat.=85.59 deg.; Clementine image: LHD6749R.318).

An example of  picturesque ruin-like formations on a hill is shown in Fig. 1. The traditional explanations in terms of crossing of impact fault systems seem inadequate for such compact and closed formations. The Moon did not have  conditions (a thin crust above melted mantle) for Venus-like tessera terrains. So the origin of these anomalies is problematical. As a rule, lunar base projects would be expected to show the rectangular patterns of subsurface constructions [7-9]. Formally, such complexes could be classified as (a) and (b) patterns.  The (c)-type bands in Fig. 2 are a puzzle.  Theirs depth from shadows (~10 m) is about the average thickness of the regolith layer on the Moon. Theirs flat bottoms and geometry remind one of modern projects for lunar regolith mining (e.g. [10]). Some depressions of (b)-type could be interpreted in mining terms too.


Figure 2
 The curious shallow depressions of ~8m-depth  and ~100m-width can be seen in the box after filtration of the image's fine structure, and again in the schematic at the top left (long.=28.31 deg.; lat.=79.11; Clementine image: LHD5502Q.290).

Of course, this visual impression  should be tested by some objective procedure. The modified fractal Carlotto-Stein method  was used for this purpose. First, the range of HIRES image brightness was increased linearly up to 256 gradations. Then convert the image into an intensity surface in a 3-D rectangular frame of coordinates (x and y are the pixel coordinates; z is its brightness). The Carlotto-Stein method [5] can be thought of as enclosing the image intensity surface in volume elements. These volume elements are cubes with a side of  2r; where r is the scale in terms of pixel coordinates or its brightness. Let Vr be the average minimal  volume of such elements enclosing an image intensity surface at some point. Then the surface area is Ar = Vr/2r. As a function of scale, Ar characterizes the size distribution of image details.  The fractal linear relation between log Ar and log r is a good approximation for natural landscapes. However, the self-similar fractals do not approximate artificial objects as a rule. That is why M.J. Carlotto and M.C. Stein used the average of the squared residuals e of the linear regression log Ar=blog r + g   as  a measure of artificiality.
          Unfortunately, e depends on the number of pixels in an image. Therefore, it is difficult to compare different images. Moreover, the shadows increase e and generate false alarms. These problems could be resolved by the non-linear regression:

log Ar = a (log r)2 +blog r + g,

          where the factor a is independent of  the image size. The shadows lead to a >0, but artificial objects have a <0.

Figure 3
The diagram of fractal properties of analyzed images: the random set of HIRES files (crosses), HIRES images of ruin-like formations (black squares), and aerospace photographs of terrestrial archaeological objects (opened squares).

This effect is shown in Fig. 3. There factors a and b are calculated for the random set of  HIRES images (crosses) and aerospace photographs of terrestrial archaeological objects (white squares). The fragments of images of the following archaeological sites were used in our analysis: Giza tombs in Egypt (KVR-1000 satellite) and El-Lejjun Roman legionary fortress, Jordan, (CORONA satellite) [11]; the Cerro Vidal trinchera , the Cerro Juanaquena trinchera and Pueblo She' in Galisteo Basin (New Mexico, aerial photographs [12]). The parameter a values for lunar ruin-like formations (black squares) is distributed between the geological background (crosses) and archaeological objects (opened squares). Some formations have a as low as the known archaeological sites.


Figure 4
The shadow effect for the parameter a of geological background (crosses) and ruin-like formations (black squares) on the Moon. The regression relating a of the random image set and zenith angle of the sun (Zsol) is shown as the dashed line. The adopted criterion for  target selection (regression - 3sa) is shown as the solid line.

The weak effect of  low sunlight could be seen in Fig. 4. At any zenith angle of the Sun (Zsol), the ruin-like formations have systematically lower a than the random set of HIRES images does. The average linear regression relating a of the random set and Zsol is shown as a dashed line. The standard deviation of the crosses from this regression is sa =0.0113. A minimal deviation of  3sa (solid line) is adopted as a formal criterion for the final selection. The selected objects on the Moon listed in Table I all have reasonable levels of archaeological interest.


