Mohamed Ismaiel

AugerPrime will play a key role in understanding the origin of Ultra-High Energy Cosmic Rays by precisely measuring the mass composition. Neutral particles will be decisive in constraining models. The identification of photon primaries depends on the ability to distinguish their air showers from a background dominated by protons and heavier nuclei primaries. More statistics are needed, as well as an accurate measurement. This thesis establishes a methodology for photon identification using the Radio Detector array, spanning commissioning, reconstruction, and quality selection, and culminating in a transparent cut-based photon–hadron separation strategy. The work quantifies detection capabilities and develops analysis tools that substantially improve event quality and enable high-purity photon-candidate selections.

Chapter 3 establishes the Radio Detector’s operational status and data quality through the development and deployment of the Monitoring Radio Detector (MoRD) system, which was essential for validating the array as a reliable instrument. Deployment progressed from the 8-station Engineering Array in 2021 through the 40-station Super Hexagon in 2023 to full array completion in November 2024. The MoRD platform, developed as part of this thesis, processes data every 12 hours, generating interactive visualizations that enable real-time identification of hardware failures, antenna misalignments, and radio frequency interference. Critical commissioning issues were identified and resolved using MoRD diagnostics: swapped antenna arms were detected via anomalous polarization ratios in the 67-MHz TV transmitter line; malfunctioning low-noise amplifiers were identified through systematically reduced gain in specific polarization channels; and antenna rotation under wind loading was quantified through time-dependent alignment monitoring.

Chapter 4 investigates the Radio Detector’s potential for photon detection through comprehensive simulation studies. Using photon-induced air shower simulations spanning 1018.4 to 1019.4 eV and zenith angles from 65◦ to 85◦, the chapter establishes that radio detection can recover many photon-induced air showers missed by Water Cherenkov Detectors at extreme zenith angles. Three reconstruction modes were evaluated: WCD-only, RD-only, and hybrid operation. The hybrid configuration achieves reconstruction efficiencies exceeding 95% above 1019 eV for zenith angles greater than 70◦, where WCD efficiency drops below 20%. The effective aperture for photons could reach 1500–2000 km sr in the hybrid mode, approaching the geometric limit of 2400 km sr and representing more than a factor of three improvement over WCD-only operation. These results demonstrate that a radio self-trigger could extend photon sensitivity to energies as low as 1018.5 eV at large zenith angles, addressing the observational

gap identified in Chapter 1.

Chapter 5 establishes the reconstruction performance of the Radio Detector and validates its reliability through systematic comparison with Water Cherenkov measurements. The Water Cherenkov reconstruction relies on fitting the muon lateral distribution and spherical air-shower front, achieving angular resolution better than 0.5◦ and energy resolution of approximately 19%. Radio reconstruction employs a two-stage process: timing fits using WCD-seeded geometry followed by lateral distribution fitting to extract electromagnetic energy with better than 5% intrinsic resolution. A critical detector calibration issue was identified and corrected: an antenna orientation error in the Offline software had reversed the east-west polarization by 180◦, which was distorting the geomagnetic emission pattern and corrupting energy reconstruction. After correction, data from January-September 2024 showed the expected geomagnetic dominance. For the full deployment period (December 2024 to October 2025), approximately 12,000 hybrid events were successfully reconstructed. The radio detector achieves reconstruction efficiency exceeding 95% at zenith angles above 73◦ and energies above 1018.6 eV, with angular precision improving from ∼ 0.3◦ at 5 stations to <0.15◦ for events with more than 15 stations. The median opening angle between radio and WCD reconstructed directions is 0.25◦, with 68% of events showing agreement better than 0.36◦. Measured data validating the reliable and unbiased reconstruction performance of two independent energy measurements for inclined air showers: EWCD (muon-dominated energy) and ERD (calorimetric electromagnetic energy), establishing the necessary foundations for the hybrid photon-hadron discrimination parameter used in Chapter 7.

