Detection of sporadic impact flashes on the Moon: Implications for the luminous efficiency of hypervelocity impacts and derived terrestrial impact rates

Abstract

We present the first redundant detection of sporadic impact flashes on the Moon from a systematic survey performed between 2001 and 2004. Our wide-field lunar monitoring allows us to estimate the impact rate of large meteoroids on the Moon as a function of the luminous energy received on Earth. It also shows that some historical well-documented mysterious lunar events fit in a clear impact context. Using these data and traditional values of the luminous efficiency for this kind of event we obtain that the impact rate on Earth of large meteoroids (0.1–10 m) would be at least one order of magnitude larger than currently thought. This discrepancy indicates that the luminous efficiency of the hypervelocity impacts is higher than 10−2, much larger than the common belief, or the latest impact fluxes are somewhat too low, or, most likely, a combination of both. Our nominal analysis implies that on Earth, collisions of bodies with masses larger than 1 kg can be as frequent as 80,000 per year and blasts larger than 15-kton could be as frequent as one per year, but this is highly dependent on the exact choice of the luminous efficiency value. As a direct application of our results, we expect that the impact flash of the SMART-1 spacecraft should be detectable from Earth with medium-sized telescopes.

Full text

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.1 (1-8)
Icarus ••• (••••)•••–•••
www.elsevier.com/locate/icarus
Detection of sporadic impact flashes on the Moon:
Implications for the luminous efficiency of hypervelocity impacts and
derived terrestrial impact rates
J.L. Ortiz a,∗,F.J.Aceitunoa, J.A. Quesada b, J. Aceituno c, M. Fernández a, P. Santos-Sanz a,
J.M. Trigo-Rodríguez d,e, J. Llorca e,f, F.J. Martín-Torres g, P. Montañés-Rodríguez h, E. Pallé h
aInstituto de Astrofísica de Andalucía, CSIC, Apt. 3004, Camino Bajo de Huetor 50, 18080 Granada, Spain
bHuétor Santillán Observatory, Avda Puente Nuevo 27, 18183 Huétor-Santillán, Granada, Spain
cCentro Astronómico Hispano-Alemán de Calar Alto, Almería, Apt. 04004, Almería, Spain
dInstitut de Ciències de l’Espai (CSIC) Campus UAB, Facultat de Ciencies, 08193 Bellaterra (Barcelona), Spain
eInstitut d’Estudis Espacials de Catalunya (IEEC), Campus UAB, Facultat de Ciencies, 08193 Bellaterra (Barcelona), Spain
fInstitut de Tècniques Energètiques, Univ. Politècnica de Catalunya, 08028 Barcelona, Spain
gAnalytical Services and Materials, Inc., 107 Research Drive Hampton, VA 23666, USA
hBig Bear Solar Observatory, New Jersey Institute of Technology, Big Bear City, CA 92314-9672, USA
Received 24 February 2006; revised 27 April 2006
Abstract
We present the first redundant detection of sporadic impact flashes on the Moon from a systematic survey performed between 2001 and 2004.
Our wide-field lunar monitoring allows us to estimate the impact rate of large meteoroids on the Moon as a function of the luminous energy
received on Earth. It also shows that some historical well-documented mysterious lunar events fit in a clear impact context. Using these data and
traditional values of the luminous efficiency for this kind of event we obtain that the impact rate on Earth of large meteoroids (0.1–10 m) would be
at least one order of magnitude larger than currently thought. This discrepancy indicates that the luminous efficiency of the hypervelocity impacts
is higher than 10−2, much larger than the common belief, or the latest impact fluxes are somewhat too low, or, most likely, a combination of both.
Our nominal analysis implies that on Earth, collisions of bodies with masses larger than 1 kg can be as frequent as 80,000 per year and blasts
larger than 15-kton could be as frequent as one per year, but this is highly dependent on the exact choice of the luminous efficiency value. As a
direct application of our results, we expect that the impact flash of the SMART-1 spacecraft should be detectable from Earth with medium-sized
telescopes.
©2006 Elsevier Inc. All rights reserved.
Keywords: Impact processes; Moon; Collisional physics; Cratering
1. Introduction
Several techniques have been used in the past to estimate the
flux of incoming bodies to the Earth. Each technique is suitable
to a size range of the impactors and no single technique is valid
for the whole range of sizes. Therefore, results from different
techniques have to be linked. The flux of large meteoroids (in
*Corresponding author. Fax: +34 958 814530.
