A 1.1–1.9 GHz SETI SURVEY OF THE KEPLER FIELD. I. A SEARCH FOR NARROW-BAND EMISSION FROM SELECT TARGETS

In the last 50 years, evidence has steadily mounted that the constituents and conditions we believe necessary for life are common and perhaps ubiquitous in the nearby galaxy. A plethora of prebiotic molecules have now been detected in molecular clouds, including amino acids and their precursors (Mehringer et al. 1997; Kuan et al. 2003), sugars (Hollis et al. 2004) and a host of other biologically important species (e.g., Lovas et al. 2006; Iglesias-Groth 2011). Such detections offer an indication that the reactants necessary for building large complex organic structures may be formed readily in proto-planetary environs. Exoplanets themselves, while once relegated to the domain of speculation, now appear to be common and numerous. While strictly Earth-size exoplanets at 1 AU from their Solar-type parents have so far eluded detection, we have evidence that the conditions necessary to maintain carbon-based life, e.g., liquid water, can exist far away from the traditional "habitable zone" (Carr et al. 1998).

As yet, no evidence exists for the presence of any kind of life outside of the Earth. However, on our own planet, life is known to have arisen early (within 1 Gyr) and flourished (Schopf et al. 2002). And while the propensity for evolution of intelligence from basic forms of life is not currently well understood, it appears that intelligence has imparted a strong evolutionary advantage to our own species. From a Copernican standpoint, the possibility that life has arisen elsewhere and perhaps evolved intelligence is plausible and warrants scientific inquiry. "Are we alone as technologically-capable intelligent beings?" is among the most profound questions we can ask as scientists, and observational astronomy represents the best means of determining an answer.

1.1. Engineered Radio Emission

SETI observations were performed during the period 2011 February–April using the Robert C. Byrd Green Bank Telescope (GBT) L-band (1.1–1.9 GHz) receiver and the Green Bank Ultimate Pulsar Processor (GUPPI) digital backend (P. Demorest et al. 2013, in preparation). For this experiment, GUPPI was configured in a novel "baseband recording" mode in which an entire 800 MHz band is digitized, channelized to 3.125 MHz with a 256-point polyphase filterbank and written to disk as 2-bit voltage data for both X and Y linear polarizations. The total aggregate data rate in this mode is 800 MBps. In order to obtain the requisite disk recording rates, the data stream was distributed to GUPPI's eight CPU/GPU computing nodes at 32 channels/node and eight 100 MHz bands were recorded separately. For the purposes of analysis, we have considered each 100 MHz band individually, hereafter Bands 0–7. We use this convention primarily because early technical problems with the GUPPI backend resulted in some computing nodes failing to record data, thus causing some observations to have non-contiguous frequency coverage in 100 MHz increments (see Table 3).

Targeted observations were performed on 86 Kepler Objects of Interest (KOIs) hosting planet candidates judged to be most amenable to the presence of Earth-like life, primarily judged by equilibrium temperature, but also cursory similarity to the Earth and the Solar system. These targets comprised KOIs hosting planet candidates in or near the traditional "habitable zone" (380 K > T_eq > 230 K) as described in Kasting et al. (1993), all KOIs hosting five or more planet candidates and all KOIs hosting a super-Earth (R_p < 3 R_⊕) in a >50 day orbit. Equilibrium temperatures were taken from Borucki et al. (2011), and the extended habitable zone range included in our search reflects their quoted uncertainty of approximately 22%. Temperatures were calculated assuming a Bond albedo, emissivity of 0.9, and a uniform surface temperature (see Borucki et al. 2011). An additional 19 KOIs located within a half power beam width of a primary target were observed serendipitously. Observations were performed using a cadence in which each target was observed interleaved with another target separated by Θ_min > 1° such that each target was effectively observed with an on-source, off-source, on-source sequence with minimal overhead. This technique is crucial for discerning a true astronomical signal from ubiquitous interference from human technologies. Details of the parameters of our observations are presented in Table 1. The observations discussed here represent part of a larger 24 hr campaign to search for technologically produced radio emissions in the Kepler field, which included both targeted observations and a raster scan of the entire field. Additional work, in preparation, will discuss a narrow-band search of the raster scan data and pulse searches over both targeted and raster observations.

