Atlantis Online
March 28, 2024, 07:27:38 am
Welcome, Guest. Please login or register.

Login with username, password and session length
News: Did Humans Colonize the World by Boat?
Research suggests our ancestors traveled the oceans 70,000 years ago
http://discovermagazine.com/2008/jun/20-did-humans-colonize-the-world-by-boat
 
  Home Help Arcade Gallery Links Staff List Calendar Login Register  

Wow! signal

Pages: 1 [2]   Go Down
  Print  
Author Topic: Wow! signal  (Read 479 times)
0 Members and 32 Guests are viewing this topic.
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #15 on: September 03, 2007, 01:51:27 pm »



Fig. 3 CITED IN TEXT  |  HI-RES IMAGES: 3a (233kb) 3b (352kb)  |       

--------------------------------------------------------------------------------
Spectra of sources at (a) R.A. = 19h22m220, decl. = -2654284 and (b) R.A. = 19h25m131s, decl. = -2708221. The resolution is 12.2 kHz.
Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #16 on: September 03, 2007, 02:03:13 pm »

No large variation in 21 cm flux was found between the 1995 and 1996 epochs, evidence that the sources are not variable on that timescale. Optical survey images were examined for coincident stellar objects, yielding a blank field at one position and a very faint possible optical counterpart for the source at the R.A. = 19h22m22s position; we have not determined whether it is stellar or Galactic. No evidence was found that either source is unusual.

2.1.2. Continuum Sources in All Fields

     We also investigated the channel 0 sources outside the error boxesapproximately 1m of right ascension on either side of the nominal right ascensions. None displayed unusual spectral features, and all of the sources detected in 1995 were also detected at 6 cm in 1996. Two sources had somewhat unusual spectral indices (see Table 2). We conclude that these too are continuum sources, with no unusual features to suggest they might be the source of the Wow emission.

2.2. Narrowband Search

     We next sought to identify narrow-bandwidth emissions from unresolved sources. Data cubes with 127 channels were made with IMAGR, having 6.1 kHz resolution (1995 observations) and 12.2 kHz resolution (1996 observations), with a typical single-channel noise of 517 and 3.6 mJy, respectively. SAD was used to create a catalog of candidate features in each single-channel image, ignoring those below a 4.2  signal-to-noise ratioa threshold below the amplitude of most but not all expected noise peaks, selected to reduce the large number of "features" that noise alone is expected produce in such large data sets (totaling more than 108 discrete samples). The flux threshold for each channel was based on the image rms noise, which had the effect of raising the threshold flux in midband channels that sampled confused, extended Galactic H I emission. The list initially included some continuum sources, but features near coordinates of continuum sources were ignored.

     To identify any sources with flux significantly stronger than the noise peaks, we computed a flux threshold above which noise peaks are not expected. For n independent samples the probability of error Pe, that one or more samples will exceed a value of Zm in the absence of a real signal, is (after Thompson, Moran, & Swenson 1994, p. 264)


Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #17 on: September 03, 2007, 02:06:20 pm »

The number of independent samples was taken as the number of synthesized beams in an image, times the number of spectral channels in each beam. In the relatively high resolution BnA array configuration the two-dimensional Gaussian response to a point source covers an area of about 17 arcsec2 and nbeams = 1.9 × 105 for a field 1811 × 1811 (a 10242 pixel image with a 9 pixel edge ignored because of increased noise). In the lower resolution D configuration a point source covers an area of about 2194 arcsec2 and nbeams = 1.5 × 103 over a 1830 × 1830 field (a 2562 pixel image with a 24 pixel edge outside the 30 primary beam ignored). For each 127-channel data cube, n was taken as 127 times larger than the number of samples in the field to account for the number of channels in the cube. To test a detection hypothesis we choose threshold fluxes Zm so that Pe = 0.01; Table 3 shows the thresholds for various data sets.

2.2.1. Wow Error Boxes

     Single-channel sources with positions inside the Ohio State coordinate error boxes would be of greatest interest. In the 1996 D configuration data set two 5.6  features just meet the threshold for the error boxes but fall short of the 5.9  threshold for the entire data set; a noise peak as high as 5.2  is expected (Pe = 0.5) in the data set. We also analyzed "extended" error boxes from 1995 BnA-array observations: 10 partly overlapping fields covering about 15 in declination (twice the Ohio State uncertainty) and with the same bounds in right ascension. No peaks exceeded a 6.6  threshold for that data set.

