ULTRA-HIGH ENERGY NEUTRINO SEARCH WITH THE ASKARYAN RADIO ARRAY - - PowerPoint PPT Presentation

SLIDE 1

ULTRA-HIGH ENERGY NEUTRINO SEARCH WITH THE ASKARYAN RADIO ARRAY

PoS(ICRC2017)966

Ming-Yuan Lu for the ARA collaboration

Wisconsin IceCube Particle Astrophysics Center, Madison, WI, USA

35th International Cosmic Ray Conference, Busan, Korea July 12-20 2017

1 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 2

Neutrino-Cosmic Ray Connection

pγ → pπ 0 nπ +

• Neutrinos are produced when cosmic rays interact

with ambient matter/radiation field

2 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 3

Cosmogenic Neutrinos

3

• Ultra-high energy (UHE) cosmic

rays observed with energies up to 1020eV

• Photohadronic interaction

between UHE cosmic rays > 1019.5eV and CMB photons:

• This is the Greisen-Zatsepin-

Kuzmin (GZK) process – 
 a ‘guaranteed’ UHE neutrino flux

1018 1019 1020 Energy (eV)

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 4

Radio Detection

• Estimated cosmogenic neutrino event rate: ~< 1/km3/yr
• Requires ~100km2 of detector effective area

4

The Askaryan Effect

~20% charge asymmetry

Peak emission 0.1~1GHz P ~ Ne

2 ~ E2

Highly polarized broadband signal Confirmed detection in ice, SLAC 2006 (Phys. Rev. Lett. 99:171101)

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 5

ARA Detection Principle

5

Full ARA37 covers ~100km2
 Most cost-effective for determining cosmic neutrino flux > 1017eV Detection method New stations 2017-18

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 6

The ARA Detector

6

Vpol Hpol

Antenna requirements:


• 1. Broadband

150~850MHz

• 2. Azimuthal

symmetry

• 3. Fit in the hole

Deployed ~40m away from station center.
 Allow in-situ calibration 450MHz notch filter removes SP comms

Each station is an autonomous detector

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 7

ARA Data Analysis

• ~10TB data/yr/station
• Filter:

• Dominated by thermal triggers – a thermal rejection of 10-10 is needed to

bring thermal events to ten times less than expected signal

• Remove corrupted data (faulty electronics), anthropogenic sources
• Events that survive filters will be reconstructed

• Interaction vertex (direction, distance)

• Neutrino kinematic parameters (momentum four-vector)
• We will introduce 3 filter techniques and 1 reconstruction method

7

ARA02 Vpol calibration pulser

Simulated 1018eV on- cone event 1.2km away

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 8

Interferometric Reconstruction

• For reconstruction, set of delays

associated with each point in the sky is computed. Cross correlation values are computed per these delays and summed

• Delays are calculated with Radiospline

– B-Spline interpolation of tabulated raytraced arrival (Beheler-Amass et al., PoS(ICRC2017)1054)

• Radiospline accounts for geometric
• ptics in ice of changing index of

refraction

• Only Vpols are used in reconstruction

(for now). An n-channel filter of 3/8 channels with Vpeak > nσnoise is applied. Only these channels are used in reconstruction.

• n (threshold) determind by 1% thermal

rejection on 10% unblinded RF events

8

P

Σ(

r ) = 1 ZLT dt⋅ vi(t +τi( r ))v j(t +τ j( r ))

j =1 NA

∑

i=1 NA

∑

T

∫

Ice Surface staHon

200m 5000m

layer

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 9

Calibration Sources

• No physics background – artificial signals for calibration

• Local calibration pulsers

• Radio transmitters on IceCube string 1 & 22 (deep pulser)

• Rooftop pulser from IC Lab

• Mobile surface pulser

9 ] ° Zenith [

• 35
• 30
• 25
• 20
• 15
• 10
• 5

Coherence 0.05 0.1 0.15 0.2 0.25 ] ° Azimuth [ 220 225 230 235 240 245 250

ARA3 run8311 evt12472

S1 1450m Pulser S1 2450m Pulser S22 1450m Pulser

For calibration and ice properties inferred by 
 deep pulser data – Kelley et al. PoS(ICRC2017)1030 Local calibration pulser Deep pulser