To maximize the signal-to-noise of the detection of a distant continuous-wave transmitter the relative motion between the transmitter and receiver must be accounted for. As we have no a priori knowledge of the specific frequency of emission from an extraterrestrial technology, the overall Doppler shift in the received signal, dominated by the radial velocity of the source, is relatively unimportant for detection. However, the time rate of change of the Doppler shift, dominated by the orbital and rotational motions of the transmitter and receiver, must be considered to integrate ∼Hz spectra over many seconds. The Doppler drift is given simply by

$$\begin{equation} \dot{f} = \frac{{d\overrightarrow{V} }}{{dt}}\frac{{f_{\rm rest} }}{c} \end{equation} \tag{ 1 }$$

where $\overrightarrow{V}$ is the line of sight relative velocity between receiver and source, f_rest is the rest frequency of the transmitter, and c the speed of light. As a point of reference, the maximum contribution from Earth's orbital motion at 1 GHz is ∼ ± 0.02 Hz s⁻¹, and from Earth's rotation is ∼ − 0.1 Hz s⁻¹. If this effect is corrected for in power spectra, the worst-case minimum achievable spectral resolution for terrestrial observations is thus about Δν = 0.3 Hz (at 1 GHz). Channelization to any finer resolution would be ineffective as the received signal would be smeared over several channels. While the Doppler drift due to Earth's motion is known, the drift due to possible motion of the transmitter is largely unknown. For observations where t_obsδf > Δν, this necessitates searching various Doppler drift rates to achieve a $\sqrt{t_{\rm {obs}}}$ increase in sensitivity. As an aside, an arbitrary Doppler drift can be removed exactly in the voltage domain with no loss in sensitivity, via multiplication by an appropriate chirp function (Leigh 1998), but this technique is computationally infeasible for most blind searches, an exception being SETI@home (Korpela et al. 2002).

We accomplished a search for narrow-band features drifting at rates up to ∼ ± 10 Hz s⁻¹ using a modified form of the "tree" dedispersion algorithm, an algorithm originally developed for searching for dispersed pulsar emission (Taylor 1974). In much the same way that dispersed pulse searches seek to find power distributed along a quadratic curve in the time–frequency plane, a search for drifting sinusoids seeks to find approximately linearly drifting features in the same plane. The difference is simply one of the dimensions and orientation of the time–frequency matrix. The tree dedispersion algorithm accelerates these searches by taking advantage of the redundant computations involved in searching similar slopes, reducing the number of additions required from n² to n log₂n, where n is equal to both the number of spectra and number of slopes searched. Figure 1 shows a diagram of the "tree deDoppler" algorithm implemented for the drifting sinusoid search, shown here for four power spectra each having N frequency channels. The tree algorithm has fallen into disuse in the pulsar community due to the fact that it intrinsically sums only linear slopes, and modern broadband pulsar observations require a more exact quadratic sum to follow the ν⁻² cold plasma dispersion relation. In the case of Doppler drifting sinusoids, the linear approximation is very good and the tree algorithm is an excellent fit to the problem.

Zoom In Zoom Out Reset image size

Our implementation of the "tree deDoppler" algorithm necessitates 2^m spectra (an integer power of 2) and searches 2^m Doppler drift rates out to a maximum drift rate of:

$$\begin{equation} {\dot{f}}_{\rm {max}} = (\Delta \nu)^2 \end{equation} \tag{ 2 }$$

where Δν is the spectral resolution. The drift rate resolution is constrained to:

$$\begin{equation} \Delta \dot{f} = \frac{{\Delta \nu }}{{T_{\rm {obs}} }} \end{equation} \tag{ 3 }$$

where T_obs is the total observing time. We performed three channelizations at resolutions ranging from 0.75 Hz to 2.98 Hz, giving the maximum drift rates and drift resolutions shown in Table 2. The maximum drift rate searched would accommodate an equatorial transmitter on a planet 5 times larger and rotating 5 times faster than Earth.

Prior to Doppler searching, each 3.125 MHz polyphase channel was further channelized to ∼Hz resolution, detected and thresholded to search for narrow-band features. After detection, both polarizations were summed to form a single spectrum. Each high resolution spectrum constructed from individual polyphase channels was corrected for the filter response imposed by first stage (GUPPI) channelization by dividing through with a polynomial fit to an average bandpass. This polynomial fit bandpass was constructed a priori by fitting to a sum of many coarsely channelized spectra exhibiting low interference (based on visual inspection). All M spectra having length N from a single observation were then fit into a matrix sized to the nearest larger matrix having dimensions 2^m × N. Because we can assume that an untargeted narrow-band signal transmitted by an extraterrestrial technology will either be drifting due to acceleration in the host system or transmitted uncorrected for the Doppler acceleration at the receiving observatory, any narrow-band signal exhibiting no drift can be ruled out as likely coming from a terrestrial source. This concept is analogous to searches for pulsars in which sources exhibiting no dispersive sweep in their pulse profiles are likely pulsed terrestrial interference. The analogy allows us to again borrow from pulsar search techniques and apply a median filter for sources exhibiting no drift (see, e.g., Eatough et al. 2009; Siemion et al. 2012). After dividing each spectral channel by its median value, the tree deDoppler algorithm was applied in-place. Each Doppler corrected spectrum was then collapsed in time and searched for any summed spectral channel exceeding 25 standard deviations above the mean, assuming Gaussian statistics, with results inserted into a database. We use the term "detection" to refer to one measurement of a unique signal or emitter. Depending on source intensity, a single signal or emitter can be detected multiple times at different drift rates and bandwidths. The set of all detections for frequencies ν, standard deviations σ, drift rates ${\dot{f}}$, and bandwidths Δν was searched to identify the detection having the largest σ within each spectral window of width ${\dot{f}}_{\rm {max}}T_{\rm {obs}}$, and a time–frequency waterfall plot around this detection was extracted and stored with the corresponding database entry. We hereafter characterize each of the highest σ detections as "candidate signals." Ultimately we were left with approximately 3 × 10⁵ candidate signals.