2.2.2. All Fields

     We also searched for single-channel sources anywhere in the fields observed, to accommodate possible errors of up to 1m in the Ohio State right ascensions. The data sets included two fields at 12.2 kHz resolution and 10 partly overlapping fields at 6.1 kHz resolution. No features met a 5.9  threshold for the two D cubes or a 6.9  threshold for the 10 BnA cubes. Figure 4 shows the distribution of single-channel peaks over 4.2  for the two data sets.

2.2.1. Wow Error Boxes

     Single-channel sources with positions inside the Ohio State coordinate error boxes would be of greatest interest. In the 1996 D configuration data set two 5.6  features just meet the threshold for the error boxes but fall short of the 5.9  threshold for the entire data set; a noise peak as high as 5.2  is expected (Pe = 0.5) in the data set. We also analyzed "extended" error boxes from 1995 BnA-array observations: 10 partly overlapping fields covering about 15 in declination (twice the Ohio State uncertainty) and with the same bounds in right ascension. No peaks exceeded a 6.6  threshold for that data set.

2.2.2. All Fields

     We also searched for single-channel sources anywhere in the fields observed, to accommodate possible errors of up to 1m in the Ohio State right ascensions. The data sets included two fields at 12.2 kHz resolution and 10 partly overlapping fields at 6.1 kHz resolution. No features met a 5.9  threshold for the two D cubes or a 6.9  threshold for the 10 BnA cubes. Figure 4 shows the distribution of single-channel peaks over 4.2  for the two data sets.

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #18 on: September 03, 2007, 02:07:23 pm »

Fig. 4 CITED IN TEXT  |  HI-RES IMAGES: 4a (149kb) 4b (224kb)  |     

--------------------------------------------------------------------------------
Distribution of single-channel peaks 4.2  in 127-channel data cubes, with channel 0 sources removed. (a) Two data cubes with 12.2 kHz channel width. (b) Ten data cubes with 6.1 kHz channel width.


Fig. 4a

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #19 on: September 03, 2007, 02:09:35 pm »

Fig. 4b

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #20 on: September 03, 2007, 02:12:15 pm »

