---
title: 'Dark100 Array: Ultra-High-Energy Cherenkov Telescopes'
url: https://www.emergentmind.com/topics/dark100-array
type: topic
---

# Dark100 Array: Ultra-High-Energy Cherenkov Telescopes

Searching arXiv for the cited Dark100/PANOSETI papers and closely related work.
arXiv search: Dark100 / PANOSETI UHE gamma-ray astronomy
Searching for: Dark100 PANOSETI ultra-high-energy gamma-ray astronomy
Dark100 Array is a planned ultra-high-energy Cherenkov telescope array based on the Panoramic Search for Extraterrestrial Intelligence (PANOSETI) telescope system and configured to operate as an imaging atmospheric Cherenkov telescope array for $\gamma$ rays in the $10\,\mathrm{TeV}$–PeV regime. In the cited descriptions, Dark100 is presented as a purpose-built system of small, cost-effective, wide-FoV telescopes optimized for Galactic PeVatron searches, ultra-heavy dark matter studies, and ultra-fast optical transients. The first phase comprises three telescopes at Palomar Observatory, with a plan to expand to six units; the array is designed to run continuously for at least five years [2508.05548].

## 1. Array definition, layout, and deployment

Dark100 is funded for up to six $0.5\,\mathrm{m}$ Fresnel-lens telescopes at Palomar Observatory, with room for future expansion. The first three telescopes will be located at the “Ferns-North,” “PTI-Heli,” and “Winter” sites, which form an approximately equilateral triangle with side lengths of order $150$–$200\,\mathrm{m}$; this is described as identical in geometry to the March/October 2024 Lick pathfinder array where the mean spacing was $169\,\mathrm{m}$. A fourth telescope will occupy an existing building at the “Dog Seismo” site, and additional piers and Fornax enclosures are pre-built at the other three pads [2508.05548].

The technical overview likewise describes an initial configuration of three telescopes arranged in a roughly triangular pattern on the Palomar plateau, followed by future expansion to six telescopes deployed around the array centroid. In that same overview, simulated Cherenkov-light pool coverage is specified through shower-generation regions of radius $400\,\mathrm{m}$ for $E_\gamma=10$–$60\,\mathrm{TeV}$, $500\,\mathrm{m}$ for $E_\gamma=100$–$200\,\mathrm{TeV}$, and $600\,\mathrm{m}$ for $E_\gamma=1$–$2\,\mathrm{PeV}$ [2508.01354].

The installation timeline is stated explicitly. Cable trenching, pier installation, and enclosure delivery are already under way as of mid-2025. The first three telescopes will be commissioned and begin science operations before the end of calendar 2025. Dark100 is designed to operate for at least five years, with a program that includes mapping the Galactic plane, monitoring transients near the Galactic Center, and searching for indirect signatures of ultra-heavy dark matter with $m_\chi>100\,\mathrm{TeV}$ [2508.05548].

## 2. Telescope optics, focal plane, and acquisition architecture

Each Dark100 telescope employs a single $0.5\,\mathrm{m}$-diameter acrylic Fresnel lens optimized for nanosecond-scale transient imaging. The focal length is approximately $0.5\,\mathrm{m}$, yielding an extremely fast, wide-angle optical system with a point-spread function of order a few arcminutes across the $10^\circ\times10^\circ$ field of view. In the technical report, the typical point-spread function is also summarized as $\lesssim0.3^\circ$, dominated by pixel size [2508.05548].

The focal plane is populated with $1024$ silicon-photomultiplier pixels arranged as a $32\times32$ array, or equivalently four $8\times8$ quadrants within the camera architecture. The pixel angular size is given as $\simeq(10^\circ/32)\simeq0.31^\circ$ on a side in one description and as $0.30^\circ\times0.30^\circ$ in the technical overview; in both cases, the total field of view per telescope is $10^\circ\times10^\circ$ [2508.01354].

