1. Ionospheric propagation effects for UHE Neutrino Detection using the lunar Cherenkov technique Dr Rebecca McFadden 1,2 Prof Ron Ekers 2 Justin Bray 2,3 1 ASTRON Netherlands Institute for Radio Astronomy, The Netherlands 2 CSIRO Astronomy and Space Science, Australia 3 University of Adelaide, Australia Moon image: Luc Viatour / www.Lucnix.be

2. Outline UHECR - UHE Neutrino Relationship Lunar Cherenkov Technique Characterisitics of Cherenkov Radio Pulse Experiments Ionospheric Calibration Techniques Future Work/Experiments

3. CR Spectrum

4. Lower energy events attributed to:  energetic processes in the sun  particles being accelerated in the shock fronts of supernova remnants

5. CR Spectrum ?

6. GZK cutoff predicts UHECR will interact with CMB photons and lose energy. ? GZK Neutrinos UHE protons interact with CMBR (GZK Interactions) Neutrino secondaries Should not observe extra galactic UHECR with distances greater than ~100Mpc Neutrinos will be produced in these interactions

7. Correlation of UHECR and nearby AGN UHECR * AGN (Cen A in white) Galactic centre Supergalactic plane

8. Back to the neutrinos … Neutrinos are produced during both the acceleration and propagation of UHECR and therefore provide an alternative method for investigating their origin. They make good messengers because:  They are weakly interacting (unlike UHECR)  Neutrinos are not deflected by magnetic fields and so should point back to where they are produced  Significant information on the UHECR spectrum is expected to be preserved in the spectrum of GZK neutrinos Some likely candidates for UHE neutrino production … GRB AGN - jets - disk Particle decay: SHDM particles (Most scenarios ruled out) Accelerating mechanisms:

9. UHE Neutrino Detection using the Lunar Cherenkov Technique neutrino Negative charge excess produces coherent Cherenkov radiation UHE particle interaction causes a cascade of secondaries Technique first proposed by Dagkesamanskii & Zhelezynkh (1989) and first applied by Hankins, Ekers and O'Sullivan (1996) using the Parkes radio telescope

10. Coherent Cherenkov Radiation ~3m ~10cm The angular distribution of the radio emission can be interpreted as a Fraunhofer diffraction pattern of the charge excess in the shower (Halzen et al 1992). The electric field spectrum rises linearly with frequency until the destructing interference effects start to take place (at the decoherence frequency or when the wavelength becomes comparable to the lateral distribution of the shower) and a maxmum is induced.

11. Characteristics of Radio Emission Pulse Profile Very narrow unmodulated pulse (ns scale) Broad receiver bandwidth for ns time resolution High speed sampling Real time trigger to avoid excessive storage requirements Spectrum Broadband continuous emission Coherent emission process Peak depends on viewing angle (< 5GHz) Coherent Regime Decoherence

12. Ionospheric Dispersion Low frequencies High frequencies Night, solar cycle min Day, solar cycle max Ionospheric dispersion destroys the characteristic coherency of the pulses The effect is worse during the day, at low frequencies, and at solar cycle max Image: C.James

13.  Advantages of using an Array of Small Radio Telescopes A rray geometry allows RFI discrimination based on the signal direction of arrival. The system temperature is dominated by thermal emission from the moon so there is little improvement in signal to noise ratio to be gained by larger apertures.  With many small dishes, their spatial separation can be used to ensure that thermal emission from the moon is incoherent between elements thus increasing the signal to noise ratio. At frequencies in the 1-2GHz range, the primary beam produced by a 22m aperture (such as the Compact Array) can see the entire moon. Beamwidth is inversely proportional to aperture size, therefore a larger dish will see less than the entire moon and any minimal improvement in sensitivity from the larger aperture will be offset by decreased coverage of the moon.