2.2.3. Other Techniques Used
     Real but weak sources of interest might, of course, lurk below a Pe = 0.01 threshold. We also screened the data for single-channel sources by (1) searching for repeated detections of weak features at the same coordinates in overlapping fields or during different observing runs, (2) searching for radio features with coordinates coinciding with objects in optical survey images or the Guide Star Catalog (Lasker et al. 1990), (3) direct inspection of portions of the UV data, (4) examining statistics on UV data, (5) analyzing 10 minute time ranges of data to identify intermittent sources attenuated by averaging over longer periods, (6) applying continuum subtraction using the AIPS task UVLIN, (7) searching for highly polarized features, and (Cool exploring data cubes with visualization software. None of these techniques identified convincing single-channel pointlike sources.
3. CONCLUSIONS
     No narrowband emission resembling the Ohio State Wow was found in searches of the apparent source positions using the VLA, to flux limits as low as 22 mJy at the nominal positions, and better than 120 mJy  0 5 north and south, with detection thresholds of 5.9  and 6.9  , respectively. Two continuum sources were detected inside Ohio State's coordinate error boxes but displayed no unusual spectral features, no unusual spectral indices, and no temporal variation.
     Our observations were sensitive enough to detect a putative underlying narrowband source  500 times weaker than the approximately 60 Jy Wow. The null result is good evidence that the Wow detection was not a continuous source, either natural or artificial, enhanced by scintillation sufficiently for Ohio State to detect on just one occasion, because the probability of a scintillation gain on the order of 100 is vanishingly small. Covering a range of frequency 3 times wider than Ohio State's observations, our observations also constrain the possibility that the original detection was due to a source drifting in frequency over a range of approximately 1.5 MHz. Our observations do not significantly constrain the possibility of highly intermittent emissions because we dwelled for no longer than 22 minutes on any field.
     R. G. thanks John Kraus, Robert Dixon, and Jerry Ehman at the Ohio State Radio Observatory for information on the Wow and Patrick Palmer at the University of Chicago for helpful suggestions on using the VLA. We thank Joseph Lazio at the Naval Research Laboratory for discussions on scintillation and the anonymous referee for suggesting improvements. This work was supported in part by the SETI Institute and The Planetary Society.
REFERENCES
•   Becker, R. H., White, R. L., Helfand, D. J., Gregg, M. D., & McMahon, R. 1996, in ASP Conf. Ser. 93, Radio Emission from the Stars and the Sun, ed. A. R. Taylor & J. W. Padres (San Francisco: ASP), 422 First citation in article
•   Cocconi, G., & Morrison P. 1959, Nature, 183, 844 First citation in article
•   Colomb, F. R., Hurrell, E., Lemarchand, G. A., & Olald, J. C. 1995, in ASP Conf. Ser. 74, Progress in the Search for Extraterrestrial Life, ed. G. S. Shostak (San Francisco: ASP), 345 First citation in article
•   Condon, J. J., Cotton, W. D., Greisen, E. W., Yin, Q. F., Perley, R. A., Taylor, G. B., & Broderick, J. J. 1998, AJ, 115, 1693 First citation in article | Full Text | NASA ADS
•   Cordes, J. M., & Lazio, T. J. W. 1991, ApJ, 376, 123 First citation in article | NASA ADS
•   Cordes, J. M., Lazio, T. J. W., & Sagan, C. 1997, ApJ, 487, 782 First citation in article | Full Text | NASA ADS
•   Dixon, R. S. 1985, in IAU Symp. 112, The Search for Extraterrestrial Life: Recent Developments, ed. M. D. Papagiannis (Dordrecht: Reidel), 305 First citation in article
•   Drake, F. D., & Sagan, C. 1973, Nature, 245, 257 First citation in article
•   Ehman, J. R. 1998, The Big Ear Wow! Signal, Ohio State University Radio Observatory, working paper First citation in article
•   Gray, R. H. 1994, Icarus, 112, 485 First citation in article | NASA ADS
•   Horowitz, P., Matthews, B. S., Forster, J., Linscott, I., Teague, C. C., Chen, K., & Backus, P. 1986, Icarus, 67, 525 First citation in article | NASA ADS
•   Kraus, J. D. 1979, Cosmic Search, 1, 31 First citation in article
•      . 1986, Radio Astronomy (2d ed.; Powell: Cygnus-Quasar) First citation in article
•   Lasker, B. M., et al. 1990, AJ, 99, 2019 First citation in article | NASA ADS
•   Perley, R. A., Schwab, F. R., & Bridle, A. H., eds. 1994, ASP Conf. Ser. 6, Synthesis Imaging in Radio Astronomy (San Francisco: ASP) First citation in article
•   Sullivan, W. T., Brown, C., & Wetherill, C. 1978, Science, 199, 377 First citation in article | NASA ADS
•   Tarter, J. 1995, Summary of SETI Observing Programs (Mountain View: SETI Institute). First citation in article
•   Thompson, R. A., Moran, J., & Swenson, G. W. 1994, Interferometry and Synthesis in Radio Astronomy (New York: Wiley) First citation in article
•   Wendker, H. J. 1995, A&AS, 109, 177 First citation in article | NASA ADS
________________________________________


http://www.journals.uchicago.edu/ApJ/journal/issues/ApJ/v546n2/40506/40506.html?erFrom=-4571361233589039105Guest
Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #21 on: September 03, 2007, 02:27:28 pm »

A Search for Periodic Emissions at the Wow Locale

Robert H. Gray
Gray Data Consulting, 3071 Palmer Square, Chicago, IL 60647; rgray@graydata.net
and
Simon Ellingsen
School of Mathematics and Physics, University of Tasmania, GPO Box 252-21, Hobart 7001, TAS, Australia; simon.ellingsen@utas.edu.au


Received 2001 May 21; accepted 2002 June 24

ABSTRACT

The Ohio State University Radio Observatory recorded a strong, narrowband emission near the 21 cm hydrogen line in 1977 during a search for extraterrestrial intelligence, an event known as the "Wow" signal. The few independent attempts to replicate the detection have failed. We have investigated the possibility of a periodic sourceperhaps rotating and illuminating us once each cycle of many hours, like a lighthousewhich prior observations would have been unlikely to detect. We used the University of Tasmania Hobart 26 m radio telescope to search for intermittent and possibly periodic emissions at the Wow locale by tracking the apparent source positions for nearly 14 hr continuously on multiple days. No emissions resembling the Wow were detected over a bandwidth of 2.5 MHz to a flux density limit of about 18 Jy, with a detection threshold of 5.9  and rms noise of 3 Jy. We conclude that the Wow was not due to a source within our flux density limits and repeating more often than every 14 hr, although the possibility of a longer period or nonperiodic source cannot be ruled out.