E-mail address: ortiz@iaa.es (J.L. Ortiz).
the 0.1–1 m diameter range) has traditionally been estimated
from the records of fireballs detected by all-sky camera net-
works (e.g., Ceplecha, 1988; Halliday et al., 1996), whereas
the flux of larger meteoroidal bodies (with diameters between
1–10 m) has recently been determined from a combination of
military satellite and infrasound data (Brown et al., 2002). On
the other hand, the flux of asteroidal bodies larger than 10 m has
been inferred from forward orbital integrations of the discov-
ered near-Earth objects by different telescopic surveys. A time-
averaged flux can also be inferred from the counts of lunar
craters using scaling laws to relate crater size to impactor diam-
0019-1035/$ – see front matter ©2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.icarus.2006.05.002

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.2 (1-8)
2J.L. Ortiz et al. / Icarus ••• (••••)•••–•••
eter. Crater counting techniques can also be applied to the Earth
in order to derive rates of impacts of large bodies, but erosion
and plate tectonics make this analysis difficult. All these tech-
niques have large uncertainties and possible systematic biases,
which are difficult to evaluate and compare. The development
of new techniques to compute impact fluxes is therefore con-
venient in order to have a good estimate of the impact rates of
meteoroids in our planet. Recently, a new technique to evaluate
impact rates of large meteoroids has been proposed and at-
tempted (Ortiz et al., 1999). The idea is to monitor the night side
of the Moon for impact flashes using CCD detectors attached to
small telescopes. This technique has the advantage that the area
covered by just one single instrument is much larger than the
area covered even with detector networks on Earth. The use of
CCD detectors is also an advantage compared to the use of pho-
tomultipliers, as proposed by Melosh et al. (1993), because the
background at each pixel is small enough so that the threshold
for flash detection is much smaller than using a photomultiplier.
Impact flashes on the Moon using the CCD technique have been
unambiguously detected during meteor showers, like the 1999
Leonid meteor storm (Dunham et al., 1999; Ortiz et al., 2000;
Yanagisawa and Kisaichi, 2002), the 2001 Leonids (Cudnik et
al., 2002; Ortiz et al., 2002), the 2004 Perseids (Yanagisawa et
al., 2006), the 2004 Leonids (Ortiz et al., 2005), the 2005 Per-
seids (Ortiz et al., in preparation) and possibly the 2005 Taurids
(Cooke et al., 2006). After the successful lunar monitoring ex-
perience of the 1999 and 2001 Leonids, a systematic search for
sporadic impact flashes was conducted by our team. Here we
present the first results of such a survey and some consequences
that can be drawn from the data.
2. Observations and reductions
Two specific instrumental setups were used for our lunar im-
pact flash survey, located at Sierra Nevada and Huétor-Santillán
Observatories in Granada, Spain. The Sierra Nevada instrumen-
tal setup consisted of two identical 0.36 m Schmidt–Cassegrain
telescopes equipped with high sensitivity CCD PAL video cam-
eras and VHS tape recorders. This pair of telescopes was de-
signed to image the same area of the Moon. We considered
that a flash was real when it was recorded by both instruments.
Cosmic rays and electronic as well as video tape noise can occa-
sionally be mistaken with impact flashes; therefore, redundancy
checks were done. Thus, if an intensity spike was detected si-
multaneously in two telescopes at the same lunar coordinates,
the flash was considered a real impact event. The lunar area
covered by this instrumental setup was 2.1×106km2±3%.
The Huétor-Santillán instrumental setup was identical to that in
Sierra Nevada except that the telescopes were a 0.2 m Schmidt–
Cassegrain telescope and a 0.4 m Newtonian telescope with
focal lengths adjusted so that both telescopes had a very sim-
ilar field of view. The lunar area covered redundantly by this
setup was 5.8×106km2±10%.
Here we report the results of 34 nights of observations
started in late 2001 and carried out mostly during 2002, 2003
and 2004. The observations were restricted to periods with lu-
nar phases ranging from 10 to 40% so that a large area of the
Tab le 1
List of observing campaigns
Date Observatory Recording time
19/02/02 Huétor S. 1 h 30 m
10/11/02 Huétor S. 3 h
06/03/03 Huétor S. 2 h
05/02/03 Huétor S. 2 h 30 m
13/09/02 Huétor S. 2 h
16/04/02 Huétor S. 2 h
13/08/02 Huétor S. 2 h
19/04/02 Huétor S. 3 h
25/04/04 Sierra Nevada 3 h
24/04/04 Sierra Nevada 3 h
01/09/03 Huétor S. 1 h 30 m
01/12/02 Sierra Nevada 1 h
02/06/03 Huétor S. 30 m
02/07/03 Huétor S. 1 h
02/09/03 Huétor S. 2 h
03/06/03 Huétor S. 1 h 30 m
03/07/03 Sierra Nevada 2 h
04/04/03 Huétor S. 1 h 30 m
04/07/03 Sierra Nevada 1 h
05/02/03 Sierra Nevada 2 h 30 m
06/02/03 Sierra Nevada 3 h
06/03/03 Sierra Nevada 30 m
06/04/03 Sierra Nevada 3 h
07/12/02 Sierra Nevada 2 h
09/03/03 Huétor S. 2 h 30 m
09/10/02 Huétor S. 30 m
12/10/02 Huétor S. 2 h 30 m
12/10/02 Sierra Nevada 3 h
25/10/04 Huétor S. 1 h 30 m
26/12/03 Huétor S. 1 h
29/12/03 Huétor S. 1 h
30/08/03 Huétor S. 1 h
18/11/01 Huétor S. 1 h 30 m
19/11/01 Huétor S. 2 h 30 m
night side of the Moon was visible without too much glare from
the dayside part. The observing log with details on the dates and
total time observed is shown in Table 1. The total number of ef-
fective hours of observation was 24 and 39 h with the Sierra
Nevada and Huétor-Santillán setups, respectively. We have de-
veloped a specific computer code that digitizes and analyzes the
stream of 25 images per second (from the video tapes) in real-
time. Basically the code searches for intensity spikes in residual
images obtained by subtracting the average of 3 previous im-
ages from the one being analyzed. The spikes detected are later
compared with the spikes detected in the second videotape from
the redundant telescope. This last step is done by visual inspec-
tion of the images containing the spikes.