Readout is FPGA-based and sampled at nanosecond resolution, approximately $1\,\mathrm{ns}$ per sample. Waveforms are buffered locally, and full-camera frames are read out on any coincident-pixel trigger. A telescope issues a trigger when at least two pixels within the same 256-pixel quadrant exceed a programmable threshold, typically $6.5$–$7.5$ photoelectrons. This two-pixel logic keeps single-telescope rates at $O(1\,\mathrm{Hz})$ under dark skies, including the October 2024 Crab run with Jupiter in the FoV. All telescopes stream triggered event packets over Gigabit Ethernet to a single Data Acquisition Unit (“DACQ”) rack in the Hale dome basement, where event building, time-stamping, and storage are handled in software [2508.05548].

These design choices are closely tied to the scientific use case. The combination of a very wide $10^\circ\times10^\circ$ FoV, nanosecond-responsive SiPM cameras, and a compact, modular array is described as enabling simultaneous monitoring of large sky regions for transients and deep pointed observations of candidate dark-matter targets and known PeV accelerators [2508.01354].

## 3. Effective area, threshold, and directional performance

Monte Carlo simulations indicate an energy threshold $E_{\mathrm{th}}$ of order $10\,\mathrm{TeV}$ for gamma rays, with full efficiency above approximately $30\,\mathrm{TeV}$. At threshold, the effective area is stated as $A_{\mathrm{eff}}(E_{\mathrm{th}})\approx10^4$–$10^5\,\mathrm{m}^2$ per telescope pair. A convenient parametrization quoted for $E\gtrsim30\,\mathrm{TeV}$ is

$$
A_{\mathrm{eff}}(E)\simeq A_0\cdot(E/100\,\mathrm{TeV})^\alpha,
$$

with $A_0\simeq5\times10^4\,\mathrm{m}^2$ and $\alpha\simeq0.3$ [2508.05548].

The technical overview gives the effective collection area for the three-telescope array as $\sim0.1\,\mathrm{km}^2$ at $10\,\mathrm{TeV}$, growing with higher energies and additional telescopes. It further states $A_{\mathrm{eff}}(10\,\mathrm{TeV})\approx0.10\,\mathrm{km}^2$, growth by a factor of $\sim3$ at $100\,\mathrm{TeV}$, and $A_{\mathrm{eff}}(1\,\mathrm{PeV})>0.3\,\mathrm{km}^2$ [2508.01354].

Angular performance depends strongly on event multiplicity. For Lick-array geometry, simulations yield a $68\%$ containment radius of $0.4^\circ$ when exactly two telescopes participate, improving to $0.19^\circ$ with three-telescope events. Energy resolution is described as not yet fully calibrated, although preliminary studies predict $\Delta E/E\lesssim50\%$ at $E\approx100\,\mathrm{TeV}$ [2508.05548].

The same body of work states that the operational energy threshold is limited by photon yield on the small mirror and places it at $\gtrsim10\,\mathrm{TeV}$. Above $100\,\mathrm{TeV}$, $\gamma$/hadron separation reaches $\epsilon_\gamma>80\%$ at $\epsilon_p<20\%$. This suggests that the intended performance regime is not the low-TeV domain usually emphasized by large-aperture IACTs, but the tens-of-TeV to PeV window where sparse coverage and large collection area become decisive [2508.01354].

## 4. Background rejection and $\gamma$/hadron separation

Night-sky-background triggers are reduced to $\lesssim1\,\mathrm{Hz}$ per telescope by the $\ge2$-pixel requirement at $7.5\,\mathrm{PE}$. For astrophysical analysis, cosmic-ray rejection exploits Hillas parameters; in the deployment paper, the “distance,” defined as the angular separation between image centroid and source direction, is singled out as showing strong discrimination. Defining

$$
\mathrm{max\_distance}=\max\{\mathrm{distance}_i\}
$$

over all imaging telescopes, $\gamma$ rays peak at $1$–$2^\circ$, while protons populate the camera edges; the quoted selection is $\mathrm{max\_distance}<2^\circ$. A further cut on reconstructed arrival direction,

$$
\theta^2<0.32^\circ{}^2,
$$

that is $\lesssim0.57^\circ$, isolates signal under a known source position [2508.05548].