Although the ISM is relatively unobtrusive to narrow-band radio emission, relative to, e.g., interstellar dust on optical light, and the atmosphere (including the ionosphere) is essentially transparent between 1 and 10 GHz, the signals being considered here are not wholly unperturbed by the intervening media. The ISM has been considered in the context of SETI for some time, notably in Cordes & Lazio (1991), Cordes et al. (1997) and references therein, with the principal results being as follows. In the strong scattering regime, narrow band sinusoids experience limited spectral broadening due to scattering in the inhomogeneous interstellar plasma, with a bandwidth Δν_broad equal to:

$$\begin{equation} \Delta \nu _{{\rm {broad}}} = 0.097{\rm {\ Hz\ }}\nu _{{\rm {GHz}}} ^{ - 6/5} {\left({\frac{{V_ \bot }}{{100}}} \right)} {\rm {SM}}^{ 3/5} \end{equation} \tag{ 4 }$$

Where V_⊥ is the transverse velocity of the source in km s⁻¹ and SM the scattering measure, a measure of the electron density fluctuations ${C_{n_e } ^2 }$, (cf. Rickett 1990) integrated along the line of sight:

$$\begin{equation} {\rm SM} = \int _0^L {C_{n_e } ^2 } (z)dz \end{equation} \tag{ 5 }$$

Further, intrinsically amplitude-stable narrow-band emission can be modulated in intensity up to 100% by strong scattering in the inhomogeneous plasma, with a characteristic time scale Δt_d equal to:

$$\begin{equation} \Delta t_d = 3.3{\rm {\ s\ }}\nu _{{\rm {GHz}}} ^{6/5} {\rm { }}{\left({\frac{{V_ \bot }}{{100}}} \right)}^{ - 1} {\rm {SM}}^{ - 3/5} \end{equation} \tag{ 6 }$$

Taking values of the SM from the "NE2001" electron density model (Cordes & Lazio 2002) for a center-field Kepler star at 0.5 kpc and assuming a transverse velocity of 25 km s⁻¹, we calculate Δt_d ≈ 3.5 hr and Δν_broad ≈ 20 μHz.

We note that the transition from strong to weak ISM scattering for our observing band occurs at a distance of about 900 ly in the direction of the Kepler field, putting many of our targets in the transition or weak scattering regimes. Although the expressions for Δν and Δt_d differ for these cases (see, e.g., Rickett 1990), the predicted broadening is similarly negligible, intensity modulations significantly lower in amplitude and modulation time scale longer in duration.

Depending on line of sight, the solar wind and interplanetary medium (IPM) can also impose significant spectral broadening on a transiting narrow-band signal. Radio scintillation due to the IPM had been known for some time prior to the discovery of strong scattering in the ISM, studied primarily through angular broadening of distant compact radio sources (see Narayan 1992 and references therein). The presence of spacecraft that could be used as monochromatic and coherent radio test sources allowed an additional probe of the IPM, notably providing a means to measure not just electron density fluctuations but also solar wind velocity (Woo 1978; Woo & Armstrong 1979), through observations of spectral broadening and phase scintillation of their narrow carrier signals. The strength of these effects depend largely on the solar impact distance R, or line-of-sight solar separation angle, but significant longitudinal and temporal (e.g., the solar cycle and coronal mass ejections) variations occur as well (Morabito 2009). Woo (2007) presents an assimilation of phase scintillation and spectral broadening observations of the S-band (2.3 GHz) carrier on Pioneer and Helios spacecraft at solar impact distances up to 200 R_☉ (adapted from Woo 1978), and find a roughly R^−9/5 dependence for spectral broadening past R ∼ 10 R_☉. A variety of models and observations suggest that the electron density fluctuations in the solar wind follow an approximate power law density spectrum, with a mean index very close to the Kolmogorov value of 5/3 (Morabito 2009 and references therein), allowing these results to be extrapolated based on Δν_broad scaling as ν^−6/5. Scaling based on the results in Woo (2007), we calculate a spectral broadening contribution from the IPM of approximately:

$$\begin{equation} \Delta \nu _{{\rm {broad}}} = 300 {\rm {\ Hz\ }}\nu _{{\rm {GHz}}} ^{ - 6/5} {\left({\frac{R}{R_{\odot }}} \right)} ^{-9/5} \end{equation} \tag{ 7 }$$

for solar impact distances greater than ∼10 R_☉.