Subject headings: extraterrestrial intelligenceradio lines: general

1. INTRODUCTION

     The Wow was a strong (30 , or about 60 Jy), narrowband (<10 kHz) emission recorded near the 21 cm H I line at Ohio State University in 1977 (Kraus 1979; Ehman 19981) during a search for extraterrestrial intelligence (SETI), but not detected again in subsequent follow-up observations at Ohio State. The event is interesting because the time-dependent intensity of the emission matched the antenna pattern signature of a transiting celestial source (Gray & Marvel 2001), which would not be expected of interference.

     Two attempts to independently replicate the detection have been reported, both unsuccessful. One attempt (Gray 1994) used the ultranarrowband Harvard/Smithsonian-META system (Horowitz et al. 1986). With 0.05 Hz channels and a chirped receiver, the observations were sensitive only to very narrowband and Doppler-corrected artificial radio signals; none were detected. A second attempt used the Very Large Array (Gray & Marvel 2001). With spectral resolutions of 6.1 and 12.2 kHz, those observations approximated Ohio State's 10 kHz bandwidth and were much more sensitive (20 mJy), capable of detecting weak underlying continuous sources that might have been briefly enhanced by interstellar scintillation to produce the Ohio Sate detection. Dwelling on the apparent source locale for no more than 22 minutes, those observations were unlikely to detect intermittent sources.

     The Ohio State detection occurred in only one beam of a dual-beam transit antenna system (Dixon & Cole 1977), and a simple explanation is that the signal may have been present only during the time one beam swept past. The two beams were side by side and separated by 2m50s of right ascension; celestial sources were usually detected first in one for 72 s, then in the other.

     One way to account for an emission present only part of the time is a rotating source, in which case the emission might appear periodically. Pulsars are a familiar example. In the context of interstellar radio signals, an example scenario would be a directional broadcast from a fixed antenna on the surface of a rotating planet, sweeping across observers once each "day" like a lighthouse. Terrestrial ballistic missile early warning system radars are an example of such emissions that are potentially detectable over interstellar distances (Sullivan, Brown, & Wetherill 1978).

     A second possible explanation for the single-beam detection is a signal drifting in frequency, because the detection occurred in channel 2 of a 50 channel filter-bank receiver. It is possible that a signal with a fortuitous frequency drift rate, in the range of ±100 to ±150 Hz s-1, drifted into or out of the band observed during the time between the passage of the two beams.

     We investigated a periodic emission hypothesis because it is possible to test by the simple expedient of sufficiently extended observations. We also incidentally investigated a drifting frequency hypothesis by observing over a band 5 times wider than Ohio State's. Other hypotheses can be conjured up, of course, including local interference, which is impossible to rule out.

     1 Available at http://www.bigear.org/wow20th.htm.

1.1. Statistics of the Wow Detection

     In this section we calculate the probability of the original Wow detection, assuming emissions of various periods and durations, with the goal of estimating how long observations must be to detect a hypothetical periodic source.

     We treat each transit observation as a trial, which might or might not be looking in the right direction at the right time to observe a periodic source. The binomial distribution gives the probability PB of an event occurring x times in n independent trials, given the probability p of the event occurring in a single trial (Zombeck 1990, p. 410). The original detection is taken as one success in 18 trials because the Ohio State transit antenna was typically kept at a constant declination for 3 days and moved in half-HPBW increments (Dixon 1985); strong continuous sources could be detected on 9 days, and the two beams yield 18 trials.

