Simulation of the effective area for the Auger Engineering Radio Array. Simulation der effektiven Fläche für das Auger Engineering Radio Array

Transcription

1 Simulation of the effective area for the Auger Engineering Radio Array Simulation der effektiven Fläche für das Auger Engineering Radio Array Bachelor Thesis at the Karlsruhe Institute of Technology (KIT) by Andreas Wickberg October 2 Referent: Betreuer: Prof. J. Blümer Dr. T. Huege

2

3 Abstract Simulation of the effective area for the Auger Engineering Radio Array To detect cosmic rays in coincidence with particle and fluorescence detectors, the Auger Engineering Radio Array (AERA) is being deployed as part of the Pierre Auger Observatory in Argentina. AERA is a radio experiment specified for the measurement of electromagnetic waves emitted by air showers in the MHz range. The goal of the presented work is to predict the efficiency of the radio array by evaluating REAS3 simulations for different primary particles, energies and arrival directions. The effect of these air shower parameters on the effective detector area is investigated and an estimation of the total number of detectable events per year is given. The present setup with its 24 radio stations should be able to record up to four events per day for primary particles in the energy range from 7 ev to 9.5 ev. Between 9 ev and 9.5 ev the fully deployed array could detect about seven cosmic ray events per year. Zusammenfassung Simulation der effektiven Fläche für das Auger Engineering Radio Array Das Auger Engineering Radio Array (AERA) wird zur Messung von kosmischer Strahlung in Koinzidenz mit Teilchen- und Fluoreszenzdetektoren, als Teil des Pierre-Auger-Observatoriums in Argentinien, aufgebaut. AERA ist darauf ausgelegt, die durch die Luftschauer im MHz-Bereich ausgesandten elektromagnetischen Wellen zu messen. Ziel dieser Arbeit ist es, die effektive Fläche des Arrays auf Basis von REAS3-Simulationen für unterschiedliche Primärteilchen, Energien und Ankunftsrichtungen zu bestimmen. Die Auswirkungen dieser Parameter auf die effektive Fläche werden untersucht und eine Abschätzung der Ereignisrate pro Jahr wird gegeben. Es zeigt sich, dass der momentane Aufbau, bestehend aus 24 Radiostationen, in der Lage sein sollte bis zu vier Ereignisse pro Tag im Energiebereich von 7 ev bis 9.5 ev zu messen. Das voll ausgebaute Array könnte bis zu sieben Schauer pro Jahr zwischen 9 ev und 9.5 ev nachweisen. i

4

5 Contents Abstract i Introduction. Radio emission and experiments AERA Motivation and goals Effective area and event rate for AERA 7 2. Simulation database Processing with Offline General approach Defining the SNR The effective area Influence of the arrival direction Variation of the antenna spacing Event rate AERA phase Influence of the radio background Conclusion 2 Bibliography 23 Appendix 25 Danksagung 28 iii

6

7 CHAPTER Introduction Every second our planet is hit by cosmic rays with energies up to 2 ev. In the atmosphere they cause a shower of secondary particles, which themselves interact with air molecules. The flux of the cosmic particles is highly dependent upon their primary energy (see fig..) and can be described by a power law with index γ: dn de E γ, (.) where the spectral index γ varies from 2.5 to 3. depending on the energy. Especially for high energies the flux is very low and at 9 ev only one particle is expected per km 2 and year. Nevertheless these high-energy cosmic rays are of special interest, as the mechanisms that accelerate particles to such high energies are still unknown. Furthermore, the deflection of these high energetic particles in magnetic fields should be negligible and their arrival direction could give a hint to possible sources. At energies up to 4 ev the particles can be detected directly by balloon experiments or satellites. For higher energies the flux becomes less and huge detector arrays at the ground are necessary to measure the extensive air shower of secondary particles. Possible detection techniques are particle detectors for the charged electrons and muons, or fluorescence detectors recording the light emitted by excited nitrogen molecules. In addition to these traditional methods the radio detection of cosmic rays proves to be a promising approach for energies from 7 ev to over 9 ev.

8 2.. Radio emission and experiments Figure.: The cosmic ray energy spectrum as measured by different experiments in their respective energy range (Blümer et al., 29). The flux is scaled by E Radio emission and experiments As the cosmic rays interact with atmosphere-molecules they produce secondary particles in inelastic collision. In large part the secondary particles are electrons and positrons. While the secondary particles continue through the atmosphere, the air shower itself grows, reaches a maximum particle number and decreases again. The position of the shower maximum X max depends on the primary particle mass and its energy. During the interactions with the molecules further electrons are drawn into the shower and simultaneously positrons are annihilated. Together these two mechanisms cause a charge excess of -2% and hence a moving and, due to the variation of the total particle number, changing net charge. At the same time every charged particle is deflected and accelerated in the Earth s magnetic field according to the Lorentz force F L = q ( v B ). (.2) Its absolute value depends on the angle θ between the Earth s magnetic field and the shower axis and hence the acceleration and the amplitude of the electric