For our observations, the nearest solar impact distance was ∼195 R_☉, giving an IPM spectral broadening contribution of ∼14 mHz.

For the parameters of this search, we can thus neglect spectral broadening due to either the ISM or IPM, and can assume a relatively steady flux for any intrinsically continuous and amplitude-stable signal over the course of our observing cadence. A key result of Cordes & Lazio (1991) was the suggestion that searches for narrow-band emissions in the strong scattering regime would have an increased likelihood of detecting a source by observing a sky location multiple times spanning many Δt_d. Although this is a very well justified strategy, the extra observing overhead associated with performing multiple on–off–on observation sequences did not permit it to be used here.

The principal complication in the otherwise straightforward data reduction involved in a narrow-band SETI experiment is the fact that human radio technology produces copious narrow band emission at ∼Hz scales. The existence of radio frequency interference is not unique to SETI experiments, of course, and over the years many techniques have been developed to mitigate its effects. It is worth noting, however, that in radio observations of most astrophysical phenomena, narrow-band features in a power spectrum can be immediately flagged and discarded because they are known to originate with technology rather than the target of the observation. Narrow-band radio SETI experiments face the more difficult task of determining whether a narrow-band feature originates with a human technology or distant intelligent life.

Our strategy to mitigate terrestrial interference was to demand that a candidate signal be both persistent and isolated on the celestial sphere. By observing in an on–off–on source cadence, we imposed this constraint by requiring that a given candidate signal be detected in both "on" source observations and not in the intervening "off" source observation. Observations in which one of the elements of the on–off–on cadence was not obtained due to technical problems were excluded completely. This technique was very effective, ruling out 99.96% of the candidate signals. Figure 2 shows a histogram of the number of detections versus signal-to-noise ratio for all detections, the detections representing the most significant detection of a single emitter (candidate signals) and only those detections passing the on–off–on automated interference excision algorithm. Figure 3 shows the number of detected signals as a function of topocentric frequency for the same detection groups. Time–frequency waterfall plots of the remaining 52 candidate signals were examined visually. Of these, 37 were ruled out immediately because candidates detected during pointings at many targets exhibited very similar modulation at nearby frequencies (±1 MHz). The remaining 15 signals were ruled out after querying the entire database of candidate signals for detections within ±1 MHz and identifying signals detected during pointings at other targets that closely resembled the modulation and drift properties of the candidate signals. Figure 4 shows two candidate signals that passed our initial on–off–on test, but were ruled out as interference based on their similar topocentric frequency and modulation. Figure 5 shows two candidate signals detected at different topocentric frequencies, but that were ruled out as interference based on their similar bandwidths and modulation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.1. Sensitivity

From the radiometer equation, the minimum detectable flux, F_i, of narrow-band emission detected in a single polarization is given by⁵:

$$\begin{equation} F_i = \sigma _{{\rm {thresh}}} S_{\rm {sys}} \sqrt{\frac{\Delta b}{t}} \end{equation} \tag{ 8 }$$

Where σ_thresh is the signal/noise threshold, S_sys is the system equivalent flux density (SEFD) of the receiving telescope, Δb is the spectral channel bandwidth and t the integration time. Assuming a flat 10 Jy SEFD for the GBT's L-band receiver, a characteristic sensitivity for the observations presented here is ∼2 × 10⁻²³ erg s⁻¹ cm⁻² or ∼3 Jy across a 0.75 Hz channel. A useful fiducial for considering the detectability of an extraterrestrial technology at radio wavelengths is the luminosity of the most powerful radio transmitter on Earth, the Arecibo Planetary Radar. Arecibo hosts two radar systems, a 430 MHz system capable of pulsed operation at ∼2.5 MW peak for a ∼5% duty cycle and a 2380 MHz system producing ∼1 MW continuous power. The equivalent isotropically radiated power (EIRP) of the higher frequency system, approaching 20 TW, is the larger of the two, owing to the ν² gain improvement. We consider this luminosity value L_AO ≈ 2 × 10²⁰ erg s⁻¹ to be approximately equal to the best-case radio emission from an Earth-level technology. Coincidently, L_AO is approximately the same as the

current total average power used by all humans on the planet Earth (Gruenspecht 2010). At 0.5 kpc, a 1 L_AO transmitter beamed in the direction of Earth would have a total flux of about 10⁻²⁴ erg s⁻¹cm⁻², placing our detectable limit for this range at ∼8 L_AO. Table 3 details total on-source times and sensitivities for all observed sources.

