Background and Transient Observer (BTO)

• BTO is a design principle that integrates continuous background characterization with rapid detection of transient events in a variety of instruments.
• It underpins systems like COSI’s NaI(Tl) detectors and optical surveys, effectively linking background surveying with real-time alert responses.
• BTO principles inform signal processing and trigger logic, enhancing the detection and analysis of phenomena such as GRBs, magnetar flares, and TGFs.

Searching arXiv for recent and foundational papers on the Background and Transient Observer (BTO) and closely related systems. arxiv_search(query="Background and Transient Observer COSI BTO", max_results=10) Background and Transient Observer (BTO) denotes both a specific instrument and a broader observational pattern. In the specific sense, BTO is a student collaboration soft gamma-ray payload for the Compton Spectrometer and Imager (COSI) Small Explorer mission, intended to measure background radiation in orbit and detect gamma-ray bursts (GRBs) and other gamma-ray transients in the $30~\mathrm{keV}$–$2~\mathrm{MeV}$ band (Gulick et al., 2024). In the broader systems sense, the same term describes an architecture that maintains a continuously updated characterization of a background—whether an optical sky survey, an alert stream, a detector environment, or a diffuse radiation field—while reacting to transient deviations with low latency and context-aware analysis (Dyer et al., 2022). This dual usage is consistent across current literature: some works treat BTO as a named mission element, while others use it as a conceptual template for survey design, alert handling, and background-aware inference.

1. Scope and meaning of the term

In the COSI literature, BTO is a proper noun: a dedicated instrument consisting of two NaI(Tl) detectors with student-designed electronics, mounted on the spacecraft to extend COSI’s spectral response below $200~\mathrm{keV}$ and to monitor the orbital background environment (Gulick et al., 2024). In parallel, several papers use “Background and Transient Observer” as a systems concept. The Gravitational-wave Optical Transient Observer (GOTO) explicitly formulates a “background” layer as a regularly updated, homogeneous survey of the visible sky and a “transient observer” layer as the responsive component that reacts to alerts, tiles error regions, and identifies new sources through difference imaging; the same paper states that GOTO is “effectively an operational prototype of such a BTO system” (Dyer et al., 2022).

This broader usage extends beyond optical surveys. The Cherenkov Telescope Array Observatory Transients Handler is described as the subsystem that watches alert streams in the background, matches them against science configurations, and reacts when a transient is important enough to interrupt or reshape the observing schedule (Collins et al., 28 Aug 2025). The H.E.S.S. follow-up system provides a similarly integrated pattern of alert ingestion, ranking, automated execution, and real-time analysis (Hoischen et al., 2022). At the algorithmic level, background-aware transient timing with Bayesian Blocks and the Astro-BEATS segmentation pipeline both implement the same logic: estimate background, normalize or subtract it, and detect coherent deviations in time or space-time (Worpel et al., 2015). A still more abstract usage appears in work on observer motion and cosmic backgrounds, where the “background” is an isotropic monopole spectrum and the “observer” is a moving detector whose transient frame introduces calculable anisotropy (Trombetti, 21 Mar 2025).

A recurrent misconception is that “background” in BTO necessarily means nuisance counts alone. In the literature, background can instead denote the soft gamma-ray and high-energy particle environment in low Earth orbit, the reference images and baseline light curves of a sky survey, the current schedule and alert state of an observatory control system, or the monopole spectrum of a diffuse cosmic field (Gulick et al., 27 Jan 2025). The common element is not a specific wavelength or hardware type, but the coupling of continuous context-building with prompt transient discrimination.

2. BTO on COSI: mission role and instrument architecture

COSI is a NASA Small Explorer mission scheduled for launch in 2027 and designed to survey the sky in the $0.2$–$5~\mathrm{MeV}$ range with germanium detectors, providing imaging, spectroscopy, and polarimetry (Gulick et al., 2024). BTO is the associated lower-energy payload. It is designed to extend COSI’s reach down to $30~\mathrm{keV}$, provide a very large field of view, and measure the gamma-ray background along the orbit. The combined system yields simultaneous spectral coverage from $30~\mathrm{keV}$ to $5~\mathrm{MeV}$, with the shared $30~\mathrm{keV}$–$2~\mathrm{MeV}$ interval serving as the principal joint-analysis band (Gulick et al., 2024).