From the analysis of the Sierra Nevada data, one impact
flash was detected in the effective 24 h of observations. The im-
pact flash as recorded by the two telescopes is shown in Fig. 1.
The analysis of the Huétor-Santillán data, with longer observ-
ing time and larger lunar coverage, yielded two impact flashes.
One of the Huétor-Santillán flashes is also shown in Fig. 1 to il-
lustrate the different image scales. The relevant data concerning
all three flashes are shown in Table 2.
The absolute calibration to standard astronomical magnitude
was carried out by referring the total data counts from the im-
pact to the best fit for 2004 Earthshine surface brightness mea-

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.3 (1-8)
Sporadic lunar impact flashes 3
Fig. 1. (a) Impact flash detected from one of the telescopes at Sierra Nevada on February 5th, 2003. (b) The same impact flash as (a) but seen from the second
telescope at Sierra Nevada. (c) Impact flash from one of the telescopes at Huetor-Santillán on December 26th, 2003. (d) The same impact flash as (c) but seen from
the second telescope at Huetor-Santillán. The Lunar area covered by the Sierra Nevada and Huetor-Santillán setups is 2.1×106km2±3% and 5.8×106km2±10%,
respectively.
Tab le 2
Characteristics of the impact flashes detected
Date (UTC) Selenographic coordinates Integration time (s) Vmagnitude
Longitude (◦) Latitude (◦)
05 Feb 2003 19:24:43 ±361±1.0W 11.1±0.4S 0.04 8.32 ±0.18
26 Dec 2003 17:36:38 ±319.4±0.8W 14.8±0.7N 0.04 7.09 ±0.25
19 Feb 2002 19:40:04 ±315±2W 20±2S 0.04 7.58 ±0.40
surements. The Earthshine Project accomplishes these continu-
ous observations for five Earthshine regions (Qiu et al., 2003;
Pallé et al., 2003). We did this rather than using standard star
calibrations in order to ensure that a possible change in the gain
setting of the cameras would not affect the results. This method
also avoided problems with extinction corrections, which were
not needed in our Earthshine calibration scheme (because the
calibration source was always at the same air-mass as the im-
pact flash).
3. Results and discussion
Our impact flashes correspond to sporadic events because no
major meteor showers were active or exhibit favorable impact
geometry on the impact dates. Other minor streams were also
discarded by checking the activity periods and impact geometry
for particles associated with minor streams lists (Cook, 1973;
Terentjeva, 1990; Jenniskens, 1994). Artificial satellites are
also easily discarded because they leave elongated brightening
streaks.
The luminous energy released in the impact events can be
computed using the known luminous flux received on Earth af-
ter correcting for the Earth–Moon distance. The initial kinetic
energy of the meteoroid has been computed in previous im-
pact flash works by means of the luminous efficiency concept,
which is the fraction of the kinetic energy that is emitted in
the visible. We have followed the same formalism and equa-
tions as in Bellot Rubio et al. (2000), and Ortiz et al. (2002) to
derive the kinetic energy. This kinetic energy can be translated
into impactor mass assuming a typical sporadic impactor speed.
According to the statistics of a large meteoroid orbit database
(Steel, 1996) this speed is approximately 20.2 km/sonEarth
and 16.9 km/s on the Moon, after correcting for the different
escape velocities of the Earth and the Moon. The value for the

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.4 (1-8)
4J.L. Ortiz et al. / Icarus ••• (••••)•••–•••
Moon is close to the 16.1 km/s average obtained by Ivanov
(2001).
Using the luminous efficiency η=0.002 (the nominal value
determined from Leonid impact flashes, e.g., Bellot Rubio et
al., 2000; Ortiz et al., 2002) the masses of the impactors would
be 1.9±3.0
0.9kg for the Sierra Nevada detection and 5.9±9.0
3.0kg
and 3.7±5.0
1.8kg for the Huétor-Santillán detections. The uncer-
tainties in the masses arise from the uncertainties in the velocity
and absolute calibration. We have used a 5-km/s uncertainty
for the impact speed (which is close to the standard deviation
of the lunar impact velocity distribution by Ivanov, 2001), but
the distribution is not Gaussian and it may well happen that
the impact speed of the particular events that we observed were
much higher or lower than the average. Therefore, the masses
calculated here are just estimates. However, the kinetic energy
is well determined and is only affected by the luminous effi-
ciency adopted. The luminous efficiency η=0.002 is already
a very optimistically high value according to numerical models
(Nemtchinov et al., 1998; Melosh et al., 1993) or based on hy-
pervelocity impact flash experiments (Ernst and Schultz, 2005;
Kadono and Fujiwara, 1996; Eichhorn, 1975). Besides, the
0.002 value was derived from 71 km/s impacts, whereas at
16 km/s, a much smaller luminous efficiency would be ex-
pected.