A more explicit classification study uses single-telescope Hillas parameters and a boosted-decision-tree classifier. The simulation chain is based on CORSIKA v7.8 with QGSJET-III-01, DPMJET III, EPOS LHC-R, and SIBYLL 2.3e as high-energy hadronic models, together with UrQMD 1.3_cors as the low-energy hadronic model. The simulated $\gamma$-ray source is a point source at camera center with zenith $=20^\circ$ and azimuth $=0^\circ$, while the proton background is diffuse within an $8^\circ$ half-angle cone. Monoenergetic energies are $E_\gamma=\{10,30,100\,\mathrm{TeV},1\,\mathrm{PeV}\}$ and $E_p=2E_\gamma$ [2508.01354].

The image parameterization extracts the classic Hillas parameters and related quantities: length $\ell$, width $w$, size $S=\sum_{\mathrm{pixel}}\mathrm{Charge}$, distance $d$, miss $m$, alpha $\alpha$, and shower core distance $R_{\mathrm{core}}$. Discrimination exploits that $\gamma$-ray images are on average narrower, shorter, oriented towards the camera center, and more symmetric. In the initial BDT study, the input features are $\{\ell,w,S\}$, with future work to add scaled parameters and multi-telescope variables. The reported figures of merit are

$$
\mathrm{tpr}=\frac{TP}{TP+FN},\qquad \mathrm{fpr}=\frac{FP}{FP+TN},
$$

$$
Q\equiv\frac{\epsilon_\gamma}{\sqrt{\epsilon_p}},
$$

and

$$
S_{\mathrm{discr}}\equiv\frac{|\langle D\rangle_\gamma-\langle D\rangle_p|}{\sqrt{\sigma^2_{D,\gamma}+\sigma^2_{D,p}}}.
$$

At $E_\gamma=1\,\mathrm{PeV}$ and QGSJET, the ROC curve yields $\mathrm{AUC}\approx0.91$; for a working point with $\mathrm{fpr}\approx0.10$, one obtains $\mathrm{tpr}\approx0.85$, hence $Q\approx0.85/\sqrt{0.10}\approx2.7$ [2508.01354].

The treatment of hadronic model dependence is notable because it addresses a standard systematic concern in Cherenkov astronomy. In the cited study, the proton-image distributions of $\ell$, $w$, $m$, $d$, $\alpha$, and $S$ at $E=200\,\mathrm{TeV}$ are described as statistically indistinguishable across QGSJET, DPMJET, EPOS, and SIBYLL, leading to the conclusion that the systematic uncertainty from hadronic model choice is negligible at this level [2508.01354].

## 5. Pathfinder deployments and Crab Nebula analysis

Before full deployment at Palomar, three PANOSETI telescopes were deployed twice in the same temporary configuration at Lick Observatory in March and October 2024. The March 2024 Lick deployment used a triangle of $169\,\mathrm{m}$ sides and provided raw trigger rates, image sizes, and Hillas distributions that agree with full-ray-trace plus shower simulations to within $\lesssim20\%$ [2508.05548].

Those pathfinder data were used to validate the reduction chain. Cuts on $\mathrm{max\_distance}$ and $\theta^2$ reduce the residual proton background by a factor $\gtrsim10$ while retaining $\gtrsim50\%$ of gammas above $30\,\mathrm{TeV}$. This establishes a direct empirical link between the simulated selection strategy and on-sky data [2508.05548].

