---
title: SPT-3G Galactic Plane Survey
url: https://www.emergentmind.com/topics/spt-3g-galactic-plane-survey
type: topic
---

# SPT-3G Galactic Plane Survey

The SPT-3G Galactic Plane Survey is a dedicated millimeter-wave time-domain survey of the Galactic Plane conducted with the South Pole Telescope using the SPT-3G camera. It is described as the first dedicated high-sensitivity, wide-field, time-domain, mm-wave survey of the Galactic Plane, targeting approximately \(100~\mathrm{deg}^2\) near the Galactic center with near-daily cadence and simultaneous observations at 95, 150, and 220 GHz. In its first two years, the survey produced blind detections of two day-scale millimeter transients associated with previously identified accreting white dwarfs, establishing the survey as a new probe of Galactic mm-wave variability and transient synchrotron phenomena [2509.08962].

## 1. Instrumentation, footprint, and observing program

The survey uses the South Pole Telescope, a 10 m, wide-field millimeter-wave survey telescope at the Amundsen–Scott South Pole Station, together with the SPT-3G receiver, a third-generation, polarization-sensitive camera with roughly 16,000 bolometric detectors observing simultaneously in three bands centered near 95, 150, and 220 GHz. The instantaneous field of view is approximately 1.9 degrees in diameter, and the camera provides arcminute-scale resolution with full Stokes \(I/Q/U\) capability. For the analysis reported, the effective band centers are \(\nu_{95}=94.2\) GHz, \(\nu_{150}=147.8\) GHz, and \(\nu_{220}=220.7\) GHz; localization and pointing calculations adopt beam widths of about \(50^{\prime\prime}\times49^{\prime\prime}\), representative of the 150 GHz band [2509.08962].

The survey footprint is divided into three subfields near the Galactic center: ra17h45dec-29, centered on Sgr A, ra17h37dec-32, and ra17h27dec-35. Each is approximately \(40~\mathrm{deg}^2\), with approximately uniform coverage across the combined footprint. The 2023–2024 campaign includes 1,692 individual 20-minute observations in total, of which 1,506 remain after quality cuts, corresponding to approximately 500 hours of on-sky time. The mapping strategy consists of raster scans at constant elevation, equivalent to constant declination at the South Pole, with back-and-forth motion at fixed speed; a typical observation contains approximately 24 scans of approximately 51 s each, stepping by \(7.5^\prime\) in elevation to tile the subfield. Each subfield was observed approximately 12 times per day during survey periods.

Single-epoch sensitivity is characterized through filtered difference maps. The median per-epoch noise rms is 4.6 mJy at 95 GHz, 6.6 mJy at 150 GHz, and 24.6 mJy at 220 GHz. In the two-year coadded maps away from the Galactic Plane, the depths are 6.8, 8.3, and \(20~\mu\mathrm{K}\)-arcmin, corresponding to point-source sensitivities of approximately 1.0, 1.6, and 4.0 mJy at 95, 150, and 220 GHz, respectively. These values indicate that the transient analysis is driven primarily by the lower-noise 95 and 150 GHz channels, while 220 GHz is more useful for post-detection spectral characterization.

## 2. Mapmaking, calibration, and transient-selection procedure

The transient pipeline begins with time-ordered data that are high- and low-pass filtered; common-mode and polynomial filters are applied to remove atmospheric fluctuations, and a survey mask is used during filtering to suppress ringing from very bright Galactic structures. Maps are made in the Lambert equal-area projection with \(0.25^{\prime\prime}\) pixels. To remove static sky components, a yearly average map for 2023 or 2024 is subtracted from each per-epoch map, leaving transient and variable emission plus residuals near the brightest sources [2509.08962].

Calibration and astrometry are controlled at the per-observation level. Relative gain calibration is performed by forcing a set of bright, static sources in each subfield to match their average flux densities. The standard deviations of the gain variations are 1.4% at 95 GHz, 3.8% at 150 GHz, and 3.2% at 220 GHz, with residual calibration uncertainties of 0.5%, 1%, and 1%, respectively. Pointing is corrected in two steps: relative alignment to bright SPT sources, followed by an absolute tie using AT20G 20 GHz sources. The typical residual pointing uncertainty per transient is approximately \(6^{\prime\prime}\), with absolute systematic uncertainties \(\sigma_{x\mathrm{decl},\mathrm{sys},\mathrm{abs}}=3.05^{\prime\prime}\) and \(\sigma_{\mathrm{decl},\mathrm{sys},\mathrm{abs}}=4.42^{\prime\prime}\).