The lunar impact rates as a function of energy can be trans-
lated into terrestrial impact rates by scaling them to the appro-
priate terrestrial surface area and using a 1.3 gravitational focus-
ing factor. A small correcting factor for the kinetic energy has
to be applied to the terrestrial case, because the larger Earth’s
gravity causes a larger impact speed on Earth as compared to
the lunar case. This factor is assumed to be 1.4, based on the
statistics of impact speeds on Earth and the Moon. The exact
correction factor depends weakly on the actual impact speed of
the meteoroid. It is useful to express the energy in kilotons in or-
der to compare with previous works in the field (note that 1 kton
is 4.185 ×1012 J). We use energy rather than mass because ex-
pressing the impact rate as a function of mass would require
a correct choice of each meteoroid’s impact speed, whereas if
the impact rate is expressed as a function of kinetic energy,
no critical assumption is made. In other words, the calculation
of masses is only done for illustrative reasons. The fluxes are
shown in Figs. 2a and 2b for two different luminous efficien-
cies (0.002 and 0.006, respectively). In Figs. 2a and 2b we plot
the results along with Brown et al. (2002) impact hazard fit and
other data as explained below.
We have added to our analysis what we consider two other
candidates to lunar impact flash data. One is the lunar flash
recorded serendipitously on film in 1953 (Stuart, 1956). Al-
though it may be argued that the duration of that flash seems
too long for an impact into a body without atmosphere, recent
works (Ortiz et al., 2002; Yanagisawa and Kisaichi, 2002)have
shown light curves of lunar impacts that lasted in the order of
1 s. Therefore the 1–8 s duration reported by Stuart is compat-
ible with the phenomenology observed in some impacts (Ortiz
et al., 2002; Yanagisawa and Kisaichi, 2002).
Taking into account the Earth’s larger area (a factor of
∼27 times that of the lunar area visible from Earth) and Earth’s
gravitational focusing factor (1.3), the rate of impacts on Earth
should be ∼35 times that of the Moon. Therefore the rate of
impacts with similar energy to the Stuart event should be at
least 35 ×(1/52)per year on Earth because 52 years have
elapsed since the Stuart event. This value is a lower limit be-
cause the Moon has not been continuously imaged during these
years. Therefore the value of the flux that we get on Earth (for
∼10 kton blasts, assuming a 0.002 luminous efficiency) is a
very conservative lower limit. The terrestrial rate of impacts
with similar energy to the 1953 event on the Moon is shown
in Figs. 2a and 2b. In these two figures, two different luminous
efficiencies have been used as indicated in the plots (0.002 and
0.006, respectively).
Another candidate to lunar impact data is the flash recorded
serendipitously on film in 1985 by Kolovos et al. (1988),al-
though the preferred explanation by the authors at the time
was an out-gassing event and subsequent flash triggered by
a piezoelectric phenomenon. Other investigators claimed that
this flash might have been related to the passage of artificial
satellites (Maley, 1991; Rast, 1991)butKolovos et al. (1992)
clearly demonstrated that the image features were inconsistent
with satellite passage. Here we have calculated the energy re-
leased using Kolovos et al.’s calibration and relate it to impactor
mass (assuming the same luminous efficiency as before). Using
a similar reasoning as with the Stuart flash, a lower limit to the
flux of ∼1 kton blasts on Earth can be derived. In this case one
has to take into account the fact that this flash would not have
been detected photographically if it had hit the sunlit portion of
the Moon, because its brightness was smaller than the surface
brightness of the sunlit side of the Moon. The effective area of
the Moon for the observation of Kolovos-like events is a factor 2
smaller than for the observation of Stuart-like events. This is be-
cause the total lunar night area that is observable on average per
lunation is half the lunar semisphere area. Therefore, the impact
rate of Kolovos-like events on Earth would be 2 ×35 ×(1/20)
per year, where 20 is the number of years elapsed since 1985.
The lower limit on the terrestrial impact rate based on the 1985
flash is shown in Figs. 2a and 2b for η=0.002 and η=0.006,
respectively.
Assuming that the terrestrial cratering rate for impactors
with diameters larger than 1 km (Grieve and Shoemaker, 1994)
is valid (actually this rate is consistent with the lunar crater-
ing rate for impactors of the same size range Werner et al.,
2002), we have estimated the influx at intermediate sizes by
fitting a power law to the data (Fig. 2a). The impacts of objects
larger than 1-km is equivalent to blasts larger than ∼100,000
Mton (assuming typical asteroid density and impact speed of
2700 kg/m3and 20.2 km/s, respectively). The resulting impact
fluxes from the fit are at least 1 order of magnitude higher than
those from an equivalent fit to the latest results on terrestrial
impact rates by Brown et al. (2002). This discrepancy indicates
that something must be revised, like perhaps the luminous effi-
ciency.