Table 3. Kepler Field Targeted Observations 2011 February–April

Objectsa	Band 0		Band 2b		Band 3		Band 4		Band 5		Band 6		Band 7
	1.1–1.2 GHz		1.33–1.4 GHz		1.4–1.5 GHz		1.5–1.6 GHz		1.6–1.7 GHz		1.7–1.8 GHz		1.8–1.9 GHz
	Ttotal	Speakc	Ttotal	Speak	Ttotal	Speak	Ttotal	Speak	Ttotal	Speak	Ttotal	Speak	Ttotal	Speak
	(s)	(erg s−1 cm−2)d	(s)	(erg s−1 cm−2)	(s)	(erg s−1 cm−2)	(s)	(erg s−1 cm−2)	(s)	(erg s−1 cm−2)	(s)	(erg s−1 cm−2)	(s)	(erg s−1 cm−2)
Kepler-10 b
	296	1.78	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74	311	1.73
Kepler-10 c
Kepler-11 b
Kepler-11 c
Kepler-11 d
	295	1.78	299	1.77	301	1.76	303	1.75	305	1.75	307	1.74	309	1.74
Kepler-11 e
Kepler-11 f
Kepler-11 g
Kepler-20 b
Kepler-20 c
Kepler-20 d	295	1.78	299	1.77	301	1.76	303	1.76	305	1.75	308	1.75	309	1.74
Kepler-20 e
Kepler-20 f
Kepler-22 b	295	1.78	300	1.77	302	1.76	304	1.76	306	1.75	308	1.75	...	...
Kepler-30 b
Kepler-30 c	295	1.78	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74	310	1.74
Kepler-30 d
Kepler-31 b
Kepler-31 c
	...	...	300	1.77	...	...	...	...	...	...	...	...	...	...
KOI 935.03
KOI 935.04
Kepler-32 b
Kepler-32 c
	...	...	...	...	...	...	...	...	...	...	302	1.75	305	1.74
KOI 952.03
KOI 952.04
KOI 51.01	293	1.78	297	1.77	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74
KOI 111.01
KOI 111.02	...	...	...	...	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 111.03
KOI 113.01	293	1.78	298	1.77	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74
KOI 174.01	293	1.79	298	1.77	300	1.77	302	1.76	304	1.76	306	1.75	308	1.75
KOI 211.01	295	1.78	299	1.77	...	...	...	...	...	...	...	...	306	1.74
KOI 260.01
	295	1.78	299	1.77	301	1.76	303	1.76	306	1.75	308	1.75	310	1.74
KOI 260.02
KOI 314.01
KOI 314.02	295	1.78	299	1.77	301	1.76	303	1.76	306	1.75	308	1.75	309	1.74
KOI 314.03
KOI 351.01
KOI 351.02	295	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.74	309	1.74
KOI 351.03
KOI 365.01	...	...	297	1.78	...	...	...	...	...	...	304	1.75	306	1.75
KOI 372.01	295	1.78	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74	310	1.74
KOI 374.01	...	...	...	...	...	...	...	...	302	1.75	304	1.75	306	1.74
KOI 375.01	...	...	...	...	...	...	...	...	...	...	304	1.75	306	1.74
KOI 386.01
	296	1.78	301	1.77	303	1.76	305	1.75	307	1.75	309	1.74	311	1.74
KOI 386.02
KOI 401.01
	296	1.78	301	1.76	303	1.76	289	1.79	307	1.75	309	1.74	311	1.73
KOI 401.02
KOI 416.01
	297	1.78	301	1.77	603	1.76	607	1.75	612	1.75	616	1.74	620	1.73
KOI 416.02
KOI 422.01	...	...	...	...	...	...	...	...	305	1.75	307	1.75	310	1.74
KOI 433.01
	294	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 433.02
KOI 448.01
	295	1.78	299	1.77	301	1.77	303	1.76	305	1.75	307	1.75	309	1.74
KOI 448.02
KOI 465.01	...	...	299	1.77	...	...	...	...	...	...	...	...	...	...
KOI 500.01
KOI 500.02
KOI 500.03	295	1.78	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74	310	1.74
KOI 500.04
KOI 500.05
KOI 536.01	294	1.78	299	1.77	...	...	...	...	...	...	...	...	...	...
KOI 542.01
	...	...	300	1.76	...	...	303	1.75	305	1.75	307	1.74	309	1.74
KOI 542.02
KOI 555.01
	295	1.78	299	1.77	301	1.76	294	1.78	306	1.75	308	1.75	...	...
KOI 555.02
KOI 564.01
	...	...	...	...	300	1.76	287	1.77	304	1.75	306	1.75	308	1.74
KOI 564.02
KOI 590.01
	295	1.78	299	1.77	...	...	...	...	...	...	308	1.75	310	1.74
KOI 590.02
KOI 618.01	...	...	...	...	...	...	...	...	...	...	303	1.75	305	1.75
KOI 622.01	...	...	...	...	...	...	...	...	...	...	301	1.75	304	1.75
KOI 682.01	295	1.78	599	1.77	302	1.76	304	1.75	306	1.75	308	1.74	310	1.74
KOI 683.01	296	1.78	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74	310	1.74
KOI 698.01	...	...	299	1.77	...	...	...	...	...	...	...	...	...	...
KOI 701.01
KOI 701.02	...	...	...	...	...	...	...	...	301	1.75	303	1.75	305	1.74
KOI 701.03
KOI 711.01
KOI 711.02	...	...	...	...	302	1.76	304	1.76	306	1.75	308	1.75	310	1.74
KOI 711.03
KOI 741.01	...	...	...	...	...	...	...	...	...	...	303	1.75	305	1.75
KOI 812.01
KOI 812.02
	296	1.78	301	1.76	303	1.76	295	1.78	307	1.75	309	1.74	311	1.73
KOI 812.03
KOI 812.04
KOI 817.01
	296	1.78	300	1.76	302	1.76	292	1.78	307	1.75	309	1.74	311	1.73
KOI 817.02
KOI 826.01	296	1.78	300	1.77	302	1.76	295	1.76	306	1.75	308	1.74	...	...
KOI 847.01	294	1.78	298	1.77	301	1.77	303	1.76	305	1.75	307	1.75	309	1.74
KOI 854.01	295	1.78	299	1.77	301	1.76	304	1.76	306	1.75	308	1.75	310	1.74