Each BTO detector is a NaI(Tl) scintillator read out by SiPMs. The flight geometry described for the COSI payload uses rectangular crystals of $2~\mathrm{MeV}$0 in $2~\mathrm{MeV}$1 aluminum housings, mounted on pedestals on opposite sides of the COSI Payload Interface Plate (Gulick et al., 2024). The associated electronics paper describes the detector modules as Scionix NaI(Tl)+SiPM assemblies with three ArrayJ-60035-4P-PCB SiPMs summed per crystal, a direct output for large-signal discrimination, and an amplified output for standard energy measurement (Nagasawa et al., 31 Aug 2025). The detectors are elevated so that each BTO field of view overlaps the full COSI field of view, while spacecraft shadowing provides some rough directional discrimination outside COSI’s nominal imaging region (Gulick et al., 2024).

Aspect	BTO on COSI
Detector count	Two NaI(Tl) detectors
Science band	$2~\mathrm{MeV}$2–$2~\mathrm{MeV}$3
Field of view	$2~\mathrm{MeV}$4 of the sky
Shared COSI+BTO band	$2~\mathrm{MeV}$5–$2~\mathrm{MeV}$6
COSI main band	$2~\mathrm{MeV}$7–$2~\mathrm{MeV}$8
Reported resolution target	$2~\mathrm{MeV}$9 FWHM
Reported achieved resolution	$200~\mathrm{keV}$0 FWHM at $200~\mathrm{keV}$1

The mission role is explicitly dual. As a background monitor, BTO measures the $200~\mathrm{keV}$2–$200~\mathrm{keV}$3 radiation environment in a low-inclination low Earth orbit, including photons and high-energy particles relevant to COSI’s sensitivity model (Gulick et al., 27 Jan 2025). As a transient observer, it is designed to detect GRBs, magnetar flares, and terrestrial gamma-ray flashes (TGFs), and to trigger a higher-fidelity readout mode when a statistically significant excess occurs (Gulick et al., 2024). This duality is intrinsic to the instrument definition rather than an after-the-fact interpretation.

3. Readout, observing modes, and background-aware signal processing

The BTO readout is partitioned into an analog board and a digital board (Nagasawa et al., 31 Aug 2025). The analog board amplifies the SiPM signals, generates triggers, drives analog-to-digital conversion, and uses separate discriminator paths for standard events and very large deposits. The digital board is built around a Microchip SAMV71 microcontroller; its firmware controls the analog chain, reads ADC data via SPI, adds timestamps, buffers event data, and interfaces with the spacecraft command and telemetry system (Nagasawa et al., 31 Aug 2025). The reported processing time is $200~\mathrm{keV}$4 per event, implying sustained rates of order $200~\mathrm{keV}$5 before dead-time losses become important (Nagasawa et al., 31 Aug 2025).

Two operating modes recur across the COSI BTO papers. In event-by-event mode, the system records individual photon energies and timestamps, which is essential for short transients such as TGFs and for fine spectral evolution in GRBs and magnetar flares (Gulick et al., 27 Jan 2025). In binned mode, BTO accumulates background information continuously in histograms. The detector trade-study describes this as background data in histograms with a relative time resolution of $200~\mathrm{keV}$6 (Gulick et al., 27 Jan 2025), whereas the mission-science description presents all counts above threshold as accumulated into histograms with $200~\mathrm{keV}$7 time bins (Gulick et al., 2024). The DAQ paper, by contrast, emphasizes the implementation detail that binned-mode science products are aggregated into coarse energy histograms over $200~\mathrm{keV}$8 intervals while discriminator counts are tracked in $200~\mathrm{keV}$9 bins (Nagasawa et al., 31 Aug 2025). This suggests an evolving implementation across design stages rather than a disagreement about the underlying two-mode concept.

The signal chain is explicitly background-aware. Heavy ions penetrating the NaI(Tl) crystal produce large pulses and long afterglow tails, so BTO uses a direct-output “SuperUpper discriminator” to identify deposits of $0.2$0, timestamps those events, and records them as indicators of time intervals likely affected by afterglow (Nagasawa et al., 31 Aug 2025). A separate upper discriminator on the amplified path marks large pulses above $0.2$1. Pole-zero cancellation is tuned to restore baseline rapidly after such deposits, and the digital system can generate veto signals and transient-trigger logic onboard (Nagasawa et al., 31 Aug 2025). The same paper reports a wide bandpass and $0.2$2 FWHM at $0.2$3, thereby exceeding the requirement of $0.2$4 FWHM over the $0.2$5–$0.2$6 range (Nagasawa et al., 31 Aug 2025).

The onboard transient trigger is a rate-excess detector operating on $0.2$7 cadence. A short-term running average over the last five bins is compared against a background estimate from the interval $0.2$8–$0.2$9 before the current time, and a trigger is issued when the short-term rate exceeds the background mean by $5~\mathrm{MeV}$0, with $5~\mathrm{MeV}$1 by default (Nagasawa et al., 31 Aug 2025). This is a standard background-plus-transient split: low-telemetry accumulation during quiescence, followed by state transition into event-by-event recording when the background model is no longer sufficient to explain the observed count stream.