Therefore we have tested the use of a much larger luminous
efficiency (η=0.02). In this case, the masses of the impactors
would be 0.19 ±0.30
0.09 kg for the Sierra Nevada detection and
0.59±0.90
0.30 and 0.37 ±0.50
0.18 kg for the Huétor-Santillán detec-

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.5 (1-8)
Sporadic lunar impact flashes 5
(a)
(b)
Fig. 2. (a) Cumulative impact rates on Earth as a function of the kinetic energy of the impactors. Minus symbols denote the results from our survey for a luminous
efficiency of 2 ×10−3and filled circles correspond to unexplained phenomena on the Moon (Stuart, 1956; Kolovos et al., 1988) that we interpret here as impact
flashes (using the same luminous efficiency of 2 ×10−3). The impact rates from the unexplained lunar events are only lower limits and this is indicated by the
arrows, whose length is arbitrary. Solid line: best fit to the lunar impact flash data and the impact rate of objects larger than 1 km by Grieve and Shoemaker (1994).
Dashed line: best impact hazard fit from Brown et al. (2002). Squares: Revelle (2001) data. See the text for a description of each data set. (b) Comparison of different
estimates of impact rates on Earth from different authors. Minus symbols: Impact fluxes from our systematic survey using a luminous efficiency of 6×10−3. Filled
circles: Impact rates from the Stuart (1956) and Kolovos et al. (1988) events interpreted here as impact flashes with a luminous efficiency of 6 ×10−3.Thesetwo
impact rates from the unexplained lunar events are lower limits and this is indicated by the arrows, whose length is arbitrary. Thick solid line: our preferred fit.
Dotted dashed line: Ceplecha (1996) data. Squares: Revelle (2001) data. Dotted line: fluxes from the lunar cratering rate (Ivanov et al., 2003) scaled to match the
1-km impactor rate. Diamond: Impact rate for objects greater than 1 km from Grieve and Shoemaker (1994). Crosses: Shoemaker (1983) fluxes. Dashed line: best
impact hazard fit from Brown et al. (2002). Triangles: data from Rabinowitz et al. (2000) scaled to match the 1-km impact rate. See the text for a description of each
data set.

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.6 (1-8)
6J.L. Ortiz et al. / Icarus ••• (••••)•••–•••
tions. Also the masses of the 1953 and 1985 impactors are
considerably reduced. Using this value for the luminous effi-
ciency the large discrepancy between our values of the terres-
trial impact rates and those from Brown et al. (2002) is re-
moved. The main difficulty here is the fact that η=0.02 for
sporadics, at 16.9 km/s, would imply at least η=0.06 for the
lunar Leonids (using a linear dependence of the luminous effi-
ciency with speed, see the last paragraph of the discussion) but
this is much larger than the maximum value compatible with
the lunar Leonids in 1999 and 2001 as reported by Bellot Ru-
bio et al. (2000) and Ortiz et al. (2002). Also, a recent work
(Yanagisawa et al., 2006) on the detection of a Perseid lunar
impact flash in 2004 seems to be incompatible with a 0.02 lu-
minous efficiency, as the size distribution of the Perseids would
be very different to what is obtained from meteor observations
on Earth.
Thus, if we consider that lunar impact flashes have a 0.02
luminous efficiency, the frequency of sporadic impact flashes
is entirely consistent with Brown et al. (2002),butη=0.02
is not consistent with previous works on luminous efficiency
and Brown et al. data is not consistent with other data sets
like those of Ceplecha (1996) and other data discussed below.
The most likely explanation to the discrepancies between our
results and those from Brown et al. (2002) is that both the lu-
minous efficiency and the impact hazard may be larger than as-
sumed. In fact, a different analysis of satellite data (Nemtchinov
et al., 1997) shows a larger terrestrial impact rate, very sim-
ilar to that of Shoemaker (1983). The impact rates of large
bolides from infrasound measurements (Revelle, 2001), plotted
in Figs. 2a and 2b are clearly above Brown et al.’s. Also, Revelle
(2001) cautions that the infrasound technique is only sensitive
to deeply penetrating bodies and therefore would miss a frac-
tion of the impactors. Results from terrestrial impact flashes
detected by military satellites in the infrared are unavailable
and were not published by Brown et al. (2002), who dealt with
visible data only. We have found an indirect citation in the liter-
ature for a rough estimate of an 80 kton blast per year (Beatty,
1994) from infrared detectors. Such a frequency is even above
our initial estimates, but there is no detailed account of the ob-
servations or the calibration system in that paper. Tagliaferri et
al. (1994) mention that “because of the scanning nature of the
infrared sensors and the manner in which the satellite data were
collected, the true number of events was at least 10 times what
we report here. The objects observed exhibited energies of ap-
proximately 10 kton equivalent down to a 1 kton; therefore our
observations indicate a much higher rate of impact of objects in
this size range than indicated in Shoemaker (1983)” and this re-
ferred to the period of time from 1975 to 1992 for which they
reported 136 impacts. Therefore, the flux would be much higher
than that by Brown et al. (2002).
