Toward Hybrid Optical/Radio/Acoustic Detection of EeV Neutrinos

1. Toward Hybrid Optical/Radio/Acoustic Detection of EeV Neutrinos Justin Vandenbroucke (UC Berkeley, justinav@berkeley.edu) with Dave Besson Sebastian Böser Rolf Nahnhauer Rodín Porrata Buford Price 2nd Workshop on ≥TeV Particle Astrophysics, Madison, August 30 2006

2. The goal: GZK physics with an IceCube extension at South Pole • ~100 GZK events (e.g. 10 yrs @ 10/yr) would give a quantitative measurement including energy, angular, and temporal distributions • Non-optical techniques must be used at these energies and their systematics are not well understood  Use a hybrid technique: same advantages of Auger and accelerator detectors

3. Goal 1: Identify UHECR sources - Neutrinos generally point to sources - However, GZK neutrinos are not produced in the source or even in its radiation field but ~50 Mpc away - But it’s still true: [D. Saltzberg] ~2 Gpc  “GZK sphere” of arbitrary B deflection/diffusion  ~ (50 Mpc) / (2 Gpc) = 1.4°

4. Goal 2: Measure N@ ECM ~100 TeV [A. Connolly] 100 events: measure Lint = 400 km ± 33%

5. The Engel, Seckel, Stanev (ESS) GZK flux model zmax = 8, n = 3 We use  = 0.7  = 0

6. A simple hybrid optical/radio/acoustic detector Monte Carlo • 1016 - 1020 eV  2 down-going neutrinos • All flavor, all interaction (first bang only) • Optical: only muons for now (no light from showers) • Radio + acoustic: hadronic shower for all channels (LPM washes out EM component), Esh = 0.2E • Vertices uniformly in fiducial cylinder • AMANDA, RICE, and SAUND code

7. LHC An example hybrid array Optical: 80 IceCube + 13 IceCube-Plus (Halzen & Hooper astro-ph/0310152) holes at 1 km radius (2.5 km deep) Radio/Acoustic: 91 holes, 1 km spacing, 1.5 km deep shift real array to avoid clean air sector

8. Acoustic simulation • Based on SAUND tools • Differences from water: • - signal ~10x higher • - noise ~10x lower, limited by sensors (not ambient)? • different refraction (opposite and smaller) • shear waves? • - Unknown ice properties to be measured by SPATS • - For now we use a model for absorption length, extrapolated from lab measurements (P. B. Price astro-ph/0506648)

9. Sound velocity profile in South Pole ice Sound channel ridge measured in firn (J. Weihaupt) Firn (uncompactified snow) in top 200 m: Vsound increasing with density refraction. Rcurvature ~200 m! predicted in bulk (using IceCube-measured temperature profile and A. Gow temperature coefficient) - measure with SPATS?

10. Strong refraction in firn Acoustic: upward Radio: downward Signals always bend toward minimum propagation speed, but: Radio adores vacuum [c = 3e8 m/s] Sound abhors vacuum [c =0]

11. …signals from surface (noise) shielded by firn source @ 1 m depth: only downward ~10° penetrate source @ 10 m depth: only downward ~40° penetrate Signals from bulk ice (neutrinos) somewhat refracted… (emit a ray every 5°) source in bulk

12. Predicted depth (temperature)-dependent acoustic absorption at ~10 kHz P. B. Price model: absorption frequency-independent but temperature (depth)-dependent In simulation, integrate over absorption from source to receiver instrumented

13. Acoustic detection contours in ice Contours for Pthr = 9 mPa: raw discriminator, no filter

14. Coincident effective volumes + event ratesfor IceCube (I), an optical extension (O), and combinations with surrounding A + R arrays (GZK events/yr) astro-ph/0512604 Curves with I/O will improve when light from cascades included

15. Event reconstruction For physics we need E and/or (, ), perhaps from (x, y, z)cascade A, R can get good pointing from cascades (O gets ~30° in ice) Multiple constraints: {O, R, A} x {timing, radiation pattern, hit amplitudes, up/down going, polarization} How best to use and combine information? 1) timing most powerful (esp. for R, A) 2) radiation pattern (R cone, A pancake, O candies) also useful 3) hit amplitude most uncertain (except for O) Hybrid reconstruction? When possible with sub-arrays but improved with hybrid array When impossible with sub-arrays but possible with hybrid array  lower multiplicity threshold (maximize physics/$)

16. [Spiesberger + Fristrup] Mono or hybrid reconstruction from timing alone • NR+NA hits determine (NR -1) + (NA -1) hyperboloids - For unscattered signals, Ni hits in sub-array i constrain source to Ni -1 hyperboloids - Alternative: exploiting cacoustic << cradio, we get (NR - 1) hyperboloids and (NA)spheres, because t(emission) = t(first radio hit) compared to acoustic hit time • Also true for O+A, even with scattering: tO ~ tR ~ few s << tA ~ s)  Reconstruction possible with 1 fewer total hits • Linear analytical solution exists for most (NO,NR,NA) with at least 4 hits • Acoustic shear waves? Another velocity