9 .. Radio emission and experiments 3 field depend on the arrival direction of the shower. Altogether the charge excess, the variation of the particle number and the deflection cause the emission of electromagnetic radiation (see Ludwig and Huege (2a), Werner and Scholten (28)). For experimental purpose the coherent part of this radiation is essential since the incoherent fraction is suppressed by a factor of about the maximum particle number. As the electrons and positrons travel with a velocity near the speed of light, the waves that are being emitted along the trajectory arrive at the observer in phase. That is true as long as the observation point is close to the shower axis. As soon as one moves away from the axis the phase difference causes interference and the signal becomes smaller. This effect becomes stronger when the air shower develops deeper in the atmosphere. Furthermore, the electromagnetic waves are emitted by many particles, all located in a pancake around the shower axis with a thickness of a few meters. Consequently, the phase difference for wavelengths in the meter range or frequencies of some MHz is insignificant and the extensive air showers can be detected by radio antennas in the MHz band. The first evidence for radio pulses from extensive air showers was found by Jelley et al. (965). Confronted with technical and theoretical difficulties this detection method was unattended for decades and it was not until the beginning of the 2st century that the interest in radio detection was renewed. This renaissance was mainly due to technical achievements like digital data processing and interferometry on the experimental side and new efforts for a quantitative understanding using simulations on the theoretical side. The goal is to establish radio detection as a standalone technique for cosmic rays at high energies and make use of the advantages this method provides. Radio measurements promise a duty cycle of nearly % and probably a new insight into the shower development and the emission mechanism, since the electric field at the observer position is a superposition of electromagnetic waves evolving from different stages of the shower. Two prototype experiments made the proof of principle and further important contributions to this new technique. The LOPES experiment (Falcke et al., 25) is situated at the Campus Nord of the Karlsruher Institut für Technologie and is co-located with the experiments KASCADE (Antoni et al., 23) and KASCADE- Grande (Apel et al., 2). LOPES is being triggered and provided with detailed shower information by KASCADE(-Grande). Measurements are performed in the range of 4 to 8 MHz and the direction reconstruction is based on radiointerferometry. LOPES showed the proportionality of the electric field strength LOPES: LOFAR PrototypE Station

10 4.. Radio emission and experiments UTM 9 South, (E = m, N = m) as offset 4 Central Radio Station Stations Stage Stations Stage2 Stations Stage3 SD grid N 2 Northing[m] Easting[m] Figure.2: The AERA layout is composed of three different stages (Fliescher (2), plot modified). Stage is already deployed and taking data. to the primary particle energy, the exponential behavior of the lateral distribution, a hint to the mass sensitivity of radio detection and more (for an overview, please see Huege et al. (2b)). The second prominent experiment is CODALEMA 2. Located at the Nançay radio observatory in France, it benefits from the low human-made radio noise level and proved the possibility of a direct reconstruction of the core-position from the antenna data. Moreover CODALEMA showed a lateral distribution consistent with the LOPES results, a north south asymmetry in the field strength and more (for an overview, please see Ravel (2)). LOPES and CODALEMA made the first step on the way to a new autonomous detection technique, but both the experiments are limited to measurements up to 8 ev and rely on other techniques as a trigger. Using the knowledge collected during the prototype experiments, a large scale experiment for radio detection is being constructed at the site of the Pierre Auger Observatory in Argentina. 2 CODALEMA: Cosmic ray Detection Array with Logarithmic ElectroMagnetic Antennas

11 .2. AERA 5.2 AERA Situated in the Argentinian Pampa, the Pierre Auger Observatory benefit from the large detection area and a relatively low human-made radiation underground in the visible as well as in the radio range. Still the transient radio signals coming from for example power lines complicate the measurements. The radio detection (RD) experiment AERA 3 (Fliescher, 2) is co-located with the surface (SD) and fluorescence detectors (FD) of the observatory, giving the opportunity to perform super-hybrid measurements of air showers by collecting complementary information both about the shower and the primary particle with SD, FD and RD. AERA, in its final stage, will be made up of 6 autonomous radio-detector stations covering an area of approximately 5 km 2. Currently 24 stations are assembled, each equipped with a logarithmic periodic dipole antenna (LPDA) measuring two perpendicular polarizations in a range from 3 to 8 MHz at the same time. The energy supply is provided by solar panels and a GPS antenna yields the timing information. The AERA layout is divided into three different antenna spacings (see fig..2). The already deployed dense core is consists of 24 antennas with a distance of 5 m. The second and third stage will be built up of 52 and 85 stations with a pitch of 25 m and 375 m respectively. The construction of a graded layout aims at a maximum number of detected events in the energy range from 7 ev to 9 ev..3 Motivation and goals Radio detection of air showers at large scale, as it is implemented at the Pierre Auger Observatory, has to overcome many challenges. As a major point to establish an autonomous technique, a shower has to be triggered and measured without aid of other experiments and there is still quite a way to go till the full potential of radio measurements is tapped. The goal of this work was to predict the effective area the radio detector array could reach and forecast the number of events per year. Special attention was put on the already deployed first stage of AERA with its 24 detector stations. Furthermore, the effect of a changed antenna spacing was investigated and compared with the actual plans for the future phases of the array. Finally, the effect of the background noise on all the results was tested by using two different noise libraries. Whereas former works covering this subject had REAS2 as their foundation, the following analysis was performed with air shower simulations based on REAS3. 3 AERA: Auger Engineering Radio Array