The Neat and Spacewatch dataset (Rabinowitz et al., 2000)
is the only asteroid survey dealing with small-size bodies (some
of them in the 10-m size range), although it does not provide a
direct estimate of the impact rate. Despite this, the size distrib-
ution from that data set can be used to derive impact rates as a
function of energy by using the impact rate of 1-km bodies to
scale the data. In order to translate the Rabinowitz et al. (2000)
absolute magnitude (H ) distribution to kinetic energy distribu-
tion one has to assume an average albedo, an average density
and an average impact speed (we have used values of 0.12,
2700 kg/m3and 20 km/s, respectively). After this is done, the
impact rate obtained is larger as well (see Fig. 2b). It must be
noted that the cometary contribution to the impact rate, sup-
posed to be between 10 and 30% of the total (e.g., Shoemaker,
1983) is not included in the asteroid survey data. By using the
lunar impact flash and the infrasound and Neat and Spacewatch
data in the smaller size end we get our best estimate for a lu-
minous efficiency of 0.006 (Fig. 2b) which implies a very small
crater for the 1953 event, much smaller than what was estimated
by Buratti and Johnson (2003), who used a smaller luminous ef-
ficiency. A value of 0.006 for the lunar sporadics implies a value
of nearly 0.02 for the Leonids (if a linear dependence of lumi-
nous efficiency with speed is assumed) and this is within the
error bars of the luminous efficiency reported from the Leonids
in 1999 and 2001. Overplotted in Fig. 2bareCeplecha (1996)
influx values (obtained from a compilation of a variety of tech-
niques), and translated into impact rates as a function of energy
by using the average 20.2 km/s impact speed on Earth. These
results are considerably higher in the 1–10 m size range, but
are in agreement in the lower energy range. Also plotted in this
figure is the Ivanov et al. (2003) impactors’ relative size distrib-
ution translated into energy distribution and scaled to match the
1-km impact rate adopted here. The size distribution is trans-
lated into energy distribution by using an average impact speed
of 20.2 km/s and average density of 2700 kg/m3.TheIvanov et
al. (2003) data are derived from the crater size distribution on
the Moon. Our best estimate implies that objects with masses
larger than 1 kg impact at least 80,000 times per year on Earth,
and blasts larger than 15 kton could be more frequent than 1 per
year.
4. Conclusions and implications
In summary, as expected, our terrestrial impact flux esti-
mates based on sporadic flashes on the Moon are highly depen-
dent on the luminous efficiency used. A value of 0.02 would
provide results consistent with the terrestrial impact flashes
by Brown et al. (2002) but incompatible with the upper limit
from previous Leonid impact flashes (which took place at much
higher speed), and orders of magnitude higher than what is
derived from laboratory experiments and numerical models.
Besides, a recent work (Yanagisawa et al., 2006) on the detec-
tion and analysis of a Perseid lunar impact flash seems to be
inconsistent with a 0.02 luminous efficiency because the size
distribution of the Perseid stream would have to be too steep.
For those reasons we suggest that the latest impact hazard esti-
mates may be somewhat too low. We have found that our best fit
requires an enhancement of at least a factor 2 in the terrestrial
impact rate.
A continuous lunar impact flash monitoring and subse-
quent cross-calibrations of the luminous efficiency against well-
known meteoroid showers other than the Leonids may allow
a more accurate sporadic impact flux evaluation in the future
(because the meteor showers are essentially monovelocity and

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.7 (1-8)
Sporadic lunar impact flashes 7
permit a good determination of the luminous efficiency at the
shower speed). Also, the exact magnitude of the flash from
the hypervelocity impact of the 290 kg ESA SMART-1 space-
craft on the Moon, scheduled for September 2006, may provide
an additional opportunity to derive luminous efficiency con-
straints under well-known velocity and geometry conditions.
This could perhaps be an adequate proxy for the lunar spo-
radic impact case, although the 2 km/s spacecraft speed would
require some scaling compared to the sporadic impact speed.
There are different estimates of the effect of impactor speed (v)
on the luminous efficiency. A v3dependence of the total in-
tensity (as shown by Ernst and Schultz, 2004) implies a linear
dependence of the luminous efficiency with speed. This is be-
cause the luminous efficiency is the ratio of the total emitted
luminous energy (proportional to v3) to the kinetic energy (pro-
portional to v2). Therefore, the result of the division is a lin-
ear dependence on v. Our own fit to the data from Fig. 3b of
another experimental work at higher speeds (Eichhorn, 1976)
indicates that the luminous efficiency has a v1.2dependence.
Our nominal estimate of the SMART-1 impact flash visual mag-
nitude is in the range of 7.7–8.9 (in the most optimistic and
pesimistic cases, respectively), well within the reach of a large
number of telescopes. Spectroscopic observations would also
be feasible by large telescopes. These estimates are made for
0.04 s flashes, which is the typical duration of most lunar flashes
recorded so far, but some flashes from meteoroid streams lasted
more than 10 times longer than that (Yanagisawa et al., 2006;
Ortiz et al., 2002). In the event that the SMART-1 spacecraft
caused a long duration flash, the peak brightness of the flash
would be decreased, making it harder to detect. In this scenario,
a 0.4 s flash would result in an integrated visual magnitude of
10.2–12.4, which would be difficult to observe within a typical
∼10 mag/arcsec2background (caused by the glare of the bright
sunlit part of the Moon). At the lightcurve maximum the mag-
nitude would lie somewhere in the middle of the values stated
above (from 9 to 10.5). Thus, the flash would be more easily
detectable at peak brightness and this requires the use of fast
imaging cameras. In any scenario, the use of fast imaging de-
vices is important to record this kind of event.