Transient identification is carried out on filtered difference maps using a point-source matched filter following the SPT-3G transient pipeline described in Guns et al. 2021. The local noise is defined as the rms in an annulus centered on each pixel with inner radius \(2^\prime\) and outer radius \(3^\prime\), a choice motivated by anisotropic noise near the Galactic Plane and residuals around bright sources. The detection threshold is intentionally conservative: at least one pixel must satisfy \(\mathrm{S/N}\geq5\) in both 95 and 150 GHz in the same epoch. A nominal \(TS>36\) threshold is also imposed, although it is not limiting for the reported events. The 220 GHz band is not required for detection because its single-epoch noise is approximately five times higher than that of the 95 and 150 GHz bands.

After candidate selection, the pipeline extracts multi-epoch lightcurves from maps of the same year and subfield and fits a multi-band Gaussian flare model jointly to the 95 and 150 GHz single-observation measurements, using a common peak time and width but independent amplitudes per band. Visual vetting removes moving thermal sources such as weather balloons and residuals from bright Galactic structures. Ten exceptionally bright sources, including Sgr A, are masked with \(5^\prime\) radius in the transient pipeline to avoid leakage-dominated false positives. A full trials-factor false-alarm calculation is not attempted because of the spatially varying Galactic background; robustness instead relies on the dual-band requirement, \(TS\gg36\), consistency across scans, and qualitative vetting. This establishes a high-purity sample at the expense of completeness for faint events and for events much shorter than a 20-minute observation or much longer than a few days.

## 3. First transient detections and source associations

The first two years of the survey yielded two transient events exceeding the dual-band \(5\sigma\) threshold. Both events are unpolarized, have durations of approximately one day, and reach peak flux densities at 150 GHz of at least 50 mJy. Both are spatially associated with previously identified accreting white dwarfs, rather than with unclassified Galactic transients [2509.08962].

| Parameter | SPT-SV J174417.2−293942 (Source 1) | SPT-SV J173508.3−292956 (Source 2) |
|---|---|---|
| Coordinates | RA \(=266.072^\circ\), Dec \(=-29.662^\circ\) | RA \(=263.785^\circ\), Dec \(=-29.499^\circ\) |
| Positional uncertainty | \(\sigma_{\mathrm{pos}}=6.1^{\prime\prime}\) | \(\sigma_{\mathrm{pos}}=6.4^{\prime\prime}\) |
| Association | Chandra 2CXO J174417.2−293944; Swift Bulge Survey source argued to be a cataclysmic variable with a subgiant donor | Chandra 2CXO J173508.2−292957; a symbiotic star with a K4III giant donor |
| Gaia distance | \(966\pm18\) pc | \(1497\pm44\) pc |
| Peak epoch | MJD 60023.42 | MJD 60384.71 |
| Peak flux densities | \(S_{95}=80.3\pm4.7\) mJy; \(S_{150}=59.4\pm6.6\) mJy; \(S_{220}=49.3\pm25.4\) mJy | \(S_{95}=58.7\pm4.5\) mJy; \(S_{150}=86.6\pm6.2\) mJy; \(S_{220}=104.4\pm23.1\) mJy |
| Duration | \(\Delta t=1.00\pm0.05\) days | \(\Delta t=0.80\pm0.08\) days |
| Spectral indices | \(\alpha_{95-150}=-0.31\pm0.08\); \(\alpha_{150-220}=-0.72\pm0.45\) | \(\alpha_{95-150}=+0.96\pm0.08\); \(\alpha_{150-220}=+0.06\pm0.27\) |
| Polarization limits | \(<10\%\) (95), \(<10\%\) (150), \(<41\%\) (220) | \(<9\%\) (95), \(<8\%\) (150), \(<33\%\) (220) |
| Test statistic | \(TS=2227\) | \(TS=1887\) |

Source 1 lies at Galactic coordinates \(l=359.229^\circ\), \(b=-0.134^\circ\) and is associated with Chandra 2CXO J174417.2−293944, identified in the Swift Bulge Survey as a cataclysmic variable with a subgiant donor. Source 2 lies at \(l=358.307^\circ\), \(b=+1.640^\circ\) and is associated with Chandra 2CXO J173508.2−292957, a symbiotic star with a K4III giant donor. In both cases, quiescent emission is not detected in the coadds. The 3\(\sigma\) upper limits from stacked maps are, for Source 1, \(<408\) mJy at 95 GHz, \(<581\) mJy at 150 GHz, and \(<2760\) mJy at 220 GHz, with the limits dominated by local background near the Plane; for Source 2, the corresponding upper limits are \(<9\) mJy, \(<18\) mJy, and \(<71\) mJy.