          It is shown that computerized archaeological reconnaissance of the Moon is practicable. The proposed methods can be used for more extensive lunar survey and for planetary SETI in general.
          Formally, there are the ruin-like formations on the Moon. Of course, the ruin-like objects could be geological formations, but they could be archaeological objects too. This second possibility is so important, that it should not be ignored a priori. Thus, any hill is natural for geologist, but an archaeologist has the right to suspect the hill as a tumulus or ancient settlement. Only direct exploration of the Moon can decide between artificial or natural origin of these unusual lunar formations. Obviously ruin-like formations are interesting as geological anomalies, at the very least.


          The author is very grateful to Dr. Y.G. Shkuratov for access to the Clementine's CDs. I also thank Dr. F.G. Graham, Dr. J. Fiebag, Dr. T. Van Flandern, Dr. L.N. Litvinenko and Dr. J. Strange for discussions and support.


          1. A.V. Arkhipov and F.G. Graham, "Lunar SETI: A Justification", in The Search for Extraterrestrial Intelligence (SETI) in the Optical Spectrum II, ed. S.A. Kingsley & G.A. Lemarchand, SPIE Proceedings, Vol. 2704, SPIE, Washington, 150-154, 1996.
          2. A.V. Arkhipov, "Earth-Moon System as a Collector of Alien Artefacts", J. Brit. Interplanet. Soc., 51, 181-184 (1998).
          3. B.V. Andrianov, Ancient irrigation systems of Aral Region. Nauka Publishing House, Moscow, 1969, p. 29.
          4. DoD/NASA, "Mission to the Moon", Deep Space Program Science Experiment, Clementine EDR Image Archive. Vol. 1-88. Planetary Data System & Naval Research Laboratory,  Pasadena, 1995 (CDs).
          5. M.J. Carlotto, M.C. Stein, "A Method for Searching for Artificial Objects on Planetary Surfaces", J. Brit. Interplanet. Soc., 43, 209-216, (1990).
          6. J. Fiebag, Analyse tektonischer Richtungsmuster auf dem Mars. Kein Hinweise auf knstliche Strukturen in der sdlichen Cydonia-Region // Astronautik, Heft 1, 9-13, 47-48 (1990).
          7. T.L. Stroup, "Lunar Bases of the 20th Century: What Might Have Been", J. Brit. Interplanet. Soc., 48, 3-10 (1995).
          8. S. Matsumoto, T. Yoshida, K. Takagi, R.J. Sirko, M.B. Renton, J.W. McKee, "Lunar Base System Design", J. Brit. Interplanet. Soc., 48, 11-14 (1995).
          9. W.Z. Sadeh and M.E. Criswell, "Inflatable Structures for a Lunar Base", J. Brit. Interplanet. Soc., 48, 33-38 (1995).
         10. J. Sved, G.L. Kulcinski, G.H. Miley, "A Commercial Lunar Helium 3 Fusion Power Infrastructure", J. Brit. Interplanet. Soc., 48, 55-61 (1995).
         11. M.J.F. Fowler, Examples of Satellite Images in Archaeological Application // URL:
        12. J. Roney, Cerro de Trinchera Archeological Sites // The Aerial Archaeology.  Newsletter. Vol. 1, No. 1, 1998 /
URL: and She_in_shadow.html

Table I. Selected Ruin-Like Formations of the Moon

Longitude      Latitude    Type    Dimensions            Image                       Description
   (deg.)           (deg.)                       (km)
________     _______    ____     _________     _____________      ____________________

  28.04           -76.45          a          5.3 x 5.6        LHD0132B.290      separate group of
                                                                                                            rectangular walls and
                                                                                                            qadrangular hills

  28.31            79.11          c          1.2 x 1.5         LHD5502Q.290      curious pattern of linear
                                                                                                             and broken band
                                                                                                             depressions of ~100m-
                                                                                                             width and  ~8m-depth

  31.06           78.84          c            0.3 x 1.3        LHD5256Q.293       rectangular zigzag band
                                                                                                             of flat depression of

 151.21         -76.24         b             0.8 x 0.8       LHD0470B.112       rectangular claster of

 246.08          81.88         a              2.2 x 2.2       LHD7638R.343       rectangular walls of
                                                                                                            100m-width  and the
                                                                                                             box-like hill of

 301.11         85.59        a-b            0.8 x 0.8       LHD6749R.318        complicated
                                                                                                             rectangular  structure
                                                                                                             on the top of a hill (Fig. 1)

(This web page produced for Alexey Arkhipov by Francis Ridge of  The Lunascan Project)
Home Page