Acknowledgments
We thank W.J. Cooke and an anonymous referee for their
help to improve the paper. We are also grateful to the Sierra
Nevada Observatory staff. Support from Projects AYA2005-
07808-C0-01, AYA2004-03250, and AYA2002-00382 is ac-
knowledged. FEDER funding is also acknowledged.
References
Beatty, K.J., 1994. “Secrets” revealed. Sky Telescope 87, 26–27.
Bellot Rubio, L.R., Ortiz, J.L., Sada, P.V., 2000. Luminous efficiency in hyper-
velocity impacts from the 1999 lunar Leonids. Astrophys. J. 542, L65–L68.
Brown, P., Spalding, R.E., Revelle, D.O., Tagliaferri, E., Worden, S.P., 2002.
The flux of small near-Earth objects colliding with the Earth. Nature 420,
294–296.
Buratti, B.J., Johnson, L.L., 2003. Identification of the lunar flash of 1953 with
a fresh crater on the Moon’s surface. Icarus 161, 192–197.
Ceplecha, Z., 1988. Earth’s influx of different populations of sporadic me-
teoroids from photographic and television data. Astron. Inst. Czech. 39,
221–236.
Ceplecha, Z., 1996. Luminous efficiency based on photographic observations
of the Lost City fireball and implications for the influx of interplanetary
bodies onto Earth. Astron. Astrophys. 311, 329–332.
Cook, A.F., 1973. A working list of meteor streams. In: C.L. Hemenway, P.M.
Millman, A.F. Cook (Eds.), Proceedings of IAU Colloq. 13. NASA SP 319,
p. 183.
Cooke, W.J., Suggs, R.M., Swift, W.R., 2006. A probable Taurid impact on the
Moon. Lunar Planet. Sci. 37. Abstract 1731.
Cudnik, B.M., Dunham, D.W., Palmer, D.M., Cook, A.C., Venable, J.R., Gural,
P.S., 2002. Ground-based observations of high velocity impacts on the
Moon’s surface—The lunar Leonid phenomena of 1999 and 2001. Lunar
Planet. Sci. 33. Abstract 1329C.
Dunham, D.W., Cudnik, B., Hendrix, S., Asher, D.J., 1999. Lunar Leonid me-
teors. IAUC 7320.
Eichhorn, G., 1975. Measurements of the light flash produced by high velocity
particle impact. Planet. Space Sci 23, 1519–1525.
Eichhorn, G., 1976. Analysis of the hypervelocity impact process from impact
flash measurements. Planet. Space Sci. 24, 771–781.
Ernst, C.M., Schultz, P.H., 2004. Early-time temperature evolution of the im-
pact flash and beyond. Lunar Planet. Sci. 35. Abstract 1721E.
Ernst, C.M., Schultz, P.H., 2005. Investigations of the luminous energy and
luminous efficiency of experimental impacts into particulate targets. Lunar
Planet. Sci. 36. Abstract 1475E.
Grieve, R.A.F., Shoemaker, E.G., 1994. The record of past impacts on Earth. In:
Gehrels, T., Matthews, M.S., Schumann, A. (Eds.), Hazards Due to Comets
and Asteroids. In: Space Science Series. Univ. of Arizona Press, Tucson,
pp. 417–462.
Halliday, I., Griffin, A.A., Blackwell, A.T., 1996. Detailed data for 259 fireballs
from the Canadian camera network and inferences concerning the influx of
large meteoroids. Meteor. Planet. Sci. 31, 185–217.
Ivanov, B.A., 2001. Mars/Moon cratering rate ratio estimates. Space Sci.
Rev. 96, 87–104.
Ivanov, B.A., Neukum, G., Bottke, W.F., Hartmann, W.K., 2003. The com-
parison of size-frequency distributions of impact craters and the planetary
cratering rate. In: Bottke, W.F., Cellino, A., Paolicchi, P., Binzel, R.P. (Eds.),
Asteroids III. Univ. of Arizona Press, Tucson, pp. 89–101.
Jenniskens, P., 1994. Meteor stream activity. I. The annual streams. Astron.
Astrophys. 287, 990–1013.
Kadono, T., Fujiwara, A., 1996. Observation of expanding vapor cloud gener-
ated by hypervelocity impact. J. Geophys. Res. 101, 26097–26109.
Kolovos, G., Seiradakis, J.H., Varvoglis, H., Avgoloupis, S., 1988. Photo-
graphic evidence of a short duration—Strong flash from the surface of the
Moon. Icarus 76, 525–532.
Kolovos, G., Seiradakis, J.H., Varvoglis, H., Avgoloupis, S., 1992. The origin
of the Moon flash of May 23, 1985. Icarus 97, 142–144.
Maley, P.D., 1991. Space debris and a flash on the Moon. Icarus 90, 326–327.