The Crab Nebula observations were conducted from Oct 30 to Nov 1, 2024, with $5$ hours of exposure. Pointing was fixed to $\mathrm{RA}=05^\mathrm{h}44^\mathrm{m}32^\mathrm{s}$, $\mathrm{Dec}=+22^\circ01^\prime03^{\prime\prime}$ to keep Jupiter out of the trigger region. The reported counts are $5$ on-source events and $248$ off-source events, using the remainder of the FoV as background, with $\alpha=1/109$. The Li and Ma significance is given by

$$
S=\sqrt{2}\left[N_{\mathrm{on}}\ln\!\left(\frac{1+\alpha}{\alpha}\cdot\frac{N_{\mathrm{on}}}{N_{\mathrm{on}}+N_{\mathrm{off}}}\right)+N_{\mathrm{off}}\ln\!\left((1+\alpha)\cdot\frac{N_{\mathrm{off}}}{N_{\mathrm{on}}+N_{\mathrm{off}}}\right)\right]^{1/2}=1.55\,\sigma,
$$

with excess

$$
N_{\mathrm{on}}-\alpha N_{\mathrm{off}}=2.7\ \mathrm{events}.
$$

That result is described as fully consistent with the Crab spectrum measured by LHAASO folded through $A_{\mathrm{eff}}\approx10^4$–$10^5\,\mathrm{m}^2$ at $E_{\mathrm{th}}\sim10\,\mathrm{TeV}$, that is, only a few $\gamma$ rays expected in $5$ h [2508.05548].

## 6. Scientific program and projected sensitivity

Dark100 is tailored to three flagship investigations: ultra-heavy dark matter annihilation or decay for $100\,\mathrm{TeV}\lesssim m_{\mathrm{DM}}\lesssim$ tens of PeV, Galactic PeVatrons, and ultra-fast optical transients. For Galactic PeVatron searches, the stated strategy is to survey the Galactic plane or target known LHAASO sources, many of which lack TeV counterparts. Over $O(1000\,\mathrm{h})$ of exposure, the array is projected to push $5\,\sigma$ detection thresholds to fluxes $\phi(E>100\,\mathrm{TeV})\lesssim10^{-14}\,\mathrm{cm}^{-2}\,\mathrm{s}^{-1}$ [2508.05548].

For dark matter, the stated target class is decaying or annihilating $\chi$ with $m_\chi>100\,\mathrm{TeV}$, which can produce final-state $\gamma$ rays above $10\,\mathrm{TeV}$. The wide FoV allows deep exposures of, for example, the Galactic Center halo. The flux relation quoted for limit setting on $\langle\sigma v\rangle$ or $\tau_\chi$ is

$$
\Phi_\gamma(E)=\frac{1}{4\pi m_\chi^2}\,\langle\sigma v\rangle\,\frac{dN_\gamma}{dE}\times J,
$$

where

$$
J=\int \rho^2(r)\,dl
$$

along the line of sight [2508.05548].

The sensitivity discussion is tied to explicit counting estimates. For any target, the detection significance after $T$ hours is computed via the standard Li and Ma formula using on/off counts obtained after optimized image and direction cuts. With six telescopes and typical $1\,\mathrm{Hz}$/telescope trigger rates, Dark100 expects $O(10^5)$ background events per $1000\,\mathrm{h}$; gamma-selection cuts reduce this to $O(10^3)$, so a $5\,\sigma$ threshold requires $O(25)$ $\gamma$ rays above background. The cited estimate states that this is reachable for sources at or above $\sim10\%$ of the Crab flux at $E>50\,\mathrm{TeV}$ over multi-year integration [2508.05548].

The same scientific framing includes dwarf spheroidals as candidate dark-matter targets and emphasizes that the array’s nanosecond-responsive cameras also support the search for ultra-fast optical transients. A plausible implication is that Dark100 is intended as a hybrid facility: its architecture is specialized for ultra-high-energy Cherenkov astronomy, yet it preserves the transient-imaging functionality inherited from PANOSETI [2508.01354].

Source: https://www.emergentmind.com/topics/dark100-array