14. Australia Telescope Compact Array Earth-rotation aperture synthesis radio interferometer Consists of six 22m antennas Five can be positioned at any of 40 stations along a 3 km east-west track The sixth antenna is fixed on a station three km further west

15. ATCA Experimental Set-Up Data Channels: 3 antennas. Dual linear polarisation. 600 MHz Bandwidth (1.2-1.8 GHz). 8-bit, 2 GHz sampling Triggering: Independent at each antenna. Adjustable threshold trigger on each polarisation. If either polarisation exceeds the threshold, 128ns buffers from both are recorded. 256 MHz clock time also recorded. Max rate 1 kHz / antenna. BasebandBuffer 1 Gbit Ethernet Link (TCP/IP) Offline Processing

16. System Detail at Each Antenna System Detail at Each Antenna LNA 1.5 Ghz LNA 2.3 Ghz LNA 2.3 Ghz LNA 1.5 GHz Polarisation and Band Splitter L-Band RF Splitter Module (F10) S-Band Splitter (C23) L-Band Splitter (C22) S-Band RF Splitter Module (F11) 1.2 – 2.5 Ghz Horn Pol A Pol B Pol A Pol B Analog De-dispersion Filters 2.048 Gs/s 8 bit ADC Buffer Ethernet Connection Adjustable Threshold Detect 5.1 σ4096 samples, 2μs 8 bit ADC CABB Sampler Board ATNF BCC Interface 100Mb/s Standard ATCA Receiving Path Customised Hardware Photo by Graeme Carrad

17. Ionospheric Dispersion The ionosphere is most stable between sunset and sunrise. Typical Night-time Dispersion: (for May/June during solar min) Mean vertical dispersion ~4ns over 1.2- 1.8 Ghz for 10pm-6am Δt = 0.00134 TEC (1/f lo 2 – 1/f hi 2 ) TEC data from NASA’s Crustal Dynamics Data Information System TEC data from NASA’s Crustal Dynamics Data Information System Signal de-dispersion is performed in analog microwave filters with a fixed dispersion characteristic. TEC depends on solar activity: Diurnal cycle Diurnal cycle Season Season 11 year solar cycle 11 year solar cycle

18. De-dispersion Method Hardware De-dispersion Filter Designed by Paul Roberts (ATNF) Signal de-dispersion is performed in analog microwave filters with a fixed dispersion characteristic. These filters were designed using a new method of planar microwave filter design based on inverse scattering. This results in a filter with a continuously changing profile, in this case a microstrip line with continuously varying width. This results in a filter with a continuously changing profile, in this case a microstrip line with continuously varying width. The width modulations on the microstrip line produce cascades of reflections which sum to produce the desired frequency response. Thick Thin Filters set for characteristic VTEC at 60° elevation.

19. New Method for Atmospheric Dispersion Calibration Proposed at the Merida ICRC 2007 (McFadden et. al.) and initially developed for the LUNASKA (see #240) Collaboration using the Australia Telescope Compact Array. Ionospheric TEC can be deduced from Faraday rotation measurements of a polarised source combined with geomagnetic field models. We propose to use this technique, with the polarised thermal radio emission from the lunar limb as our polarised source, to obtain instantaneous and line-of-sight TEC measurements. Lunar polarisation distribution – radially aligned Faraday rotation from the ionosphere

20. Lunar Polarisation Distribution Lunar (planetary) polarisation distribution is due to Brewster angle effects and the changing viewing angle (with respect to the planetary surface normal). First described by Heiles and Drake (1963)

21. Lunar Polarisation Distribution measured by SPORT collaboration (2002)

22. New Method for Atmospheric Dispersion Calibration Using a radio interferometer, Faraday rotation analysis usually starts with imaging to spatially resolve the data. Lunar imaging requires: - short baselines (antenna spacings) - many hours of data to obtain data samples as the Earth rotates (Earth rotation aperture synthesis) NOT ideal for the Lunar Cherenkov technique. Therefore, we have had to develop Faraday rotation estimation techniques in the visibility domain (i. e. data taken in the far field and not converted to the imaging/source plane).