KOI 882.01	294	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 892.01	295	1.78	299	1.77	301	1.76	303	1.76	306	1.75	308	1.75	309	1.74
KOI 902.01	...	...	...	...	299	1.76	301	1.76	303	1.75	305	1.75	308	1.74
KOI 947.01	...	...	...	...	...	...	294	1.77	311	1.72	313	1.72	315	1.71
KOI 974.01	294	1.78	299	1.77	...	...	...	...	...	...	304	1.75	306	1.75
KOI 986.01
	296	1.77	301	1.76	303	1.76	299	1.75	307	1.74	309	1.74	311	1.73
KOI 986.02
KOI 998.01	294	1.78	298	1.77	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74
KOI 1010.01	294	1.78	298	1.77	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74
KOI 1032.01	294	1.78	298	1.77	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74
KOI 1099.01	296	1.78	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74	311	1.73
KOI 1113.01
	296	1.77	301	1.76	303	1.76	299	1.75	307	1.74	309	1.74	311	1.73
KOI 1113.02
KOI 1118.01	296	1.78	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74	311	1.73
KOI 1159.01	...	...	...	...	...	...	...	...	...	...	303	1.75	305	1.75
KOI 1162.01	296	1.78	300	1.77	302	1.76	...	...	...	...	...	...	...	...
KOI 1168.01	...	...	300	1.77	302	1.76	...	...	...	...	...	...	...	...
KOI 1192.01	294	1.78	298	1.77	...	...	...	...	...	...	...	...	...	...
KOI 1199.01	295	1.78	299	1.77	301	1.77	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1203.01
KOI 1203.02	295	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1203.03
KOI 1208.01	295	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1210.01	295	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1226.01	295	1.78	299	1.77	301	1.77	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1261.01
	...	...	...	...	...	...	...	...	...	...	307	1.75	309	1.74
KOI 1261.02
KOI 1268.01	295	1.78	299	1.77	301	1.76	303	1.76	306	1.75	308	1.75	310	1.74
KOI 1302.01	292	1.79	296	1.78	298	1.77	300	1.76	303	1.76	305	1.75	...	...
KOI 1328.01	296	1.78	300	1.76	302	1.76	304	1.75	306	1.75	309	1.74	310	1.73
KOI 1355.01	297	1.78	599	1.77	603	1.76	600	1.75	611	1.75	616	1.74	620	1.74
KOI 1358.01
KOI 1358.02	294	1.78	299	1.77	301	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1358.03
KOI 1361.01	...	...	...	...	302	1.76	295	1.78	306	1.75	308	1.75	310	1.74
KOI 1372.01	296	1.78	300	1.76	302	1.76	304	1.75	306	1.75	308	1.74	310	1.73
KOI 1375.01	295	1.78	300	1.77	302	1.76	304	1.76	306	1.75	308	1.75	309	1.74
KOI 1377.01	297	1.78	599	1.77	603	1.76	600	1.75	611	1.75	616	1.74	620	1.74
KOI 1379.01	297	1.78	599	1.77	603	1.76	600	1.75	611	1.75	616	1.74	620	1.74
KOI 1423.01	...	...	...	...	...	...	300	1.76	303	1.75	...	...	...	...
KOI 1426.01
KOI 1426.02	...	...	300	1.77	302	1.76	304	1.76	306	1.75	308	1.75	310	1.74
KOI 1426.03
KOI 1429.01	...	...	...	...	...	...	297	1.78	303	1.75	305	1.75	307	1.74
KOI 1463.01	297	1.78	301	1.77	303	1.76	305	1.75	307	1.75	309	1.74	615	1.74
KOI 1472.01	...	...	...	...	...	...	292	1.78	304	1.75	306	1.74	308	1.73
KOI 1478.01	295	1.78	299	1.77	...	...	...	...	...	...	304	1.74	306	1.74
KOI 1486.01
	295	1.78	299	1.77	301	1.76	295	1.78	306	1.75	308	1.75	310	1.74
KOI 1486.02
KOI 1503.01	296	1.78	300	1.77	302	1.76	304	1.76	306	1.75	308	1.75	310	1.74
KOI 1508.01	...	...	...	...	...	...	292	1.78	304	1.75	306	1.74	308	1.73
KOI 1527.01	295	1.78	299	1.77	301	1.76	303	1.76	305	1.75	308	1.75	309	1.74
KOI 1528.01	...	...	...	...	...	...	292	1.78	304	1.75	306	1.74	308	1.73
KOI 1535.01	...	...	300	1.76	...	...	303	1.75	305	1.75	307	1.74	309	1.74
KOI 1561.01	295	1.78	300	1.77	302	1.76	304	1.75	306	1.75	308	1.74	310	1.74
KOI 1564.01	296	1.78	300	1.77	302	1.76	295	1.76	306	1.75	308	1.74	...	...
KOI 1574.01	294	1.78	298	1.77	300	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 1582.01	295	1.78	299	1.77	...	...	...	...	...	...	...	...	...	...
KOI 1596.01
	295	1.78	...	...	...	...	...	...	...	...	307	1.74	309	1.74
KOI 1596.02
KOI 1598.01
KOI 1598.02	293	1.78	297	1.77	299	1.76	301	1.76	303	1.75	306	1.75	308	1.74
KOI 1598.03
KOI 1648.01	...	...	...	...	...	...	300	1.76	303	1.75	...	...	...	...
KOI 1749.01	...	...	...	...	300	1.76	287	1.77	304	1.75	306	1.75	308	1.74
KOI 1819.01	...	...	...	...	302	1.76	304	1.76	306	1.75	308	1.75	310	1.74
KOI 2248.01
KOI 2248.02
	...	...	...	...	...	...	297	1.78	303	1.75	305	1.75	307	1.74
KOI 2248.03
KOI 2248.04
KOI 2418.01	295	1.78	...	...	...	...	...	...	...	...	307	1.74	309	1.74
KOI 2493.01	294	1.78	298	1.77	300	1.76	303	1.76	305	1.75	307	1.75	309	1.74
KOI 2534.01
	297	1.78	302	1.76	304	1.76	296	1.78	308	1.75	310	1.74	312	1.73
KOI 2534.02