12

13 CHAPTER 2 Effective area and event rate for AERA The effective area is the detector size that is actually sensitive to detect extensive air showers. Of course this area depends on various parameters like the primary particle energy, its mass and arrival direction or the antenna spacing. In the following analysis these different parameters were investigated to understand their significance for the measurement. Furthermore, the effect of the antenna spacing was studied and compared with the actually planned AERA layout. Having determined the effective area for the whole array, the number of expected events per year was calculated. Finally the obtained results were specified for the already running first phase of AERA. 2. Simulation database To get a good estimation of the total effective area a precise simulation of the radio signal is necessary. In this work REAS3 (Ludwig and Huege, 2b) was used to simulate showers at different observer positions, with different arrival directions and energies. The air shower development, the trajectories of each electron and positron and their creation and annihilation is calculated with CORSIKA and the particle distribution is passed to REAS3 by means of histograms. REAS3 calculates the emission from each particle and sums it up at defined observer positions, which are co-located with the antenna positions in the following analysis. Caused by this approach every shower simulation is tied to a specific point in the array and can not be relocated without a detailed rerun of the code. Basically a simulation for the radio emission from air showers can be mi- 7

14 Processing with Offline croscopic, like REAS3, or macroscopic like for example MGMR (Werner and Scholten, 28). One of the main advantages of REAS3, compared to macroscopic models, is the basic approach and the abdication of parametrizations. Especially close to the axis the results gained from REAS3 and MGMR differ, whereas both models converge for larger distances (Huege et al., 2a). The mentioned differences are founded in an over simplified shower model used for MGMR (Ludwig and Huege, 2). Since for the first phase of AERA the signals at a distance of about m are of utmost importance, REAS3 formed a solid basis for the following calculations. The underlying database consists of 475 REAS3 events for proton and 225 iron events at energies between 7 ev and 9.5 ev, thus giving the opportunity to study the influence of the primary particle type and energy on the radio signal. The energy distribution follows E. The arrival directions follow a uniform angle distribution from to 36 degree for the azimuth angle and a distribution proportional to cos 2 (θ) for the zenith angle between 5 and 65 degree. The shower cores are randomly spread over the AERA array, having and 25 events for each proton and iron in the area of the first and second phase respectively and 25 proton events in the area of the third phase (see fig..2). 2.2 Processing with Offline At the Pierre Auger Observatory the software framework Offline is used to analyze the recorded data and combine the results of all detector components and extensions into one interface (Argirò et al. (27), Abreu et al. (2)). Besides the evaluation of experimental measurements, it offers the possibility to mimic the experimental data pipeline for simulated events. The processing with Offline is subdivided into several little programs, the socalled modules. Each module is specified for certain steps in the sequence from raw signal to the reconstructed event. Starting with a simulated event the same modules as for recorded antenna signals but in opposite direction and order are applied. The result corresponds to a noiseless experimental cosmic ray signal as it is supposed to be measured by the antennas with all detector effects, like for example the different responses of the cables, filters and amplifiers depending on the signal frequency, incorporated (for further information please see Abreu et al. (2)). Contrary to these pure signals, the real signals at the detector site are overlayed by various natural and human made radio sources, forming a radio background that complicates the measurement and makes it harder to determine the cosmic ray component. To simulate these complications and to get as close as possible to the experimental circumstances, a noise library with over 6 noise events from

15 2.3. General approach 9 the AERA detection site was put together and is randomly added to the noiseless signal, creating something comparable to a real experimental trace. From this on there is no more difference between the module sequence used for experimental and simulated data. The complete module sequence with a description of every module can be found in the appendix. 2.3 General approach The most accurate method to calculate the effective area would be to process tens of thousands of simulations with varying parameters, spread over the whole detector array and count the fraction of reconstructed events. Unfortunately this method shows to be unrealistic due to limited computation time and resources. The indirect approach used in this work, is based on the idea to calculate the lateral sensitivity in detecting a clear signal for a standalone station and use this to estimate the probability for threefold coincidences between radio stations as presented in Erdmann et al. (29). The idea was refined by considering not only the influence of the primary particle type and energy but also its zenith angle. By means of the sensitivity, specified for different shower particle types, energies and angles, Monte Carlo generated shower core positions and arrival directions were evaluated to determine the effective area for AERA Defining the SNR To calculate the lateral sensitivity it was necessary to re-examine the signal to noise ratio (SNR) as used in the conventional event reconstruction. The goal is to get an idea of the individual antenna sensitivity in seeing a definite signal, whilst being sure that it is not just noise triggering the station. The preset definition of the signal to noise ratio S used in Offline is the following: N S N = A2 σ 2 noise, (2.) where A is the maximum of the enveloped total electric field strength and σ noise denotes the noise RMS. The SNR requirement influences the reconstruction by excluding signals that are barely higher than the radio background. To get a feeling of this effect and to find an appropriate value, minimum bias data taken at the AERA detection site was used. These events were recorded using a ten second trigger without any indication for a cosmic ray. It can be assumed that they should contain very few or even no reconstructible showers at all. The fraction of misleadingly