17. Proof-of-principle Monte Carlo • Demonstrate we get a single solution with reasonable precision • Choose source and module locations randomly for each event (array and radiation pattern independence) • Time resolution: smear by ± 5 ns (R) and ±10 s (A) • No refraction (will worsen resolution) • No noise hits (will require higher multiplicity) • No receiver location error (will add absolute resolution floor)

18. 5 R + 0 A: 48.8 m 0 R + 5 A: 2.0 m (hyperboloids  planes) 0 R + 6 A: 0.3 m 6 R + 0 A: 7.2 m 1 R + 4 A: 1.7 m (spheres planes) all using fast analytical solution (~1000 evts/s): Cascade location reconstruction results 5 acoustic hits: 2.0 m 5 radio hits: 48.8 m

19. Instead of using timing only, we could use radiation pattern geometry only (no amplitudes) • Radio beamed in thin cone, acoustic in thin pancake • Bad for event rate, good for reconstruction • Acoustic: even with pancake thickness and refraction,very flat  fit a plane through the hit modules, upward normal points to the GZK source • Only requires 3 hits on 3 strings • What about E? Need vertex not just direction • But now a 2D problem: transform to the plane and intersect hyperbola within it (need 3-4 hits) • Similar for radio: 5 parameters determine a cone (known opening angle)  need 5 hits

20. Another demo MC: pointing resolutionusing acoustic radiation pattern only (no timing) determine hits, fit plane, compare neutrino direction actual radiation pattern no refraction no noise hits 0.5 km hole spacing isotropic 1019 eV ‘s overflow bin

21. Conclusions • Optical high energy neutrino detection proven by AMANDA with thousands of atmospheric neutrinos • GZK physics will require new techniques with large uncertainties • Bootstrap them using coincidence with IceCube and with each other • Join efforts with a large hybrid array with hybrid advantages • R/A: shallower narrower cheaper holes • ≥ 10 GZK events per year are possible • Hybrid reconstruction techniques are promising • South Pole possibly best place on Earth for all 3 techniques • Such a detector could discover UHECR sources and measure a cross section at 100 TeV ECM

22. Extra slides

23. O(91) radio/acoustic strings for a fraction of the IceCube cost? • Holes: ~3 times smaller in diameter (20 cm) and ~1.5 km deep • Don LeBar (Ice Coring and Drilling Services) drilling estimate: $33k per km hole length after $400k drill upgrade to make it weatherproof and portable (cf. SalSA ~$600k/hole) • Sensors: simpler than PMT’s • Cables and DAQ: Only ~5 radio channels per string (optical fiber). ~300 acoustic modules per string, but: • Cable channel reduction: Send acoustic signals to local in-ice DAQ module (eg 16 sensor modules per DAQ module) which builds triggers and sends to surface • Acoustic bandwidth and timing requirements are easy (csound ~10-5 clight!) • Acoustic data bandwidth per string = 0.1-1 Gbit, could fit on a single ethernet cable per string

24. Acoustic event rate depends on threshold (noise level) and hole spacing Trigger: ≥ 3 strings hit ESS GZK events per year: Need low-noise sensors (DESY) and low-noise ice (South Pole?) Frequency filtering may lower effective noise level For hybrid MC, set threshold at 9 mPa = a few sigma

25. Optical simulation • Check Halzen & Hooper’s rate estimate with standard simulation tools; run a common event set through optical, radio, and acoustic simulations • For now, only simulate the muon channel (cascades in progress) • Use standard AMANDA simulation tools: muon propagation, ice properties, detector response • Define a coincidence to be hits at 2 of 5 neighboring modules on one string within 1000 ns • Require 10 coincidences in the entire array within 2.5 s • For optical-only events, require > 182 channels hit (a muon energy cut proxy) to reject atmospheric background • Do not apply Nch requirement when seeking coincidence with radio or acoustic

26. Radio simulationUsing RICE Monte Carlo • Dipole antennas in pairs to resolve up-down ambiguity • 30% bandwidth, center frequency = 300 MHz in air • Effective height = length/ • Radio absorption model: based on measurements by Besson, Barwick, & Gorham (accepted by J. Glac.) • Trigger: require 3 pairs in coincidence • Use full radio MC

27. Resolution results: one sub-array alone, 6 hits acoustic radio

28. Resolution results: 1 radio + 4 acoustic hits intersect 4 spheres: without the radio hit we would not know the sphere radii, or would have too few hyperboloids