     The probability of a detection success during one daily transit of one beam is simply the fraction of the terrestrial day a signal is presentfor example, p = (144 × 24)/86,400 = 0.04 for a 144 s signal repeating 24 times per dayand the two beams yield two trials per day. We consider signal durations of 72 (a lower limit for the Wow, since it was present during the full transit of one beam), 144, and 288 s. Durations shorter than about 100 s guarantee detection in only one of the two beams, so that the trials are independent. Durations longer than about 250 s could be detected in both beams, which begins to violate the assumption of independent trials.

     The probability of detecting a periodic source exactly once in 18 trials is given in Figure 1 for various periods and durations. For periods under about 2 hr the probability reaches over 0.35. For periods over 12 hr the probability of detection falls below 0.10, and over 24 hr is less than 0.05 for all durations considered. The Wow detection would seem rather lucky for a source with such a long period, although the Ohio State survey ran for several years and covered approximately half of the sky, so if many such sources existed, detecting one would not be too surprising. This analysis suggests that a search for periodic emissions should be extended in time, as long as 10 or 20 hr, but not very much longer.

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #22 on: September 03, 2007, 02:28:29 pm »



Fig. 1 CITED IN TEXT  |  HI-RES IMAGE (71kb)  |   

--------------------------------------------------------------------------------
Probability of exactly one detection in 18 trials. The solid line represents a signal duration of 144 s every n hours. The upper dotted line represents a signal duration twice as long, the lower dotted line half as long. The probability of exactly one detection falls off rapidly for periods less than about 1 hr as a result of multiple detections.
Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #23 on: September 03, 2007, 02:29:18 pm »

We note that the range of periods where detection is not too improbable (up to 1020 hr) includes the length of short planetary days. Sidereal rotation periods probably cannot be much less than 6 hr because of dynamical considerations (Lightman 1984). In our solar system four of nine planets have periods between 10 and 17 hr, two others 2425 hr, while the remaining three are much longer. The range of periods consistent with the Wow thus includes the intriguing scenario of an emission from the surface of a rotating planet, although periodicity could arise from other mechanisms as well.

1.2. Confirming Rotating Source Emissions

     Ohio State's 100 follow-up transit observations would not significantly constrain the possibility of a source with a period of more than several hours because they sampled sources for only 72 s twice each terrestrial day. For sources that are themselves periodic, periodic sampling makes a second, confirming detection extremely difficult because in many cases the source will be out of phase with the observations for many subsequent transits.

     A simple discrete computer simulation was performed to investigate this effect, using as its starting point Ohio State's detection, and continuing over 500 subsequent days at 1 s intervals, for source periods up to 36 hr in 0.01 hr increments. The source emission was taken as illuminating the Earth for 144 s during each extraterrestrial daylong enough for the observed 72 s detection, yet brief enough to avoid detection in Ohio State's other beam. A simulated detection was declared whenever the two simulated rotating beams overlapped for even a few seconds.

     The simulation results show that Ohio State's subsequent observations scattered over several hundred days would have had few opportunities for additional detections, making them a poor test for emissions with periods comparable to planetary days. For nearly half of the periods considered, a second detection would not occur for 100 days or more, when both the terrestrial antenna and the source point in the same direction again, illustrated in Figure 2.

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #24 on: September 03, 2007, 02:30:32 pm »



Fig. 2 CITED IN TEXT  |  HI-RES IMAGE (147kb)  |       

--------------------------------------------------------------------------------
Number of days until a second opportunity for detection using the Ohio State transit telescope, for sources with periods up to 36 hr. Prospects are especially poor for sources with periods near but not equal to the terrestrial 23.9345 hr day.
Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #25 on: September 03, 2007, 02:32:54 pm »

There is no evidence that the Wow was periodic, but a rotating source is one mechanism that naturally accounts for the single detection and failure of subsequent detection efforts.

2. OBSERVATIONS

     The apparent coordinates of the Wow emission were R.A. = 19h22m22s or 19h25m12s (both ±5s), decl. = -2703 ± 20 (B1950.0; J. D. Kraus 1990, private communication). The two right ascensions result from the dual-feed antenna system forming two beams. Ohio State recorded the difference in intensity between the two beams, but not the sign, so there was an ambiguity in which beam the emission was detected. Revisions to the coordinates have been proposed (Ehman 1998) but are small enough to neglect in the present work.


