First application of the optimised next neighbour cleaning to data from imaging atmospheric Cherenkov telescopes

summary

The gist

Imaging atmospheric Cherenkov telescopes (IACTs) cameras are constantly subject to signals from the night sky background and electronic noise during event recordings.

In short

The optimized next neighbour (ONN) technique was implemented to clean data from VERITAS imaging atmospheric Cherenkov telescopes. By using two parameters—minimum pixel charge ($q_{min}$) and maximum time difference ($\delta t$)—the method suppresses noise while retaining Cherenkov light signals. This resulted in significantly increased effective areas, especially at lower energies, leading to up to an 82-fold increase in reconstructed excess events at energies below 1 TeV.

Key concepts

Optimised Next Neighbour (ONN) Technique
This method cleans noisy image data by dynamically setting selection cuts based on the minimum pixel charge ($q_{min}$) and the maximum arrival-time difference ($\delta t$) within a cluster of neighboring pixels. It filters out noise clusters by ensuring their properties fall below a threshold probability ($f_{acc}$), balancing noise suppression with signal retention.
Cleaning Thresholds ($q_{min}$, $\delta t$)
The cleaning process is defined by two parameters: $q_{min}$, the minimum total integrated charge required for a pixel cluster, and $\delta t$, the maximum allowed difference in pulse arrival time within that cluster. These parameters are tuned to maximize noise suppression while preserving pixels containing actual Cherenkov light signals.
Effective Area Improvement
The ONN method substantially increases the effective area of the telescope, particularly in the lower energy range (60-90 GeV). This gain is achieved by better retaining faint Cherenkov signals and reducing energy thresholds, allowing for more accurate reconstruction of gamma-ray spectra compared to standard cleaning methods.

Terminology used across episodes

This episode discusses

The paper

First application of the optimised next neighbour cleaning to data from imaging atmospheric Cherenkov telescopes · Read on arXiv

M. Kherlakian, G. Maier

Deutsches Elektronen-Synchrotron DESY · Humboldt-University zu Berlin · Ruhr-University Bochum Astronomisches Institut

Imaging atmospheric Cherenkov telescopes (IACTs) cameras are constantly subject to signals from the night sky background and electronic noise during event recordings. Image cleaning methods are typically employed to eliminate pixels that are mostly noise-contaminated and lack sufficient Cherenkov light. In this paper, we present the implementation of the optimised next neighbour (ONN) technique to the Very Energetic Radiation Imaging Telescope Array System (VERITAS) reconstruction pipeline. With this method, cleaning thresholds are derived in a two-parameter space, composed of the minimum total integrated charge of a cluster of pixels, q min, and by the maximum difference in the pulse arrival time within the cluster, δ t. The cleaning contours in q min - δ t are adjusted to set a high probability of noise suppression while optimising the retention of pixels containing Cherenkov light. In comparison to the VERITAS double-pass image cleaning method, effective areas are increased at the lower energy range (about 60 - 90 GeV) by a factor of up to about 25. Energy thresholds are reduced by up to about 22%. Testing the method on Crab Nebula data resulted in an increased number of reconstructed excess events at energies < 1 TeV by a factor of up to about 82. Finally, an increase in the differential sensitivity of VERITAS by a factor of up to about 8 was observed at about 20 GeV. With the addition of BDT-based γ /hadron separation cuts, the method improves the detection significance of a 20 hour Crab Nebula dataset, while excess event rates are enhanced by about 16%. In summary, the ONN image cleaning provides a reduced energy threshold and a higher rate of event reconstruction. The technique is especially advantageous for the analysis of gamma-ray sources presenting transient and soft spectra.

DOI: 10.1051/0004-6361/202659160

Transcript

Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.

Vera: We're looking at the paper: "First application of the optimised next neighbour cleaning to data from imaging atmospheric Cherenkov telescopes".

Jocelyn: Here's the gist of the paper. Imaging atmospheric Cherenkov telescopes (IACTs) cameras are constantly subject to signals from the night sky background and electronic noise during event recordings.

Vera: In this paper, we present the implementation of the optimised next neighbour (ONN) technique to the Very Energetic Radiation Imaging Telescope Array System (VERITAS) reconstruction pipeline.

Jocelyn: With this method, cleaning thresholds are derived in a two-parameter space, composed of the minimum total integrated charge of a cluster of pixels, qmin, and by the maximum difference in the pulse arrival time within the cluster, δt.

Vera: The cleaning contours in qmin − δt are adjusted to set a high probability of noise suppression while optimising the retention of pixels containing Cherenkov light.

Paper summary: Vera: So this paper is about a new image cleaning technique called the optimised next neighbour or ONN for the VERITAS reconstruction pipeline. It tackles the problem of keeping good Cherenkov light signals while getting rid of that constant noise from the night sky background and electronic noise during event recordings.

Jocelyn: Exactly, Vera. The main idea is to derive cleaning thresholds in a two-parameter space defined by something called qmin, which is the minimum total integrated charge of a cluster of pixels, and delta t max, which is the maximum difference in pulse arrival time within that same cluster.

Subrahmanyan: That sounds like you're looking at how to dynamically set the cuts based on local pixel properties and timing differences rather than using fixed thresholds. How does this approach actually balance keeping good signal pixels with suppressing noise?

Vera: Well, they found that by adjusting the cleaning contours in the qmin versus delta t max space, they can set a high probability of noise suppression while still optimizing how many pixels containing Cherenkov light they keep. They found a maximum probability of noise cluster retention called facc equal to five times ten to the negative four provides that sweet spot.