16 2.4. The effective area Probability stations which pass SNR-cut reconstructed events SNR Fraction of stations which pass the SNR-cut.2 sim with noise SNR Figure 2.: With the minimum bias data the fraction of reconstructed events and stations that pass the SNR-cut was calculated (lhs). As a check the fraction of activated stations was determined for the simulation database (rhs). reconstructed events for Offline set to a certain SNR and the fraction of stations which pass the SNR-cut are shown in figure 2.. Both indicators suggest that a SNR of about 9 should minimize the percentage of wrongly detected showers to under % and guarantee a signal that can be well distinguished from the radio background in every station that passes the cut. To be sure not to exclude a significant number of real signals by this choice, the fraction of activated stations, that means stations passing the SNR-cut, was also calculated for the simulated events (see fig. 2.). For a small SNR the number of activated stations is huge because the noise induces a trigger in nearly every station even if no signal can be seen. For a higher SNR the fraction first decreases rapidly to about.2. Afterwards it shows just a small change and decreases by 8% from 8 to. A signal to noise ratio of 9 was fixed for the further analysis. This seems to be a reasonable choice, restraining the noise induced triggering on the one hand and granting a high number of real signals on the other hand. 2.4 The effective area The total number of simulated signals, for a certain distance between shower axis and radio station, was compared to the number of signals passing the SNR-cut as defined in By this the lateral sensitivity for different energy intervals between 7 ev ev was derived. Up to this step the approach is identical to the method presented in Erdmann et al. (29). The improvement made in this work was to introduce a further differentiation to take into account not only the energy but also the arrival direction of the air shower. In spherical coordinates the direction can be described by the zenith angle θ and the azimuth angle φ.

17 2.4. The effective area Energy log(e/ev) Energy log(e/ev) Distance to station Distance to station Energy log(e/ev) Distance to station Figure 2.2: The single-station detection probability for proton induced air shower at energies of 7 ev ev and different zenith angle intervals. The first plot depicts the results for θ [ ; 22 ). The second and third plot represent θ [22 ; 44 ) and θ [44 ; 66 ) respectively. The best approach to implement the arrival direction into the analysis would be to calculate single-station detection probabilities for specified particle type, energy, zenith angle and azimuth angle. Due to the limited database of air shower simulations the differentiation in this work was limited to the zenith angle θ that showed a stronger influence on the detection probability than the azimuth angle did. On the one hand the zenith angle influences the position of the shower maximum due to geometrical effects and a changed atmospheric depth, the shower has to pass, before arriving at the radio station. With increasing zenith angle θ the position of the shower maximum and with it the whole shower is displaced to a bigger geometric distance from the Earth s surface. On the other hand the angle between shower axis and geomagnetic field vector determines the electric field strength of the measured signal. In first order this dependency can be explained the Lorentz force as described in section.. With θ as a third parameter a single-station detection probability was calculated dependent on the primary particle type, its energy, the zenith angle of the air shower and the distance between shower axis and radio station (see fig. 2.2).

18 The effective area The histograms indicate that, at least in most cases, the lateral sensitivity grows with energy and zenith angle of the primary particle. It is mentioned in Erdmann et al. (29) that there is a systematical underestimation of the effective area caused by averaging over showers with varying arrival directions before determining the detection probability for different inclinations. The refinement made in this work was to calculate a single-station detection probability specified for a certain particle type, energy and zenith angle before the effective area, averaged over varying arrival directions, was calculated. The two different approaches will be referred to as the averaged method as presented in Erdmann et al. (29) and the differentiated method which is the basis for the results presented below. The use of histograms and binning introduces errors that are caused by the finite interval width. Showers falling into the same bin are averaged over and especially in case of the energy binning at low energies the miscalculations can be important as showers at 7 ev are rarely even reconstructed, whereas the probability is much bigger for primary energies of 7.5 ev. The number of particles is determined by the cosmic ray flux that can be described with a spectral index γ = 3 in the considered energy range. As the events in the database already follow a distribution according to E, each event is weighted with an additional factor of E 2 to account for the cosmic ray flux and to avoid an overrated influence of the events at the upper bound of every single energy bin. In the next step a Monte Carlo code, providing randomly generated shower directions and core positions, was used to create millions of cosmic rays within all three stages of the detector array. For each individual of these shower a signal detection probability is appointed to the surrounding antennas, by determining the distance to the generated shower axis and the lateral sensitivity for the corresponding energy and zenith angle. In order to reconstruct the arrival direction of a shower, at least three antennas with a clear signal are necessary. According to that premise the likeliness for a shower to be detected was calculated by multiplying the lateral sensitivity of the three closest antennas. Finally, to calculate the effective area for AERA the geometrical surface, subdivided into bins with an area A i, is weighted by the detection efficiency w i of every single bin: A eff = i w i A i. (2.2) Figure 2.3 shows the detection probability for proton and iron induced shower in different energy intervals.