23. Lunar Polarisation Studies with B Sault, University of Melbourne Lunar (planetary) polarisation distribution is due to Brewster angle effects and the changing viewing angle (wrt to planetary surface normal). First described by Heiles and Drake (1963) and more recently measured by SPORT (2002) PA Dist QURing Variation on Hankel transform - angular dependence - angular dependence - trivial radial dependence - trivial radial dependence Standard Hankel transform - radial dependence - no angular dependence

24. Lunar Polarisation Studies Lunar (planetary) polarisation distribution is due to Brewster angle effects and the changing viewing angle (wrt to planetary surface normal). First described by Heiles and Drake (1963) and more recently measured by SPORT (2002)

25. Stokes Parameter modelling in MIRIAD U Q Position Angle Distribution Stokes Distributions: Data generated in MIRIAD, using uvgen – this method also produces full uv data sets

26. Modelled Visibility Data Data generated in MIRIAD using uvgen – ‘moon’ placed at South Pole to look at U,Q structure without baseline projection effects

27. Visibility domain Stokes distributions U Q Stokes Distributions: Ring (for comparison) Lunar Distribution Radiation pattern in the far field of the moon (ie uv plane) Data generated in Matlab - to model images in the visibility domain

28. Visibility domain PA Distribution U Q LunarDistribution uv plane distribution Data generated in Matlab - to model images in the visibility domain not to scale (zoomed in to show structure)

29. PA Measurements in the visibility domain Position Angle Distribution in the uv plane (not to scale) Position Angle map generated in Matlab Visibility domain Position Angle measurements from data modelled in Miriad

30. PA Measurements in the visibility domain Position Angle map UV plane Position Angle measurements from Modelled data For and East-West array, baseline projection traces an ellipse through the uv plane

31. PA Measurements in the visibility domain Position Angle map generated in Matlab UV plane Position Angle measurements from data modelled in Miriad Baseline projection traces an ellipse through the uv plane

32. PA Measurements in the visibility domain Position Angle map generated in Matlab UV plane Position Angle measurements from data modelled in Miriad Baseline projection traces an ellipse through the uv plane

33. Mathematically … Position Angle Distributions (not to scale)

34. Mathematically … Position Angle Distributions (not to scale) Ring (for comparison) Variation on Hankel transform - angular dependence - angular dependence - trivial radial dependence - trivial radial dependence Standard Hankel transform - radial dependence - no angular dependence

35. Position Angle Measurements Simulation Real Data Data generated in MIRIAD using uvgen and imported into Matlab for analysis (no Faraday rotation) Real data taken on the ATCA, Sep 2008

36. Faraday Rotation Estimates Real Data (ATCA Sep 2008) – assume deviations in linear uv angle vs PA relationship are due to Faraday Rotation

37. GPS Comparison (TEC)

38. Format of Ionospheric Products Conversion of VTEC to Slant TEC via Single Layer Model

39. Discrepencies ? GPS vs. foF2 Possible Explanations: - Unreliable measurements on the shortest baselines - Missing TEC component from plasmasphere

40. Nanosecond Pulse Generators Analogue – ECL gate triggered by a 5ns square wave and filtered with a bandpass Digital – FPGA followed by 2GHz DAC

41. ATCA Experiment Results Results from May 2007 We can use the dispersion to discriminate against terrestrial RFI E.g. we were observing regular, strong triggers of ~20ns duration (opposite). Was the origin: A) Terrestrial (no dispersion) B) Lunar (characteristic dispersion) C) Moon bounce (double dispersion)? Most detected RFI events however are narrowband … A) Terrestrial Impulse? B) True Event? C) moon-bounce? Predictions – from a 3am Analysis Observed Trigger

42. Timing Calibration 2 GHz clocks on each antenna with unknown relative offset. Point antennas at a point source (bright quasar). Trigger off the noise cal pulse. Correlate resulting buffers between antennas. (Chris Phillips, ATNF)

43. Field Programmable Gate Array Semiconductor device containing programmable logic components Can be programmed to duplicate logic functions (such as AND, OR, NOT, XOR) Or more complex functions such as decoders or simple math functions (DSP applications: filtering, transforms, modulation) A Hardware Description Language (HDL) can be used to define the FPGA's behaviour High level design tools may be used to raise the level of abstraction in the design process

44. Introduction to ATCA Upgrade Upgrade planned to increase current 600MHz bandwidth to 2 GHz – Compact Array Broad Band Upgrade (CABB) Due to bandwidth and data storage limitations, the only way to exploit this new bandwidth is by implementing real time detection algorithms These algorithms can be programmed into more complex FPGA boards currently being built by the ATNF for the CABB Upgrade