To produce a flat baseline, a third-degree polynomial was fitted to those regions of the spectrum free from H I emission and interference. A noise diode on the receiver was compared to the system total power once every 10 minutes to determine the system temperature. The system was calibrated against Virgo A, assumed to have a flux density of 211 Jy at 1420 MHz (Baars et al. 1977).

     Reference spectra are typically obtained through off-source observations, but one of our goals was to observe continuously over many hours. Since we were searching for a strong, time-variable signal, we were able to use an alternative approach: averaging smoothed on-source spectra taken at times near those of the signal spectrum. Reference spectra were formed by averaging groups of 30 s spectra between system calibrations at 10 minute intervals. This procedure has the drawback of including spectral features of potential interest in the reference spectrum, which would then be removed from the quotient spectrum. To avoid this, we used several techniques to exclude features from the reference spectra. To remove single-channel features, we median-smoothed each reference spectrum. To remove broader features, including H I, apparent radio frequency interference (RFI), and potentially interesting intermittent features, we first used only prior spectra to form a reference, then identified features in subsequent spectra, then "bridged" the reference spectrum with a straight line connecting channels across the base of the feature, an iterative procedure.



     Since objects at the Wow declination are visible for only 46 hr daily from most observatories in the northern hemisphere, a southern hemisphere site was necessary. We used the University of Tasmania Hobart 26 m radio telescope, which allowed us to track continuously for 14 hrthe maximum time the target coordinates are above that telescope's horizon.

     We tracked each of the two nominal positions during two days in 1998, for nearly 14 hr each. Additional observations were made in 1999, 15 north and south of the nominal coordinates, to fully cover the Ohio State 40 declination HPBW with the Hobart beam.
Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #26 on: September 03, 2007, 02:34:06 pm »

2.2. Sensitivity

     The theoretical rms () channel sensitivity of a spectrum obtained with the Mount Pleasant system is


where Tsys is the system temperature in Jy, Nchan is the number of spectral channels, B is the observed bandwidth of the spectrometer, t is the integration time in seconds, and Npol is the number of averaged polarizations. The factor of /2 is the decrease in sensitivity due to the 1 bit approximation made in the correlator, and the factor of is the increase in sensitivity due to Hanning smoothing. This yields  = 2.46 Jy for a single 30 s spectrum and 0.55 Jy for a reference spectrum consisting of 20 averaged spectra. The sensitivity expected for our quotient spectra is then 2.5 Jy (from standard propagation of errors), which is close to the 2.63.0 Jy rms obtained during our various observations.

     Our 3 Jy sensitivity was sufficient to yield an unambiguous detection of a 60 Jy sourcethe estimated Wow fluxwith a signal-to-noise ratio of 20.

2.3. Interference Identification and Excision

     The radio frequency interference environment was sampled for approximately 1 hr before and after each observing run, usually with the antenna pointed overhead. Signals detected during these off-source observations were assumed to be local interference, and those channels were ignored in subsequent analysis.

3. RESULTS AND DISCUSSION

3.1. H I

     The hydrogen emission line was prominent in all spectra, with a measured flux of 220 ± 20 Jy. The H I peak was 1520 kHz below the LSR-corrected H I frequency, indicating that interstellar hydrogen in the directions observed has a radial velocity of 34 km s-1.

     We used several strategies to accommodate the possibility that a signal of interest might be obscured by the H I background. To identify signals partly buried in H I, we calculated a running baseline flux for each channel, averaging the nine prior measurements in the channel, and subtracted it from the flux for each spectrum, effectively removing the constant H I profile. Ohio State used a somewhat similar running baseline removal method. To identify signals entirely buried in H I emission during one set of observationsand not fixed in the same LSR reference framewe staggered observations over approximately 6 months, so the H I emission would be Doppler shifted to different sky frequencies.

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #27 on: September 03, 2007, 02:35:18 pm »

3.2. Statistical Analysis

     To identify spectral features of possible interest, we computed a flux density threshold above which noise peaks are not expected. For n independent samples the probability of error Pe that one or more samples will exceed a value of Zm in the absence of a real signal is (after Thompson 1991)


3.3. Spectral Features

     Several apparent narrowband features above our detection threshold remained after excluding known interference and removing the H I profile. Some were traced to a fault in the local oscillator (LO) used for Doppler tracking, which caused sudden frequency shifts of 5 kHz at certain settings, shifting the H I profile one channel and causing the running baseline calculation to encounter a spurious flux density increase. Removing those, two apparent features remained.