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 10

Coherence [arb. unit] 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Cumulative ratio

• 4

10

• 3

10

• 2

10

• 1

10 1

Noise expo log(E/eV)=16 log(E/eV)=17 log(E/eV)=18 log(E/eV)=19 log(E/eV)=20 log(E/eV)=21

Simulation reconstruction

10

• Data set: 1018eV neutrino vertices randomly scattered around

an ARA station, up to 5km

] ° Reco Zenith - True Zenith [

• 30 -20 -10

10 20 30 Count 1 10

2

10

3

10

Mean: -0.13 RMS: 1.30

• Gaus. Fit

: 0.0049 µ : 0.35 σ

] ° Reco Azimuth - True Azimuth [

• 10
• 5

5 10 Count 10

2

10

3

10

Mean: -0.0059 RMS: 0.31

• Gaus. Fit

: -0.0044 µ : 0.29 σ

10-8 rejection:
 0.1082

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 11

Other Filters – Waveform RMS method

11

• Wavefrom RMS: wavefront RMS" is an estimator for how closely the arrival

time differences agree with each other

• Hit time is determined by power threshold crossing with 5ns integration
• Compare the hit times in similar pairs of antennas (e.g. vertically aligned

pairs) by finding the RMS of the delays in each pair

• Threshold and RMS cut are tunable parameters

5 10 15 20 25 30 0.2 0.4 0.6 0.8 1

Efficiency 3rd Highest Vpeak/RMS for all channels

= Face RMS Cut = Time Sequence Quality Parameter Cut

= A-type pairs = B-type pairs t1 t2 t3 t4

θA,i θA,ii

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 12

Other Filters – Surface filter

12

• Minimum arrival angle:


A2: ~35° A3: ~40°


• The signal arrival angle is

reconstructed with a plane wave fit using the timing from 4-hit combinations, then averaged over combinations

• Zenith resolution: 2.5°

zenith angle [degrees] 20 40 60 80 100 120 140 160 180 events [a.u.] 0.02 0.04 0.06 0.08 0.1

above-ice radio sources

• interactions

ν in-ice

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 13

Summary

• Multiple filters are being
• characterized. Can be

combined to maximize sensitivity

• UHE neutrino search will be

carried out on
 ARA2: 28.2 month
 ARA3: 27.2 month

• Radiospline reconstruction

readily accounts for ray- bending, and can reach ~<1° vertex resolution

• Going forward, optimization

study shows 25%-30% sensitivity increase in larger station geometry*

• Phased array approach will

allow triggering at lower energy thresholds**

• 6 ARA stations at the end of

this coming Pole season

13

[eV]

ν

E

16

10

17

10

18

10

19

10

20

10 ]

• 1

sr

• 1

s

• 2

) [GeV cm

ν

F(E

ν 2

E

• 10

10

• 9

10

• 8

10

• 7

10

• 6

10

• 5

10

• 4

10

• 3

10

• 2

10

Kotera et al. (2011) Ahlers & Halzen (2012) p @ 100EeV: 100% p @ 100EeV: 10% p @ 100EeV: 1%

IceCube (2015) Auger (2015) ANITA II (2010) ARA (2016) IceCube (2016) ARA 2 Stations '13-'15 (Trig. Lev.) ARA 200 Stations 5yr (Trig. Lev.)

