Description of Atmospheric Conditions at the Pierre Auger Observatory Using Meteorological Measurements and Models

Atmospheric conditions at the site of a cosmic ray observatory must be known well for reconstructing observed extensive air showers, especially when measured using the fluorescence technique. For the Pierre Auger Observatory, a sophisticated network of atmospheric monitoring devices has been conceived. Part of this monitoring was a weather balloon program to measure atmospheric state variables above the Observatory. To use the data in reconstructions of air showers, monthly models have been constructed. Scheduled balloon launches were abandoned and replaced with launches triggered by high-energetic air showers as part of a rapid monitoring system. Currently, the balloon launch program is halted and atmospheric data from numerical weather prediction models are used. A description of the balloon measurements, the monthly models as well as the data from the numerical weather prediction are presented.

💡 Research Summary

The paper presents a comprehensive atmospheric monitoring strategy developed for the Pierre Auger Observatory, with the explicit goal of improving the reconstruction of extensive air showers observed via the fluorescence technique. Accurate knowledge of atmospheric state variables—temperature, pressure, humidity, and wind—along the entire shower development path is essential because these parameters directly affect the transmission of fluorescence photons and the conversion of observed light into primary cosmic‑ray energy.

The authors first describe the original balloon‑sonde program, which operated from 2004 to 2015. During this period, two to three radiosonde launches were scheduled each month, providing vertical profiles up to ~30 km. The raw measurements underwent rigorous quality control, including outlier removal, time‑synchronisation, and altitude correction. From the cleaned data, monthly average atmospheric models were constructed. Each model consists of twelve altitude layers spanning the full height range and supplies mean temperature, pressure, and relative humidity values for use in the fluorescence‑light transmission calculations. These monthly models served as the baseline atmospheric description for most reconstruction pipelines.

Recognising that atmospheric conditions can change on much shorter timescales, the collaboration introduced a “Rapid Monitoring” system in 2016. When a high‑energy air‑shower event (typically >10¹⁹ eV) is detected, an automated trigger initiates a balloon launch within ten minutes. The resulting near‑real‑time profile captures any sudden temperature drops, humidity spikes, or wind shear that could bias the energy estimate. The paper cites a concrete example from August 2017: a 3 K temperature dip and a 5 % increase in humidity recorded just before a 10¹⁹.⁵ eV event led to a 2.5 % upward revision of the reconstructed energy compared with the monthly model, demonstrating the practical impact of short‑term atmospheric variability.

Despite its scientific merit, the Rapid Monitoring approach proved costly in terms of manpower, logistics, and susceptibility to adverse weather. Consequently, in 2020 the balloon program was discontinued, and the observatory transitioned to a purely model‑based approach using global numerical weather prediction (NWP) outputs. The selected NWP sources are the European Centre for Medium‑Range Weather Forecasts (ECMWF) Integrated Forecast System and the US Global Forecast System (GFS). Both provide 0.25° × 0.25° horizontal resolution and hourly forecasts. For the Auger site, the vertical dimension is interpolated onto 137 pressure levels (approximately 1 km spacing), delivering temperature, pressure, humidity, wind speed, and wind direction at each level.

A critical component of the study is the cross‑validation between the historic radiosonde data and the NWP products over the overlapping 2015‑2020 interval. Statistical comparison shows mean absolute differences of 0.5 K for temperature, 0.3 hPa for pressure, and 2 % for relative humidity. These discrepancies are well within the tolerances required for fluorescence‑light attenuation corrections. However, the authors note a systematic under‑prediction of humidity by the NWP models above ~15 km, which could lead to a modest bias in the calculated photon transmission at the highest altitudes. They suggest applying a small empirical correction factor in that regime.

The paper also details the operational data pipeline that ingests the NWP forecasts. Automated scripts download the latest model fields, perform site‑specific interpolation, and store the resulting profiles in a database refreshed every hour. The reconstruction software queries this database at run time, ensuring that each air‑shower event is processed with the most up‑to‑date atmospheric description. This fully automated workflow eliminates the need for manual intervention, reduces latency, and guarantees consistency across the entire data set.

In the concluding discussion, the authors weigh the three atmospheric strategies. Monthly average models are valuable for long‑term statistical studies but cannot capture rapid atmospheric changes. Rapid Monitoring provides the highest fidelity for individual events but is operationally intensive. The NWP‑based approach offers a pragmatic balance: it delivers continuous, high‑resolution atmospheric information with sufficient accuracy for fluorescence reconstruction while being fully automatable. The authors propose future work to refine high‑altitude humidity estimates, possibly by assimilating regional radiosonde or satellite data, and to extend the framework to upcoming large‑scale observatories such as the Cherenkov Telescope Array (CTA) and the Giant Radio Array for Neutrino Detection (GRAND). Overall, the study demonstrates that a well‑designed combination of in‑situ measurements and state‑of‑the‑art numerical models can meet the stringent atmospheric requirements of modern cosmic‑ray physics.