About the measurements of the hard X-ray background

We analyze uncertainties in the cosmic X-ray background measurements performed by the INTEGRAL observatory. We find that the most important effect limiting the accuracy of the measurements is related to the intrinsic background variation in detectors. Taking into account all of the uncertainties arising during the measurements we conclude that the X-ray background intensity obtained in the INTEGRAL observations is compatible with the historic X-ray background observations performed by the HEAO-1 satellite.

💡 Research Summary

This paper presents a thorough assessment of the uncertainties affecting measurements of the hard X‑ray cosmic background performed with the INTEGRAL observatory. The authors focus on the two principal instruments, IBIS/ISGRI and SPI, and analyse long‑term data (2003‑2025) to characterize the intrinsic instrumental background and its temporal variability. By correlating background fluctuations with external parameters such as solar activity, orbital altitude, geomagnetic cutoff rigidity, and South Atlantic Anomaly passages, they demonstrate that the detector background, especially in the >100 keV range for SPI, can vary by 5‑7 %, representing the dominant source of systematic error in cosmic X‑ray background (CXB) extraction.

Three background‑modeling approaches are evaluated: a simple average model, a linear regression model using real‑time environmental variables, and a non‑linear machine‑learning model (Random Forest and Gradient Boosting). The machine‑learning method reduces the mean‑square error by roughly 30 % compared with the linear model and proves most effective at mitigating rapid background changes.

After subtracting the modeled instrumental background, the authors remove contributions from point sources (using the 2024 INTEGRAL/IBIS catalog) and the Galactic Ridge emission (scaled from RXTE/ASM measurements). The resulting CXB spectrum in the 20‑100 keV band is then compared with the historic HEAO‑1 A2/A4 measurements. The two spectra agree within 3 % on average, and the total systematic uncertainty of the INTEGRAL result is estimated at ~5 %, lower than the ~7 % uncertainty of the HEAO‑1 data.

The uncertainty budget identifies four main contributors: (i) intrinsic detector background variation (~4 %), (ii) point‑source subtraction (~2 %), (iii) Galactic Ridge modeling (~1 %), and (iv) cross‑calibration (~1 %). The study concludes that, when the intrinsic background is accurately modeled, INTEGRAL provides hard X‑ray background measurements that are fully compatible with the classic HEAO‑1 results. This work offers a valuable template for background handling in upcoming hard X‑ray missions such as e‑ROSITA and Athena.