19 2.4. The effective area 3 Energy bin log(e\ev): iron Energy bin log(e\ev): Energy bin log(e\ev): Energy bin log(e\ev): Energy bin log(e\ev): Figure 2.3: Efficiency map showing the probability to detect a shower with proton (lhs) or iron (rhs) as primary particle with different energies. The coordinates on the x- and y-axis are given in meters. The efficiency is drawn on the z-axis.

20 The effective area Effective area [ km 2 ] proton, [ ; 65 ) iron, [ ; 65 ) geometric area Effective area [ km 2 ] differentiated method, [ ; 65 ) averaged method, [ ; 65 ) geometric area Energy bin log(e\ev) Energy bin log(e\ev) Figure 2.4: The left plot shows the effective area for proton and iron induced showers calculated with the more differentiated method taking into account the effect of the zenith angle. On the right a comparison between the averaged and the differentiated approach for proton is shown. The total number of particles in a shower depends on the primary particle energy and so does the electric field strength, being the sum of all emitted electromagnetic waves at the observer position. With increasing shower energy the electric field amplitude and hence the effective area grows, because the signal can be seen over larger distances. At primary energies of about 7.25 ev the detector efficiency is quite small but the sheer amount of cosmic rays in this range is sufficient to compensate for that. The first phase of AERA gains full efficiency, meaning an effective area as big as the actual geometrical detector size, at about 8.25 ev. Going to higher energies the whole array reaches its full potential at about 9.25 ev. With a reliable probability, iron induced showers at these energies can be detected even outside the detector array, whereas the reconstruction of the shower core position is supposed to be very difficult in this case. The effective area as calculated according to equation 2.2 is depicted in figure 2.4. For proton induced showers it is systematically smaller than the effective area for iron induced air showers. The only exception is the second energy bin, where the slightly bigger proton area can be explained by the limited statistics in the energy intervals from 7 ev to 8 ev. Caused by the weak radio signal only few showers are reconstructed at all, decreasing the statistical foundation of the calculations in these bins. On the other hand the shower detection works very well for higher energies, creating a solid statistical basis and reliable results. The difference in the effective areas at these energies is caused by the differing shower development of the primary particles. Iron, having a larger cross section than proton, causes air showers that evolve earlier in the atmosphere and have their shower maximum at

21 2.4. The effective area 5 Sensitivity [ ; 22 ) [22 ; 44 ) [44 ; 66 ) Effective area [ km 2 ] p, [ ; 22) p, [22; 44) p, [44; 66) Fe, [ ; 22) Fe, [22; 44) Fe, [44; 66) geometric area Distance [m] Energy bin log(e\ev) Figure 2.5: The detection probability is once more displayed for proton induced showers in the energy range from 8.5 ev to 9 ev. The sensitivity at large distances grows with the zenith angle θ (lhs). The effective area for different zenith angle intervals is shown on the right. a smaller atmospheric depth. For geometrical reasons the coherence at some distance from the shower axis decreases more slowly if the waves are emitted further away from the surface (cf. sec..). Therefore the lateral development of the electric field is, in general, more steep for proton than for iron and the signal of an iron induced shower can be detected over larger distances. As mentioned above the area calculated by the method presented in this work is supposed to be bigger than the area as calculated with the approach presented in Erdmann et al. (29). A comparison shows that the effective area determined with the improved method is systematically bigger and the difference is about 6% in the lowest and 9% in the highest energy bin (see fig. 2.4) Influence of the arrival direction To investigate the effect of the zenith angle in detail, the effective area was calculated once more, restricting θ in the Monte Carlo code to intervals as they were already used to calculate the lateral sensitivity. The results show an effective area for a specified inclination (see fig. 2.5). Especially at higher energies the tendency towards a bigger effective area for more inclined showers is apparent. For these showers, developing deep in the atmosphere, a variation of the shower maximum caused by an inclination of the axis, has a grave effect on the lateral shower coherence and therefore on the electric field amplitude. The effective area for highly inclined showers in the energy bin from 9 ev to 9.5 ev is up to 25% bigger than for vertical showers with θ =.