*: PoS(ICRC2017)938 **:P. Allison et al, these proceedings

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 14

Backup slide

14 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 15

Askaryan Signal Modelling

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu 15

DetecHon in silica sand (Phys. Rev. LeW. (2001) 86.2802) DetecHon in rock salt (Phys. Rev. D72(2005) 032002 DetecHon in ice, SLAC (Phys. Rev. LeW (2007). 99:171101)

SLIDE 16

Askaryan Signal PolarizaHon

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu 16

SLIDE 17

LPM Effect & NC/CC InteracHons

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu 17

SLIDE 18

LPM Effect & NC/CC InteracHons

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu 18

SLIDE 19

Radio AWenuaHon In Ice

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu 19

SLIDE 20

Radio AWenuaHon In Ice

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu 20

• J. Appl. Phys. 80, 5884 (1996)

SLIDE 21

Delays - Raytracing

• Ice index of refraction varies with depth. Change is most

drastic near surface (firn). As a result, EM waves travel in curved paths – raytracing

• Ideal vertex direction/distance reconstruction need to take into

account raytracing effect

21

Varying Index of Refraction

index of refraction

• vs. depth

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 22

Raytracing and Radiospline

• Semi-analytic approach to compute ray paths -> Delays computed by this

approach tabulated and fitted with B-Spline

22 Antenna depth: 25m Antenna depth: 200m

Region with raytrace solutions

Horizontal Distance (m) 100 200 300 400 500 600 700 800 Depth (m)

• 800
• 700
• 600
• 500
• 400
• 300
• 200
• 100

Solution 0 Solution 1

Horizontal Distance (m) 500 1000 1500 2000 2500 3000 3500 4000 Depth (m)

• 900
• 800
• 700
• 600
• 500
• 400
• 300
• 200

Solution 0 Solution 1

Target Source Source Target Distance 4km Distance 1km

Region with raytrace solutions Region without raytrace solutions Region without raytrace solutions

Raytrace shadow boundary

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 23

Radiospline

23

• Solution: Cartesian coordinates to Cylindrical, multi-step process

Source location (zsource, radius) firn shadow tables

antenna

(ztarget, radius)

source in firn shadow?

No Solution

source in air? source in ice? air spline table ice spline table

Time Delays

random sources in air random sources in ice

sample size = 100,000 sources

• M. Beydler

Faster than raytracer by factor > 500

Random source/target locations (2.3 GHz Core i7)

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 24

Ray Tracing

• Can extrapolate curved paths by applying

Euler’s method to Snell’s law

• This is very simple, but horribly slow

∆s ∆s ∆s ∆s ∆s ∆θ

24 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 25

Ray Tracing

• The problem is a system of ODEs so one can use

standard ODE techniques, such as Runge-Kutta

• In particular, such methods can use an adaptive step

size to move very quickly where the ray varies slowly

(where τ is amplitude and A(z,f) is attenuation length as a function of depth and frequency)

25 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 26

26

Identifying Solutions

• In general, we have two endpoints, and we

want to know the properties of the ray(s) which connect them

• Solving the differential equations requires an

initial angle which is unknown

• In general, must use trial and error to find a

ray which travels from the source to the target, within some margin of error

• Allowing for reflections complicates matters

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 27

Illustrative Example

Distance (m) Depth (m)

27 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 28

Identifying Solutions as Roots

Initial Angle (radians) Vertical Miss Distance (m)

Direct Ray Surface Reflected Ray Bedrock Reflected Ray Maximum Above Ray Maximum Below Ray

28 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 29

Semi-Analytical Approach

• Casting lots of test rays is expensive!
• Sohda et al.: “Image formation in an optically

stratified medium: optics of mirage and looming” (1967)

• For the case of a pure exponential index of

refraction, the authors derive a functional form for the direct ray path

• Our index of refraction is a very similar form:

Sohda, et al.: South Pole Ice:

29 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 30

Schematic

Image formation by strat8ed media

505

• 3. Ray tracing in a stratified m

e d i u m

Analytically the problem of ray tracing in a stratified medium can be approached with

1 ~ ~ 0

dizerent methods. The medium may be considered to be constituted of many discrete layers: and the path of the ray through such a discontinuous medium with well-defined horizontal stratifications may be investigated, or the medium may be considered to vary continuously with height. The former approach has been used extensively for studying electromagnetic wave propagation in various regions of the earths stratified atmosphere (e.g. Wait 1962), but the latter approach has been preferred in the present analysis because

it seems to be more akin to reality.