4. Science program and expected return

The principal science drivers are GRBs, magnetar giant flares, and TGFs (Gulick et al., 2024). For GRBs, BTO supplies the low-energy side of the prompt spectrum, where $5~\mathrm{MeV}$2 frequently lies below $5~\mathrm{MeV}$3, while COSI constrains the harder tail up to several MeV. The combined instrument therefore improves recovery of $5~\mathrm{MeV}$4, the low-energy spectral index, the total fluence, and the separation of photospheric or thermal components from non-thermal emission (Gulick et al., 2024). The simulations described for one long GRB and one short GRB show BTO spectra spanning $5~\mathrm{MeV}$5–$5~\mathrm{MeV}$6 and COSI spectra extending to $5~\mathrm{MeV}$7, with the overlap region supporting cross-calibration and joint fitting (Gulick et al., 2024).

The estimated GRB detection rate is one of the most concrete published performance claims. Using MEGAlib and BATSE-like long GRBs, Earth-occultation cuts, and a background rate of $5~\mathrm{MeV}$8 per detector, the mission-science paper estimates approximately $5~\mathrm{MeV}$9 GRBs per year, with most events being long bursts (Gulick et al., 2024). The same discussion notes that comparison with Fermi GBM suggests this estimate may be slightly high, leading to a more realistic expectation of $30~\mathrm{keV}$0–$30~\mathrm{keV}$1 GRBs per year. For short GRBs, the paper reports about $30~\mathrm{keV}$2–$30~\mathrm{keV}$3 detections per year when GBM lightcurves are rescaled to BTO effective area and assessed with a $30~\mathrm{keV}$4 threshold (Gulick et al., 2024). The significance metric used in the GRB simulations is

$30~\mathrm{keV}$5

with $30~\mathrm{keV}$6 taken from the simulated orbital background (Gulick et al., 2024).

Magnetar giant flares occupy a complementary niche because their spectra peak in the soft gamma-ray band and their brightest Galactic prompt spikes can saturate large-area detectors. The BTO paper argues that BTO’s smaller effective area may be advantageous in such cases, because it should recover the decaying prompt flux and tail more quickly after the brightest interval (Gulick et al., 2024). A specific simulation based on the extragalactic giant-flare candidate GRB231115A adopts a Comptonized spectrum with $30~\mathrm{keV}$7, $30~\mathrm{keV}$8, and fluence $30~\mathrm{keV}$9 in $30~\mathrm{keV}$0–$30~\mathrm{keV}$1, and shows that BTO can obtain useful statistics up to about $30~\mathrm{keV}$2 for such an event (Gulick et al., 2024).

TGFs are a distinct case because COSI’s main germanium telescope suppresses photons from below through its anti-coincidence design, while BTO does not (Gulick et al., 2024). The BTO science paper therefore identifies BTO as the only instrument on the spacecraft expected to downlink TGF event data. Using a TGF spectral model

$30~\mathrm{keV}$3

together with atmospheric propagation, the simulations show BTO sensitivity to the $30~\mathrm{keV}$4–$30~\mathrm{keV}$5 portion of the TGF spectrum and a visible $30~\mathrm{keV}$6 annihilation feature in an illustrative $30~\mathrm{keV}$7 burst injected at $30~\mathrm{keV}$8 altitude (Gulick et al., 2024).

5. Detector trade study and the centrality of afterglow

The most decisive hardware choice in the published BTO literature is the selection of NaI(Tl) over CsI(Tl) (Gulick et al., 27 Jan 2025). The comparison was driven not by nominal light yield alone but by heavy-ion-induced afterglow, which directly affects dead time, false triggers, and low-energy threshold. In irradiation tests with $30~\mathrm{keV}$9 helium and $5~\mathrm{MeV}$0 carbon beams, both materials showed afterglow, but the measured durations were substantially longer in CsI(Tl) (Gulick et al., 27 Jan 2025).

Material	Afterglow duration	Operational implication
CsI(Tl)	$5~\mathrm{MeV}$1 for C, $5~\mathrm{MeV}$2 for He	Stronger afterglow, larger dead time
NaI(Tl)	$5~\mathrm{MeV}$3 for C, $5~\mathrm{MeV}$4 for He	Faster decay, lower dead time

The duration ratios are explicit: CsI afterglow lasts $5~\mathrm{MeV}$5 longer than NaI for the carbon beam and $5~\mathrm{MeV}$6 longer for the helium beam (Gulick et al., 27 Jan 2025). The study further concludes that CsI would incur a $5~\mathrm{MeV}$7 larger dead time per afterglow event and require a $5~\mathrm{MeV}$8 higher energy threshold to suppress afterglow-induced false triggers (Gulick et al., 27 Jan 2025). Since BTO’s science case requires sensitivity down to $5~\mathrm{MeV}$9, this threshold penalty is mission-defining.