22 The effective area Effective area [ km 2 ] spacing 5m spacing 75m spacing m spacing 25m spacing 5m spacing 75m spacing 2m spacing 3m spacing 4m spacing 5m spacing 6m Energy bin log(e\ev) Figure 2.6: The effective area was calculated in different energy bins with varied antenna spacing Variation of the antenna spacing Besides the shower parameters, the detector layout and especially the antenna pitch has a crucial influence on the detection efficiency in the different energy regions. For low energies a narrow grid is supposed to have the best effect, whereas a compromise has to be found for higher energies, where a big detection area is necessary to compensate for the small cosmic ray flux. To study the influence of the station spacing on the effective area, a square grid with a constant number of 44 radio stations was used. The distance between the individual stations was varied from 5 to 6 m causing an increasing geometrical detector area. Although the square grid is not identical with the hexagonal layout used for AERA, it is sufficient for a first analysis of the effects. In the energy bin from 7 ev to 7.5 ev the effective area grows with the spacing and the geometrical dimension of the array as long as the distance is small enough to see a signal in at least three radio stations (see fig. 2.6). At about 25 5 m it reaches a maximum and decreases as there are not enough stations close enough to get an event reconstruction for all showers. For medium energies a distance of 3 seems to be appropriate, whereas at the highest energies a pitch of about 4 5 m promises to maximize the effective area. Compared to these results the actual plans (cf. sec..2) based on 5 m, 25 m and 375 m for the different stages of the experiment, seem to cover the desired energy range quite well.

23 2.5. Event rate 7 ] km 2 sr yr E 3 J(E) [(ev) 2 38 KASCADE-Grande data Auger SD infill data 2 Auger SD data 2 yr Events p : event rate p : integrated rate Fe: event rate Fe: integrated rate log(e\ [ev]) Energy bin log(e\ev) Figure 2.7: The energy spectrum of cosmic rays as used for this analysis (lhs). Dropped data points are displayed in grey. The estimated event rate for pure proton or iron showers (rhs), showing the integrated event rate of all events over a certain energy. 2.5 Event rate Results gained from an experiment are just as good as their statistical foundation. Especially at high energies the number of detected cosmic rays is small and the statistical error huge. To get an estimation of the event rate in case of AERA the number of detectable events was calculated using the results obtained above. The energy flux J is the number of events per solid angle, area, time and energy: dn dω da dt de = J(E). (2.3) To get the total number of events the equation has to be solved and integrated: N = 2π T J(E) A(E, θ) cos(θ) de sin(θ)dθ. (2.4) Where T is the observation time and A(E, θ) cos(θ) the projection of the effective area depending on E and θ. As the effective area was only calculated for certain angle intervals θ j and energy intervals E i, it is approximated as a constant for the appropriate choice of E and θ. N 2π T i,j [ J(E i ) A(E i, θ j ) (θ)] E i 2 sin2 E i (2.5) θ j The flux J(E) covers an energy range that demands data from different cosmic ray experiments. In this case measurements from KASCADE-Grande (Bertaina

24 Event rate proton, [; 65) p : event rate Effective area [ km 2 ] iron, [; 65) geometric area yr Events p : integrated rate Fe: event rate Fe: integrated rate Energy bin log(e\ev) Energy bin log(e\ev) Figure 2.8: On the left hand side the effective area for the first phase of AERA is calculated. The dashed line marks the geometrical area covered by the radio stations. The event rate for iron and proton is shown on the right hand side. et al., 2) and the Pierre Auger Observatory (The Pierre Auger Collaboration et al., 2), with extra data points provided by the SD infill extension (Schulz, 2), were used to create an all particle energy spectrum in the desired range (see fig. 2.7). The experiments show large uncertainties at the end of their individual energy range. These points were dropped and more stable data from other experiments was used. Evaluating equation 2.5, the event rate for specified energy bins was calculated, taking the effective area for proton or iron as a basis. The integral over the energy was calculated by means of a Riemann sum using the flux data. The actual particle type composition of cosmic rays at these energies is still not finally identified, so the expected number of events per year, as depicted in figure 2.7, outlines the limits for pure proton or iron induced air showers. According to the estimation a huge amount of events at lower energies is expected, giving the possibility to perform multiple measurements in combination with SD and FD. Still one should keep in mind that especially at these energies the fluctuations due to the small detection probability are significant and influence the event rate accordingly (cf. sec. 2.4). At the other end of the spectrum, towards high energies, there is still a considerable number of events that will allow the study of cosmic rays at high energies using radio detection. These results show that the detector layout, with its graded spacing, facilitates the detection of events in the aspired energy range. Compared to the calculations in Erdmann et al. (29) the event rates differ by 7% when taking iron as primary particle and by over 5% in case of proton induced showers.

25 2.6. AERA phase 9 75 μv/m/mhz Frequency [MHz] yr Events noise library March 2 noise library November November 2 (8h) Time March 2 (5h) Energy bin log(e\ev) Figure 2.9: A comparison of the noise libraries from November 2 and March 2 shows a systematically higher noise level in the 2 sample (lhs). The electric field strength in is shown on the z-axis. This changed radio background has a huge effect on the calculated event rate especially in the lowest energy bin. Based on proton as primary particle the event rate of the fully deployed array is depicted for both samples (rhs). 2.6 AERA phase All considerations made above have the full deployed radio array as their foundation, although only the first phase is taking data at the moment. In this section the calculations for the present state of AERA, with its 24 antennas, will be performed to illuminate the possibilities the experiment offers already. The effective area reaches the geometrical detector area in the second energy bin, making the array very efficient in this range (see fig. 2.8). Towards high energies the effective area grows beyond the area covered by the antennas and, despite the low flux, there should be still about one event per year between 9 ev and 9.5 ev (see fig. 2.8). Altogether the estimation yields about 2 to 4 events per day in the whole energy range, providing a good basis for the study of cosmic rays and the radio emission mechanism. 2.7 Influence of the radio background The detection of cosmic rays by means of the radio method would be easy if the antennas would only record the actual shower signals. Unfortunately these signals are overlayed by a radio background that is produced by a great number of natural and human made sources. Although the detection site in the Argentinian Pampa has a quite low continuous radio background there are several sources of transient signals in the MHz band like for example power lines, complicating the