     Field 1922N contained a 14  feature spanning two channels in both linear polarizations. It does not resemble the Ohio State event sufficiently to be of interest for two reasons. First, it lasted only one 30 s integration period (compared to at least 72 s for the Wow), and second, its frequency was 456 kHz higher than the Wow. Field 1922N contained a second possible feature, weak but noted because its frequency corresponds to the Wow. At 5.6  it was below our 5.9  detection threshold (Pe = 0.05) and below the 5.7  maximum expected over all observations (Pe = 0.5) but above the typical 5  maximum noise peak during a single day's observations. It appears as a sudden increase at the peak of the H I profile, during just one integration period, approximately 15 Jy above the flux in adjacent channels and observations. No simultaneous increase was found in the other polarization, evidence that it was not an LO shift. A check of those channels over the entire observing period found no other peaks of interest. Although the feature matched the Wow frequency, it was too near the statistical noise peaks to consider as a redetection.

     Given that RFI was observed in some channels, we must presume that features are RFI in the absence of evidence otherwise, such as obvious resemblance to the Wow in frequency and strength, or a characteristic such as repetition. Excluding these features as probable RFI, no spectral features remain that noticeably exceed the noise.

3.4. Constraints on Periodic Sources

     Our observations would have detected a sufficiently strong source with a period shorter than 14 hr. For sources with longer periods, the probability of detection is shown in Figure 3. During a single run that probability is approximately 14/t, where t is the period in hours, which applies to the areas observed more than 15 north and south of the nominal positions, observed only one time. The nominal positions were observed twicedirectly, and partly covered during observations north and southfor a binomial probability of detection of over 0.90 for sources with periods of up to about 20 hr.

Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #28 on: September 03, 2007, 02:36:15 pm »



Fig. 3 CITED IN TEXT  |  HI-RES IMAGE (63kb)  |     

--------------------------------------------------------------------------------
Probability of detecting sources of various periods. The lower line represents one observing run of 14 hr, and the higher lines represent the binomial probability of detection for two, three, or four 14 hr runs, respectively.
Report Spam   Logged
Jennie McGrath
Superhero Member
******
Posts: 4349



« Reply #29 on: September 03, 2007, 02:37:34 pm »

 These constraints on period are statistical and therefore approximate, as a result of our small number of observing runs. A single 14 hr observing run, for example, would have no chance of detecting a source with a 24 hr period if the run happened to start just after the emission ended.