45. Parkes Experiment 1.2-1.5 MHz Multi-beam Rx - 3 on-Moon beams (1 on Cen A position) - 1 off-Moon beam Digital dedispersion 5 x Xilinx 5 FPGAs

46. Additional Processing Power from FPGA Upgrade Wavefront D   Path = D sin  Delay = D/c sin   Wavefront 2 We can form multiple Beams to Cover the Limb of the Moon It will also be possible to implement real time de-dispersion algorithms in this hardware and we have developed a technique for obtaining measurements of the ionospheric TEC which are both instantaneous and line-of-sight to the lunar observations. The ionospheric TEC can be deduced from Faraday Rotation measurements of a polarised source combined with geomagnetic field models, which are more stable than ionospheric models when the direction of signal propagation is parallel to the geomagnetic field vector. We propose to use this technique, with the polarised thermal radio emission from the lunar limb as our polarised source, to obtain instantaneous and line-of-sight TEC measurements.

47. ATCA Experiment Summary May 5th-7th 2007 (5 hr) 10pm-6am. Simultaneous baseband recording (64 Mhz at 1.2 GHz). Targeted Sag A region. Centre-pointing. Only ms timing available! Feb 26th-28th 2008 (14 hr) Repeat of ’07 set-up. 14 hrs stable configuration. March 17th-19th 2008 (20 hr) Targeted Cen A. Pointed at limb to maximise Cen A sensitivity. 2007 Data Reduction 120,000 candidates per antenna. 300 remain after 1 ms coincidence and ‘dumb’ RFI cuts. 4 remain after polarisation + ‘smart’ RFI cuts. We expect ~6 false events. ASSUME: No neutrinos. 2008 Analysis ~100 000 triple coincidence events 60 from the direction of the Moon All attributed to RFI/near-field events

48. Conclusions and Future Work The Lunar Cherenkov technique uses the moon as a large volume detector for UHE neutrinos. Real time dedispersion is required to recover the full pulse amplitude. Knowledge of the ionospheric dispersion characteristic can also be used as a ‘signature’ of lunar events. We propose to use lunar polarisation measurements to calibrate ionospheric dispersion effects line of sight to the moon and in real time. Our technique exploits the radial symmetry of polarised planetary emission and uses the interferometric response as an angular spatial filter to obtain meaningful position angle measurements in the visibility domain. A new dataset has already been taken using the WRST to explore baseline dependence of data quality. Strategies to calibrate plasmaspheric TEC content are being considered and require further development. This technique must also be adapted for use with other instruments such as LOFAR (NuMoon project, see #86, #1042).

49. Format of Ionospheric Products Conversion of VTEC to Slant TEC via Single Layer Model

50. GPS vs. foF2 foF2 data – available near real time, error < 5 TECU GPS data – raw STEC measurements, error < 1 TECU VTEC measurements, available with a few days delay*, error < 2-3 TECU Greater accuracy may be possible by converting from satellite to source STEC in the ‘Slant Domain’

51. Work conducted by the Lunatic team will form a pathway for UHE neutrino detection using the proposed SKA radio telescope. The Square Kilometre Array will be 100 times more sensitive than the best present day radio instruments. The current designs proposed for the SKA consist of large numbers (~10 4 ) small dishes (6-12m) to achieve a square kilometre of collecting area in the 0.1-3GHz range. Lunar UHE Neutrino Astrophysics with the Square Kilometre Array (LUNASKA)

52. UHE Neutrinos (~10 20 eV) Cosmic Ray energy spectrum extends from below 10 10 eV to at least 10 20 GeV In the rest frame of an UHECR proton, the CMB are blue shifted to gamma ray energies and the threshold for Bethe-Heitler pair production and pion photoproduction is reached. The origin of UHECR above this threshold (GZK cut-off) is not fully understood. For extragalactic UHECR almost all spectral information above the GZK cut-off is lost however significant information is preserved in the spectrum of neutrinos. UHE neutrino astronomy will be able to provide more insight into the origin of UHECR.

53. Neutrino-induced Showers Deep inelastic scattering neutrino interactions