26 Influence of the radio background measurements and in particular the triggering enormously. In this work noise events recorded by AERA were added to the REAS3 simulated signals to mimic a real air shower. During the analysis two different noise libraries were used. The first one was recorded in November 2, the second one in March 2. For some, at the moment unknown, reasons the second library shows a significantly higher mean noise level than the first one (see fig. 2.9). All results presented above are based on this second noise sample. To investigate the effects of a changed radio background the complete analysis was repeated with the noise library recorded in November 2. A lower radio background leads to more events especially at lower energies where the field amplitude of the air shower is barely higher than the noise level. In the lowest energy bin the event rate is over two times larger than in the calculations based on the 2 noise library, whereas the differences towards higher energies are very small (see fig. 2.9). The increased radio background at the Pierre Auger Observatory in March 2 leads to a diminished event rate especially caused by the loss of many low energy events. If this variation is due to the surrounding human made radio sources and not just an effect of changed detector properties or natural sources like for example the passage of the seasons, an effort to go back to a noise level as it was in November 2 is desirable.

27 CHAPTER 3 Conclusion In this work the effective area for the Auger Engineering Radio Array was calculated and an estimation of the event rate was given. For the first time the whole analysis was based on REAS3 simulations. As part of the Pierre Auger Observatory AERA is taking data since October 2 and the first hybrid measurements in coincidence with the surface detectors have already been detected. According to the results derived in section 2.4.2, the chosen antenna spacings will be appropriate to cover the energy range from 7 ev to 9.5 ev. Altogether over 67 events per year are expected for the fully deployed array and even at high energies between 9 ev and 9.5 ev about 7 events per year should be seen. For the first stage of the experiment an event rate of 2-4 events per day was forecasted. In general a systematically bigger effective area for iron induced air showers compared to proton induced showers was found. Furthermore it was shown that the effective area grows with the zenith angle of the shower axis. This work has once more shown that AERA bears the possibilities to be a milestone for the radio method in the MHz range, but only the next years can show whether radio detection can be established as a standalone detection technique and supplement our knowledge both about the air shower radio emission mechanism and the cosmic rays themselves. 2

28

29 Bibliography Abreu, P. et al. (AUGER) (2). Advanced functionality for radio analysis in the Offline software framework of the Pierre Auger Observatory. arxiv: Antoni, T. et al. (KASCADE) (23). The Cosmic ray experiment KASCADE. Nucl. Instr. Meth. B, A53: doi:.6/s68-92(3)276-x. Apel, W. D. et al. (KASCADE Collaboration) (2). The KASCADE-Grande experiment. Nucl. Instr. Meth. A, 62: Argirò, S. et al. (27). The Offline Software Framework of the Pierre Auger Observatory. Nucl. Instr. Meth., A58: doi:.6/j.nima arxiv: Bertaina, M. et al. (KASCADE Collaboration) (2). The cosmic ray energy spectrum in the range 6-8 ev measured by KASCADE-Grande. Astrophys. Space Sci. Trans., 7: 229. Blümer, J., Engel, R. and Hoerandel, J. R. (29). Cosmic Rays from the Knee to the Highest Energies. Prog. Part. Nucl. Phys., 63: doi:.6/j. ppnp arxiv: Erdmann, M. et al. (29). Simulation of the Efficiency to Detect Air Showers with the Auger Engineering Radio Array AERA. AERA GAP note Falcke, H. et al. (LOPES Collaboration) (25). Detection and imaging of atmospheric radio flashes from cosmic ray air showers. Nature, 435: doi:.38/nature

30 24 Bibliography Fliescher, S. (2). Radio detection of cosmic ray induced air showers at the Pierre Auger Observatory. Nucl. Instr. Meth. A, In Press, Corrected Proof:. doi:.6/j.nima Huege, T., Ludwig, M., Scholten, O. and de Vries, K. (2a). The convergence of EAS radio emission models and a detailed comparison of REAS3 and MGMR simulations. Nucl. Instr. Meth. A, In Press, Corrected Proof:. doi:.6/j. nima Huege, T. et al. (LOPES Collaboration) (2b). The LOPES experiment - recent results, status and perspectives. doi:.6/j.nima Proceedings of the ARENA 2 conference, Nantes, France. Jelley, J. V. et al. (965). Radio Pulses from Extensive Cosmic-Ray Air Showers. Nature, 25: 327. Ludwig, M. and Huege, T. (2a). REAS3: A revised implementation of the geosynchrotron model for radio emission from air showers. Nucl. Instr. Meth. A. doi:.6/j.nima Ludwig, M. and Huege, T. (2b). REAS3: Monte Carlo simulations of radio emission from cosmic ray air showers using an end-point formalism. Astropart. Phys., 34: doi:.6/j.astropartphys Ludwig, M. and Huege, T. (2). Analysis of air shower radio signals with REAS3. In Proceedings of the 32th ICRC, Beijing, China. Ravel, O. (2). The CODALEMA experiment. Nucl. Instr. Meth. A, In Press, Corrected Proof:. doi:.6/j.nima Schulz, A. (2). Private communication. The Pierre Auger Collaboration et al. (2). The Pierre Auger Observatory I: The Cosmic Ray Energy Spectrum and Related Measurements. arxiv: Werner, K. and Scholten, O. (28). Macroscopic Treatment of Radio Emission from Cosmic Ray Air Showers based on Shower Simulations. Astropart. Phys., 29: doi:.6/j.astropartphys