The orbital implications were quantified with simulated background rates for a BTO-like low Earth orbit. For the full two-detector system, the study finds a total background of $30~\mathrm{keV}$0 over $30~\mathrm{keV}$1–$30~\mathrm{keV}$2, or $30~\mathrm{keV}$3 per detector (Gulick et al., 27 Jan 2025). Using the irradiation results to define afterglow-inducing events, the authors estimate one such event every $30~\mathrm{keV}$4 in NaI and every $30~\mathrm{keV}$5 in CsI. The corresponding observing time lost per orbit is then about $30~\mathrm{keV}$6 for NaI and about $30~\mathrm{keV}$7 for CsI (Gulick et al., 27 Jan 2025). Although CsI offers higher light yield and radiation hardness, the same paper concludes that NaI is the better choice for BTO because it preserves low threshold, minimizes dead time, and simplifies trigger logic (Gulick et al., 27 Jan 2025).

This directly answers a common misconception in scintillator selection: higher raw light output does not automatically imply higher mission utility. In this case, afterglow behavior under the relevant heavy-ion environment dominates the system-level trade.

6. BTO as an observatory and algorithm design pattern

The generalized BTO pattern is clearest in GOTO. Its optical network maintains a fixed sky grid, builds reference images and baseline light curves, and aims to “survey the entire visible sky every two nights,” while the scheduler can interrupt the survey every $30~\mathrm{keV}$8 to prioritize new alerts (Dyer et al., 2022). The same system reports a minimum latency of $30~\mathrm{keV}$9 from alert receipt to exposure start and a prototype median coverage of $2~\mathrm{MeV}$0 per gravitational-wave alert during O3a (Dyer et al., 2022). In BTO language, the survey is the background layer and alert-driven tiling is the transient-observer layer.

In high-energy observatories, the pattern shifts from sky tiling to alert brokering and schedule control. The CTAO Transients Handler is responsible for filtering “thousands of internal and external alerts per night,” matching them to science configurations, computing observation windows, and creating scheduling blocks for the Short-Term Scheduler; the paper states that it is designed to determine the optimum observation strategy within four seconds of receipt (Collins et al., 28 Aug 2025). H.E.S.S. implements a similar but mature stack: about $2~\mathrm{MeV}$1 incoming alerts per month are reduced to about $2~\mathrm{MeV}$2 science-case-matched alerts, simple GRBs are processed in $2~\mathrm{MeV}$3–$2~\mathrm{MeV}$4, and end-to-end automatic reaction from alert reception to CT5 data taking is typically $2~\mathrm{MeV}$5 (Hoischen et al., 2022). In these systems, the background is no longer only detector noise or sky brightness; it is also the ongoing observation program that must be preserved except when transient priority warrants interruption.

Background-aware transient inference appears at the data-analysis level as well. For photon timing, the Bayesian Blocks study develops several background-subtraction variants and finds that the “weighted photons” method—combining source and background photons in a single time-ordered list with exposure-area weights—recovers eclipse ingresses and egresses with precision comparable to the interval between individual photons (Worpel et al., 2015). For fluorescence microscopy, Astro-BEATS constructs a dynamic background estimate, subtracts it frame by frame, normalizes the residual by local noise, and then uses DBSCAN in $2~\mathrm{MeV}$6 space to find coherent transients; it is presented explicitly as a background-and-transient observer for Ca$2~\mathrm{MeV}$7-imaging videos and outperforms threshold-based alternatives on its target task (Fan et al., 19 Mar 2026). These works show that BTO is not limited to telescope hardware; it also names a recurring computational architecture.

A further abstraction appears in work on observer motion and cosmic background monopoles. There the “background” is an isotropic radiation field in its rest frame, while the “observer” is a moving instrument whose Lorentz boost transfers monopole power into dipole and higher multipoles (Trombetti, 21 Mar 2025). The paper’s formalism separates intrinsic background properties from observer-induced distortions. This does not describe a mission payload, but it preserves the same conceptual division: a background with its own state, and an observing frame whose transient configuration must be modeled if the underlying signal is to be interpreted correctly.

Taken together, these usages show that BTO is best understood as a family resemblance across systems rather than a single platform definition. The most literal instance is the NaI(Tl)-based COSI payload. The most general formulation is a design principle: maintain a quantified background, react rapidly to departures from it, and preserve enough contextual information that the transient can be interpreted rather than merely detected.