Notes. ^aObjects in italic text were observed serendipitously with a primary target, see the text. KOI 326, one of our original targets, has been omitted from the table as its single identified planet candidate was determined to be a false positive. ^bWe exclude Band 1 here, as the band 1200–1330 MHz was not searched due to the presence of a bandpass filter used to mitigate heavy aircraft radar interference contaminating this region. ^cPeak sensitivity quoted for 0.75 Hz channelization. ^d/ 10⁻²³.

Download table as:  ASCIITypeset images: 1 2 3 4 5

Figure 6 shows the range at which a transmitter similar to the Arecibo Planetary Radar (∼5 × 10¹³ erg s⁻¹ transmitted through a 305 m parabolic reflector) could be detected using the parameters of this experiment (150 s integrations, 0.75 Hz channelization) applied to all heterodyne receivers at the GBT. These limits also apply to an intrinsically uncertainty-limited broadband pulse having approximately 1000 times the total radiated energy, broadened assuming the "NE2001" ISM model (Cordes & Lazio 2002) to t = 0.17 μs with pulse bandwidth = 800 MHz centered on our observing band. Here we have used system temperature and gain values from the GBT Proposer's Guide, neglected the galactic synchrotron background and for frequencies above 15 GHz we assume a 50% weather quantile. We have assumed a 25σ detection threshold, as was used in the analysis described here. The approximate median distance to a Kepler catalog star is also indicated. Although Figure 6 suggests higher frequencies might be preferred, again owing to the transmitter gain improvement, the additional scheduling difficulties due to weather constraints and pointing correction overhead make lower frequency observations more tractable at present.