31 Appendix Offline The used SVN revision number is Using the configuration file bootstrap.xml two options differing from the default values were set: <configlink id="rdstationsignalreconstructor"> <RdStationSignalReconstructor> <MinSignalToNoise> 9 </MinSignalToNoise> </RdStationSignalReconstructor> </configlink> <configlink id="rdchanneltimeseriesclipper"> <RdChannelTimeSeriesClipper> <ClippedTimeSeriesLength> 248 </ClippedTimeSeriesLength> </RdChannelTimeSeriesClipper> </configlink> The module sequence as it was used for the analysis is the following: Module sequence <!-- A sequence for processing simulated radio events --> <sequencefile xsi:nonamespaceschemalocation="/home/wickberg/offline/ install/share/auger-offline/config/modulesequence.xsd"> <enabletiming/> <modulecontrol> 25

32 <loop numtimes="unbounded"> <module> EventFileReaderOG <!-- read in a simulated file --> </module> <module> RdStationAssociator </module> <!-- associate simulated pulses to antenna stations --> <module> RdAntennaStationToChannelConverter </module> <!-- calculate antenna response per channel on simulations --> <module> RdChannelResponseIncorporator </module> <!-- apply the forward channel response to the data --> <module> RdChannelResampler </module> <!-- resample channel data to experimental sampling rate --> <module> RdChannelTimeSeriesClipper </module> <!-- clip time series to number of samples in experiment --> <module> RdChannelVoltageToADCConverter </module> <!-- convert channel voltages to ADC counts --> <module> RdChannelNoiseASCIIImporter </module> <!-- Noise from a library consisting of real noise samples is added to the trace --> <!-- At this point of the module sequence, the simulated signal has all detector effects incorporated. Consequently, the following module sequence is identical to the one for measured data as provided in the RReconstruction example. --> <module> RdChannelADCToVoltageConverter <!-- convert ADC counts to voltage --> ><module> RdChannelPedestalRemover <!-- remove a possible DC offset --> </module> </module> <module> RdChannelResponseIncorporator </module> <!-- remove the channel response from the data --> 26

33 <module> RdChannelRFISuppressor </module> <!-- suppresses narrow - band signals using a median filter --> <module> RdChannelUpsampler <!-- up - sample the data --> </module> <loop numtimes="unbounded"> <module> RdAntennaChannelToStationConverter </module> <!-- use antenna pattern to reconstruct e - field vector --> <module> RdStationSignalReconstructor </module> <!-- reconstruct pulses on station level data --> <module> RdDirectionConvergenceChecker </module> <!-- checks whether the direction reconstruction via RdPlaneFit has converged, breaks the loop either after convergence or iterations. --> <module> RdPlaneFit </module> <!-- fit the arrival direction of the pulses --> </loop> <!-- Deselected for efficiency study <module> RdEventPostSelector </module> --> <!-- select only events with successful direction reconst. --> <module> RecDataWriterNG <!-- write out ADST file --> </module> </loop> </modulecontrol> </sequencefile> 27

34 Danksagung Ich möchte all den Menschen danken, die mich während der Arbeit an dieser Thesis unterstützt und begleitet haben. Ich danke Tim Huege und Maximilien Melissas für die Betreuung, die Diskussionen und die Hilfe bei der Überwindung des ein oder anderen Hindernisses. Ich danke auch Marianne Ludwig und Nunzia Palmieri sowohl für die angenehme Zeit im Büro, als auch für Rat und Tat. Weiterhin möchte ich mich bei der gesamten Radio-Gruppe für die freundliche Aufnahme und die unzähligen Ratschläge bedanken. Es war eine schöne Zeit bei euch! Mein Dank gilt auch Andreas Haungs für viele wichtige Hinweise und den letzten Feinschliff meiner Arbeit. Schließlich danke ich meiner Familie, meiner WG und meinen Freunden, die jederzeit für mich da waren. 28

35 Hiermit versichere ich, dass ich die vorliegende Arbeit selbstständig verfasst und keine anderen, als die angegebenen Quellen und Hilfsmittel benutzt habe. Andreas Wickberg Karlsruhe, den. Oktober 2