3.5. Wow and H I Frequency      

We note that the Wow frequency given by Kraus corresponds to the frequency of the peak H I emission found in our observations: both were some 20 kHz below the LSR, implying a radial velocity of approximately 4 km s-1. This could be taken as dynamical evidence that the Ohio State source was moving with the gas and hence was unlikely to be terrestrial interference, but that may be a coincidence. Since the Ohio State receiver had only 50 channels, there was a 2% chance that any detection would fall in the channel corresponding to the H I peak.
3.6. Rotating Sources in SETI
     Several SETI surveys have reported one-time detections with some characteristics expected of interstellar signals, including narrow bandwidth (<1 Hz) and the terrestrial Doppler signature (Horowitz & Sagan 1993; Colomb et al. 1995). None have been confirmed during follow-up observations, typically tracking the apparent source coordinates for on the order of 1 hr.
     One explanation that has been advanced is brief flux density enhancements of continuous emissions, caused by interstellar scintillation and noise variations (Cordes, Lazio, & Sagan 1997), but the probability of large scintillation gains is small. Rotating sources may provide an alternative explanation. Such sources might appear intermittent during transit survey observations and fail to be confirmed with brief reobservations, but they might prove easily repeatable with sufficiently extended observations perhaps a single "day" for planetary sources. While the length of extrasolar days is unknown and probably quite variable, observations over 25 hr would, for example, encompass the periods of more than half the planets in our solar system. It may be prudent to anticipate that the length of extrasolar planetary days could affect interstellar broadcasts.
4. CONCLUSIONS
     No signals resembling the Ohio State Wow were detected in observations dwelling for up to 14 hr at the coordinates where the signal was reported. This result constrains a putative periodic emission to a period greater than 14 hr because our observations were sufficiently sensitive and extended in time to detect emissions with a shorter period at least once. Since the probability of the original detection has been shown to be rather low (<0.10) for sources with periods over approximately 12 hr and a range of durations, the Ohio State detection would have been rather lucky if due to a periodic emission with a much longer period.
     Our observations cannot entirely rule out the possibility of a period longer than 14 hr or an emission that is not periodic at all. We also cannot rule out the possibility of a signal outside of the 2.5 MHz band we observed, although it was 5 times wider than Ohio State's frequency coverage and sufficiently wide to encompass the Doppler shifts expected of a planetary-based signal broadcast at the H I frequency.
     While these observations were undertaken to investigate the Ohio State signal, they also constitute a general search for periodic signals near H I over approximately 1 deg2 of sky, for periods as long as short planetary days. No prior SETI experiments appear to have searched for signals with such long periods (Tarter 1995). The sensitivity of the search was sufficient to detect a 1000 MW broadcast at over 100 lt-yr, assuming a 300 m transmitting antenna, or 1016 W radiated isotropically.
     R. G. thanks John Kraus, Robert Dixon, and Jerry Ehman at the Ohio State University Radio Observatory for providing information on the Wow emission. This work was supported in part by the SETI Institute.
REFERENCES
•   Baars, J. W. M., Genzel, R., Pauliny-Toth, I. I. K., & Witzel, A. 1977, A&A, 61, 99 First citation in article | NASA ADS
•   Colomb, F. R., Hurrell, E., Lemarchand, G. A., & Olald, J. C. 1995, in ASP Conf. Ser. 74, Progress in the Search for Extraterrestrial Life, ed. G. S. Shostak (San Francisco: ASP), 345 First citation in article
•   Cordes, J. M., Lazio, T. J. W., & Sagan, C. 1997, ApJ, 487, 782 First citation in article | Full Text | NASA ADS
•   Dixon, R. S. 1985, in IAU Symp. 112, The Search for Extraterrestrial Life: Recent Developments, ed. M. D. Papagiannis (Dordrecht: Reidel), 305 First citation in article
•   Dixon, R. S., & Cole, D. M. 1977, Icarus, 30, 267 First citation in article | NASA ADS
•   Ehman, J. R. 1998, The Big Ear Wow! Signal: What We Know and Don't Know About It after 20 Years, OSURO working paper First citation in article
•   Ellingsen, S. 1996, Ph.D. thesis, Univ. Tasmania First citation in article
•   Gray, R. H. 1994, Icarus, 112, 485 First citation in article | NASA ADS
•   Gray, R. H., & Marvel, K. B. 2001, ApJ, 546, 1171 First citation in article | Full Text | NASA ADS
•   Horowitz, P., Matthews, B. S., Forster, J., Linscott, I., Teague, C. C., Chen, K., & Backus, P. 1986, Icarus, 67, 525 First citation in article | NASA ADS
•   Horowitz, P., & Sagan, C. 1993, ApJ, 415, 218 First citation in article | NASA ADS
•   Kraus, J. D. 1979, Cosmic Search 1, 31 First citation in article
•   Lightman, A. P. 1984, Am. J. Phys., 52, 211 First citation in article | NASA ADS
•   Sulivan, W. T., Brown, C., & Wetherill, C. 1978, Science, 199, 377 First citation in article | NASA ADS
•   Tarter, J. 1995, Summary of SETI Observing Programs (Mountain View: SETI Institute) First citation in article
•   Thompson, A. R. 1991, in Interferometry and Synthesis in Radio Astronomy, ed. A. R. Thompson, J. M. Moran, & G. W. Swenson, Jr. (Malabar: Krieger), 264 First citation in article
•   Zombeck, M. V. 1990, Handbook of Space Astronomy and Astrophysics (2d ed.; Cambridge: Cambridge Univ. Press) First citation in article
________________________________________

http://www.journals.uchicago.edu/ApJ/journal/issues/ApJ/v578n2/54084/54084.html
Report Spam   Logged
Pages: 1 [2]   Go Up
  Print  
 
Jump to:  

Powered by EzPortal
Bookmark this site! | Upgrade This Forum
SMF For Free - Create your own Forum
Powered by SMF | SMF © 2016, Simple Machines
Privacy Policy