Choose a Cartesian coordinate system with the origin on the surface of ground, the x axis being along the earth's surface (assumed to be plane) and the y axis pointing vertically upwards. Let the coordinates of an object S be (XO,JO) and that of an observer E be

(se.

je)

as shown in figures 1 and 2. Let P(xm,

ym)

be the point where the direction of the ray (making an angle ii with the vertical at the object) is horizontal; this point will also be

an extreme position (either the maximup. or the minimum) ir, the curved ray path a h g

ivhich this ligi,t beam travels.

V i P, 1

7 ' ' 0 L

p2

Figure 1. Path of the rays in mirage. Figure 2. Path of the rays in looming. Consider a layer of very small thickness dy and suppose that the light is incident at an angle i + d i

• n one surface and leaves it at an angle i from the other surface. Suppose

further that the refractive index of this layer is p and that of a similar layer immediately adjacent to it is p -

• dp. According to Snell's !aw

sin (i - di) -

p

• ~

sin i

p - dp

which, since di is small, leads to the equation

• cot i di =

dplp.

(2)

At this point we may mention that if instead of examining the changes in the angle of inci- dence, we had investigated the turning of the wave front as it propagates up or down the horizontal stratificatioos, we would have arrived at an exact!y identical equation. Using (1) we obtain Integrating and using the boundary conditions

• cot i

di =

CY dy.

. .

I =

11 or

' i ~

• il at y =

yo (figures 1 and 2 respectively)

R(x,z) dz z x S(x0,z0)

P(xm,zm)

I(xi,zi)

n + dn

I(xi,zi)

θ θ0 θ+dθ ∆θ

n

(Base figure from Sodha et al.)

30 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 31

31

Sketch of Derivation

• From Snell’s law:
• Substituting n(z) and solving for θ gives:
• Also, we have:
• Substituting and rearranging gives:
• UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 32

Condition for Direct Ray

• We arrive at:
• This cannot be solved analytically for σ0, but

it can be done numerically

• This is much faster than doing the same

thing (root finding), but using a ray-cast for each evaluation

14

where

32 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 33

AraSim

• Signal chain noise figure calibraHon results from T. Meures are

applied.

• EffecHve area results are consistent with previously obtained.
• This data set should be used for future analysis

33

[eV]

ν

E

16

10

17

10

18

10

19

10

20

10

21

10 ]

2

Aeff [m

• 2

10

• 1

10 1 10

2

10

3

10

4

10

5

10

ARA Single Station (Trigger Level)

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 34

Coherence reconstrucHon quality cut

• Different normalizaHon from previous slide

34

Coherence [arb. unit] 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Cumulative ratio

• 4

10

• 3

10

• 2

10

• 1

10 1

AraSim

AraSim expo log(E/eV)=16 log(E/eV)=17 log(E/eV)=18 log(E/eV)=19 log(E/eV)=20 log(E/eV)=21

ExponenHal fit range (0.082, 0.098) y = e55.4885 – 682.864x At noise pass rate 10-8, coherence is 0.1082 Passing rate

Noise 1e-8 1016eV 0.253 1017eV 0.399 1018eV 0.531 1019eV (3/5 size) 0.650 1020eV 0.738 1021eV 0.812

UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu

SLIDE 35

Nchnl>=3 Cut efficiency

Trig Nchan>=3 2013 A2 RF 10944954 113617 (1.04%) 1016eV 92.791 5.007 (5.40%) 1017eV 199.185 49.385 (24.79%) 1018eV 1150.953 548.996 (47.70%) 1019eV 3445.290 2093.14 (60.75%) 1020eV 5098.867 3552.08 (69.66%) 1021eV 5413.973 4053.74 (74.88%)

35 UHE Neutrino Search with ARA, ICRC2017, M.-Y. Lu