Jocelyn: So that specific value, facc equals five times ten to the negative four, is what they used as the optimal balance between suppressing noise, keeping Cherenkov signals, and preserving those image shapes. That sounds like a very specific tuning process for their cleaning algorithm.

Subrahmanyan: I see. It’s about finding that precise parameter setting where you get the best trade-off across all three goals: noise suppression, signal retention, and image morphology preservation. What were the initial benchmarks they used to determine this facc value?

Vera: They looked at Monte Carlo simulations of gamma ray events and cleaned them assuming different values for facc to see what happened. They found that when they use this method, compared to the VERITAS double-pass cleaning method, they see an increase in effective areas at the lower energy range, specifically around sixty to ninety giga-electron volts by a factor of up to twenty five.

Jocelyn: That's a significant gain in sensitivity for that lower energy band. And they also managed to reduce the energy thresholds by up to twenty two percent with this ONN method. That sounds like it helps them see fainter things better at the low end of their spectrum.

Subrahmanyan: Reducing the energy threshold by twenty two percent is a substantial improvement when you're trying to reconstruct spectra, which is what we really care about in cosmology. What did they find when they tested this method on real data from the Crab Nebula?

Vera: They tested it on Crab Nebula data and found that there was an increased number of reconstructed excess events at energies below one tera electron volt by a factor of up to eighty two. That's quite a jump in the event counts for those lower energy ranges.

Paper summary: Jocelyn: Eighty two times more events below one tera electron volt is impressive, especially when you consider the overall sensitivity improvement they reported later, which was about eight times higher differential sensitivity around twenty giga-electron volts.

Subrahmanyan: So we're talking about a factor of eight increase in differential sensitivity at twenty giga-electron volts. That points toward better resolution in that specific energy regime, which is important for probing structure formation models. What’s the big picture here for us?

Vera: This method seems especially advantageous for faint and transient signals because it allows them to use that lower energy threshold more effectively when reconstructing energy spectra. It should be suitable for any IACT data, including CTAO, MAGIC, and HESS, as long as they have arrival timing of extensive air showers available in the nanosecond range.

Jocelyn: So the paper "First application of the optimised next neighbour cleaning to data from imaging atmospheric Cherenkov telescopes" suggests that this ONN technique offers a way to clean data that is more tailored to the local noise environment and signal characteristics. It’s not just another fixed filter, but something dynamic based on how close pixels are and how long their light pulses arrive.

Subrahmanyan: I think the real implication is how much cleaner we can get our fundamental measurements from these instruments, allowing us to probe fainter sources or lower energy phenomena more reliably. If this works as well across different telescopes like CTAO, it sets a new standard for data processing pipelines in this field.

Vera: Right. So they’ve shown that by using this ONN procedure, they can get better effective areas at low energies and improve the energy resolution below one tera electron volt by about fifteen percent for their specific configuration. That directly relates to reducing the energy bias when you're looking at those sub-hundred giga-electron volt events.

Jocelyn: So what this means for pulsar timing observations is that if we can clean our Cherenkov images this well, we might be able to detect fainter signals or see features in the light curve that are currently lost in the noise floor because of poor image quality.

Subrahmanyan: We need to keep an eye on whether this method scales as well when you move from VERITAS data to other array systems. The success of this ONN technique hinges on that timing information being available with good resolution, which is a key limitation they mention for its general applicability.

Vera: That's the caveat they flag: it depends entirely on having that arrival timing of extensive air showers available at a reasonable nanosecond resolution, otherwise the method’s power is limited. That’s the practical constraint we have to keep in mind.

Jocelyn: So we see a method that boosts event retention significantly in real data like Crab Nebula, gives us better energy thresholds, and improves sensitivity across the board while being flexible enough for different instruments. It seems like a solid step forward for how we clean these Cherenkov images.

Conclusion: Vera: So to wrap up this paper, we're looking at how they put their new optimised next neighbour cleaning technique onto real data from VERITAS.

Jocelyn: It was all about taking a complex noise problem and finding a way to clean the image better than before.

Subrahmanyan: The core of it is using these two parameters, qmin for minimum charge and delta t max for the maximum arrival time difference within a pixel cluster.

Vera: Exactly. They found that by adjusting those boundaries in this parameter space, you can set a cleaning probability called facc equal to five times ten to the negative four.

Jocelyn: That specific tuning allows them to balance suppressing noise against keeping the actual Cherenkov light signals intact while preserving the shape of the image.

Subrahmanyan: And they showed that this method improves things in real data, like on Crab Nebula, by increasing counts below one tera electron volt by a factor of eighty two.

Vera: It also boosts the differential sensitivity around twenty giga electron volts by a factor of eight. That means better resolution at those lower energies.

Jocelyn: This whole thing seems really useful for faint or transient signals because it lets you use that lower energy threshold more effectively when you're trying to reconstruct an energy spectrum.

Subrahmanyan: It should be applicable across different IACT instruments like CTAO and HESS, as long as we have that nanosecond timing information available for the air showers.

Vera: The authors pointed out a major caveat though: this method really depends on having that good arrival timing resolution in the nanosecond range.

Jocelyn: So, they've shown a way to clean data more intelligently based on local noise and signal characteristics, offering real gains in sensitivity and energy resolution.

Subrahmanyan: That kind of tailored cleaning procedure is what we need when trying to push our limits on probing the structure formation of the universe.

More episodes

← Home