Chapter 6 develops a comprehensive quality selection framework that ensures high-purity data essential for photon searches by operating at both station and event levels to suppress background contamination while preserving genuine air-shower signals. At the station level, three complementary physics-motivated checks replace the standard SNR threshold to achieve substantial background rejection: (1) a dual-window linear SNR criterion requiring SNR′ ≥ 3.5 in the v×B component, which achieves 15× better noise rejection than the standard SNR ≥ 7.22 cut by adapting to local noise conditions rather than using fixed amplitude thresholds; (2) a cosine-similarity test against a bipolar template waveform constructed from 32,000 high-quality traces, requiring correlation ≥ 0.70 over a 200 ns window, which rejects 45.6% of false positives (primarily RFI and hardware transients) while retaining 96.7% of genuine air-shower signals; and (3) a polarization purity check requiring the geomagnetic v×B component to contribute at least 60% of the signal power at peak amplitude, eliminating residual outliers where the perpendicular component dominates contrary to air-shower physics. These station-level cuts, when combined, achieve a signal efficiency minus noise leakage fraction of 41.5% and reduce false positives by more than an order of magnitude compared to the standard method. At the event level, strict geometric and reconstruction quality requirements ensure reliable photon-hadron discrimination: minimum station multiplicity NRD ≥ 5 (39% of events), at least one station sampling the Cherenkov cone where radio emission peaks (84% of 5+ station events), lateral distribution fit quality χ2

ν ≤ 3 (78% retention), directional consistency between radio and WCD with opening angle ≤ 1◦ (92% retention), and thunderstorm flagging using meteorological data and satellite lightning observations to exclude periods of atmospheric electric field contamination (66% retention). Applied to the pilot dataset spanning October 2024 to February 2025, the complete quality selection retains 16% of initially reconstructed events (1005 of 6383).

Chapter 7 establishes photon–hadron separation using the hybrid energy mismatch between radio and Water Cherenkov reconstructions. The relative energy difference δ =(EWCD − ERD)/ERD serves as the primary discriminator, as photon-induced air showers yield systematically higher radio energies relative to WCD measurements compared to hadronic air showers. A background-only robust linear detrending removes residual instrumental and reconstruction trends, yielding an energy-stabilized observable δ′

. A single transparent cut, δ′ <δ′

cut, provides strong separation in the inclined hybrid simulations: for a representative working point retaining ∼ 80% of photons, no background event is observed to pass in the available hadronic sample, and the first nonzero leakage appears only when loosening the cut to achieve ϵγ ≃ 92%. These results validate the hybrid detection strategy for UHE photon searches at large zenith angles and demonstrate that a simple, physically motivated selection can deliver the high purity required for a rare-photon search. Chapter 8 builds the analysis chain needed to turn the selected hybrid data set into a quantitative photon search, defining how the available observing time and detector coverage translate into an exposure that can be used for flux constraints. Chapter 8 develops the exposure calculation and establishes a framework for applying the photon–hadron separation selection to real data. Using the period from December 2024 to October 2025, station operational status was tracked hourly by integrating trace-based activity, bad-period records, and array geometry. Active hexagons were counted in 12-hour bins to construct a time-dependent geometric exposure. For the zenith range 65◦–80◦, the cumulative geometric exposure reaches 850.4 km2 sr yr, while the fully efficient domain 73◦–80◦ accumulates 316.9 km2 sr yr. While this full-efficiency statement applies to hadronic air showers in the chosen domain, the expected photon sensitivity is obtained by folding the operated exposure with photon-specific reconstruction and selection efficiencies, which rise with energy. The chapter concludes by proposing a burn-sample strategy: 10% of the 2025 data (selected via event ID modulo 10 = 3) would provide ∼ 45 quality-selected events to validate the separation cut on real data while preserving 90% (corresponding to ∼ 400 events/year) for the final photon search, ensuring statistical rigor through blinded analysis protocols.

In conclusion, rigorous detector commissioning, data quality selection, and a transparent hybrid separation strategy provide the necessary tools and a statistical framework to conduct the next generation search for ultra-high-energy photons at the Pierre Auger Observatory. Crucially, the Radio Detector array is now fully deployed and operating smoothly, supported by the validated monitoring and quality frameworks developed herein, allowing for trusted data collection and enabling this new photon search to contribute significantly to the existing methods.