Melosh, H.J., Artemjeva, N.A., Golub, A.P., Nemchinov, I.V., Shuvalov, V.V.,
Trubetskaya, I.A., 1993. Remote visual detection of impacts on the lunar
surface. Lunar Planet. Sci. 24. 45.
Nemtchinov, I.V., Svetsov, V.V., Kosarev, I.B., Golub, A.P., Popova, O.P., Shu-
valov, V.V., Spalding, R.E., Jacobs, C., Tagliaferri, E., 1997. Assessment
of kinetic energy of meteoroids detected by satellite-based light sensors.
Icarus 130, 259–274.
Nemtchinov, I.V., Shuvalov, V.V., Artem’eva, N.A., Ivanov, B.A., Kosarev, I.B.,
Trubetskaya, I.A., 1998. Light flashes caused by meteoroid impacts on the
lunar surface. Solar Syst. Res. 32, 99–114.
Ortiz, J.L., Aceituno, F.J., Aceituno, J., 1999. A search for meteoritic flashes on
the Moon. Astron. Astrophys. 343, L57–L60.
Ortiz, J.L., Sada, P.V., Bellot Rubio, L.R., Aceituno, F.J., Aceituno, J., Gutiér-
rez, P.J., Thiele, U., 2000. Optical detection of meteoroidal impacts on the
Moon. Nature 405, 921–923.
Ortiz, J.L., Quesada, J.A., Aceituno, J., Aceituno, F.J., Bellot Rubio, L.R., 2002.
Observation and interpretation of Leonid impact flashes on the Moon in
2001. Astrophys. J. 576, 567–573.

ARTICLE IN PRESS YICAR:7956
JID:YICAR AID:7956 /FLA [m5+; v 1.60; Prn:23/06/2006; 12:57] P.8 (1-8)
8J.L. Ortiz et al. / Icarus ••• (••••)•••–•••
Ortiz, J.L., Aceituno, F.J., Santos-Sanz, P., Quesada, J.A., 2005. A study of
Leonid impact flashes on the Moon in 2004. In: American Astronomical
Society, DPS Meeting 37, 17.05.
Pallé, E., Goode, P.R., Yurchyshyn, V., Qiu, J., Hickey, J., Montañés Ro-
driguez , P., Chu, M.-C., Kolbe, E., Brown, C.T., Koonin, S.E., 2003. Earth-
shine and the Earth’s albedo. 2. Observations and simulations over 3 years.
J. Geophys. Res. 108 (D22). 4710.
Qiu, J., Goode, P.R., Pallé, E., Yurchyshyn, V., Hickey, J., Montañés Rodriguez,
P., Chu, M.-C., Kolbe, E., Brown, C.T., Koonin, S.E., 2003. Earthshine and
the Earth’s albedo. 1. Earthshine observations and measurements of the lu-
nar phase function for accurate measurements of the Earth’s bond albedo.
J. Geophys. Res. 108 (D22). 4709.
Rabinowitz, D., Helin, E., Lawrence, K., Pravdo, S., 2000. A reduced estimate
of the number of kilometre-sized near-Earth asteroids. Nature 403, 165–
166.
Rast, R.H., 1991. The “Moon” flash of 1985 May 23 and orbital debris.
Icarus 90, 328–329.
Revelle, D.O., 2001. Global infrasonic monitoring of large bolides. In: Warm-
bein, B. (Ed.), Proceedings of the Meteoroids 2001 Conference, 6–10 Au-
gust 2001, Kiruna, Sweden, ESA SP-495. ESA Publications Division, No-
ordwijk, p. 483.
Shoemaker, E.M., 1983. Asteroid and comet bombardment of the Earth. Annu.
Rev. Earth Planet. Sci. 11, 461–494.
Steel, D., 1996. Meteoroid orbits. Space Sci. Rev. 78, 507–553.
Stuart, L.H., 1956. A photo-visual observation of an impact of a large meteorite
on the Moon. Strolling Astron. 10, 42–43.
Tagliaferri, E., Spalding, R., Jacobs, C., Worden, S.P., Erlich, A., 1994. Detec-
tion of meteoroid impacts by optical sensors in Earth orbit. In: Gehrels, T.
(Ed.), Hazards Due to Comets and Asteroids. Univ. of Arizona Press, Tuc-
son, pp. 199–220.
Terentjeva, A., 1990. Fireball Streams. In: Lagerkvist, C.I., Rickman, H., Lind-
blad, B.A. (Eds.), Proceedings of AMC, vol. 89, p. 579.
Werner, S.C., Harris, A.W., Neukum, G., Ivanov, B.A., 2002. The near-Earth
asteroid size-frequency distribution: A snapshot of the lunar impactor size-
frequency distribution. Icarus 156, 287–290.
Yanagisawa, M., Kisaichi, N., 2002. Lightcurves of 1999 Leonid impact flashes
on the Moon. Icarus 159, 31–38.
Yanagisawa, M., Ohnishi, K., Takamura, Y., Masuda, H., Ida, M., Ishida, M.,
Ishida, M., 2006. The first confirmed Perseid lunar impact flash. Icarus 182,
489–495.