Zoom In Zoom Out Reset image size

We use the Arecibo Planetary Radar example simply as a point of reference. While this transmitter is highly directional and the probability of interception is thus fairly low, observing systems in which the ecliptic plane is viewed edge-on increases the probability of detecting a radar used for local planetary system ranging and imaging. If we assume that an exo-Arecibo has a duty cycle similar to Earth's, a characteristic ecliptic (±5°) illumination of about 100 hr year⁻¹ (P. Perillat 2012, private communication), the overall probability of being in the radar beam during an observation is ∼2 × 10⁻⁸. While this figure is indeed low, it represents an order of magnitude improvement over an isotropic assumption.

Our search of 104 KOIs identified no evidence of advanced technology indicative of intelligent life. If we assume a low false positive rate for KOIs, in the simplest terms this result indicates that fewer than ∼1% of transiting exoplanet systems are radio loud in narrow-band emission between 1 and 2 GHz at the ∼8 L_AO level. If we take the orbital inclination requirement for detection to be ±5° and conservatively estimate the total fraction of FGK stars hosting planets of any type to be ∼15% (Marcy & Howard 2011), we estimate that fewer than ∼10⁻⁴ FGK stars host civilizations detectable via orbital plane narrow-band radio emission in the same band and luminosity. For the GBT, this implies a surface density of < ∼ 5 × 10⁻² deg⁻² detectable sources. For the upcoming Square Kilometer Array, a facility that will be perhaps 100 times more sensitive than the GBT, the similar surface density of detectable sources is < ∼ 100 deg⁻² (l = 0, b = 0; Robin et al. 2003).

Although the observations described here were part of a campaign targeting specific KOIs, the size of the telescope beam probed a much larger population of stars at a concomitantly higher luminosity limit. When probing advanced technology luminosities that represent a reasonable extrapolation of terrestrial technology, i.e., ≈Kardashev type I (Kardashev 1964), describing the target population and quantifying limits based on number of Sun-like stars or number of Earth-like planets is quite logical. However, when we begin to probe luminosities (and energy usage) that are many orders of magnitude larger than Earth's, our uncertainty in the bounds of life in general render these measures inadequate. For large luminosities, total stellar mass is a much more useful measure of the amount of energy-delivering capacity of a surveyed area. Integrating the GBT's beam out to and encompassing the Milky Way's halo stars (Gnedin et al. 2010), a characteristic total mass is ∼5 × 10³ M_☉. At 80 kpc, our sensitivity equates to an EIRP limit of ∼10⁵ L_AO, or approximately an order of magnitude larger than the total solar insolation incident on Earth. A civilization capable of truly isotropic emission at these power levels would likely be capable of harnessing vastly greater amounts of energy from their parent sun than incident on their home planet, and thus would be approaching the Kardashev type II class. Taking our 86 observations as independent samples of a ∼5 × 10³ M_☉ column, we estimate the number of 1–2 GHz narrow-band-radio-loud Kardashev type II civilizations in the Milky Way to be ${<}10^{-6}\ M^{-1}_ \odot$.

Ultimately, experiments such as the one described here seek to firmly determine the number of other intelligent, communicative civilizations outside of Earth. However, in placing limits on the presence of intelligent life in the galaxy, we must very carefully qualify our limits with respect to the limitations of our experiment. In particular, we can offer no argument that an advanced, intelligent civilization necessarily produces narrow-band radio emission, either intentional or otherwise. Thus we are probing only a potential subset of such civilizations, where the size of the subset is difficult to estimate. The search for extraterrestrial intelligence is still in its infancy, and there is much parameter space left to explore. The exponential growth in semiconductor technology over the last decades has been an incredible boon to SETI experiments, allowing orders of magnitude improvements in spectral coverage. Within the next decade, we will have the ability to examine significantly larger portions of the electromagnetic spectrum, including instantaneous analysis of the entire 10 GHz of the terrestrial microwave window. In addition to radio searches, new technology will extend SETI into regions of the electromagnetic spectrum never before observed with high sensitivity (Siemion et al. 2011). Extending searches to encompass much larger classes of signals is crucial to producing robust and meaningful limits.

We thank John Ford and Scott Ransom for technical assistance during our observations and Gerry Harp for comments on an early draft of this manuscript. The work presented here was partially funded by NASA Exobiology Grant NNX09AN69G and donations from the Friends of Berkeley SETI and the Friends of SETI@home. We also acknowledge the financial and intellectual contributions of the students, faculty and sponsors of the Berkeley Wireless Research Center. This research used resources of the National Energy Research Scientific Computing Center, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH1123. A.P.V.S. gratefully acknowledges receipt of student observing support from the National Radio Astronomy Observatory. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.