Polarization analysis uses Stokes \(I/Q/U\) maps filtered identically to the intensity maps. The polarization fractions are measured by stacking all observations from the day of each flare and computing \(\sqrt{\langle Q^2+U^2\rangle/\langle I^2\rangle}\) at the source position. Both events are consistent with no linear polarization, with 95% upper limits below 10% at 95 and 150 GHz.

## 4. Derived energetics and source-physics constraints

The survey derives isotropic luminosities using
\[
L_\nu = 4\pi d^2 S_\nu,
\qquad
\nu L_\nu = 4\pi d^2 \nu S_\nu.
\]
At 150 GHz, Source 1 has \(\nu L_\nu(150)=(9.95\pm1.15)\times10^{30}~\mathrm{erg}~\mathrm{s}^{-1}\), while Source 2 has \(\nu L_\nu(150)=(3.48\pm0.32)\times10^{31}~\mathrm{erg}~\mathrm{s}^{-1}\). Peak \(\nu L_\nu\) values at 95 and 220 GHz are reported to be of the same order: \(8.54\times10^{30}\) and \(1.21\times10^{31}~\mathrm{erg}~\mathrm{s}^{-1}\) for Source 1, and \(1.50\times10^{31}\) and \(6.15\times10^{31}~\mathrm{erg}~\mathrm{s}^{-1}\) for Source 2 [2509.08962].

A band-limited flare-energy proxy is then estimated from
\[
E \approx \nu L_\nu \times \Delta t.
\]
Using the 150 GHz luminosity and fitted duration, Source 1 yields \(E\approx(8.6\pm1.0)\times10^{35}\) erg, and Source 2 yields \(E\approx(2.41\pm0.36)\times10^{36}\) erg. The paper explicitly notes that this is a lower bound to the bolometric synchrotron flare energy and depends on the assumption that the spectrum near 150 GHz dominates the mm-band power during the event.

Brightness temperature estimates are based on
\[
T_b \approx \frac{c^2 S_\nu}{2k_B \nu^2 \Omega},
\qquad
\Omega \approx \frac{\pi \theta^2}{4\ln 2},
\]
for a Gaussian source of FWHM \(\theta\). Two illustrative angular scales are considered: \(\theta\approx1.5\) mas, corresponding to a “micronova/shock shell” scale at \(d\approx1\) kpc for approximately one day of expansion at several \(10^3~\mathrm{km\,s^{-1}}\), and \(\theta\approx0.15\) mas for a compact reconnection or jet knot. At 150 GHz, Source 1 gives \(T_b\approx1.4\times10^6\) K for \(\theta=1.5\) mas and \(1.4\times10^8\) K for \(\theta=0.15\) mas; Source 2 gives \(2.1\times10^6\) K and \(2.1\times10^8\) K for the same two scales. The reported interpretation is that brightness temperatures in the \(10^6\)–\(10^8\) K range require nonthermal synchrotron emission, whereas thermal free-free emission from nova ejecta would imply \(T_b\ll10^6\) K at mm wavelengths.

The synchrotron parameter estimates use
\[
\nu_c \approx \frac{3}{2}\gamma^2 \frac{eB}{2\pi m_e c}
\approx 4.2\times10^6 \gamma^2 (B/\mathrm{G})~\mathrm{Hz}
\]
and
\[
t_{\rm syn} \approx \frac{6\pi m_e c}{\sigma_T \gamma B^2}
\approx \frac{7.7\times10^8}{\gamma B^2}~\mathrm{s}.
\]
For mm-wave emission near \(1.5\times10^{11}\) Hz, the required Lorentz factors are modest: \(\gamma\approx190\) for \(B\approx1\) G, \(\gamma\approx60\) for \(B\approx10\) G, and \(\gamma\approx19\) for \(B\approx100\) G. Day-scale durations imply \(\gamma B^2\approx8900\)–11000; the examples given in the paper include \(B\approx10\) G with \(\gamma\approx90\)–110, and \(B\approx20\) G with \(\gamma\approx22\)–28. This is presented as consistent with mm-wave synchrotron from mildly relativistic electrons in magnetic fields of order 10–30 G, with the observed durations governed by opacity evolution and radiative cooling.

## 5. Emission mechanism, source class, and alternative explanations

The observational combination of spectral shape, duration, and brightness temperature is interpreted as favoring synchrotron emission. Source 1 has a falling spectrum between 95 and 150 GHz and a further falling spectrum between 150 and 220 GHz, which is described as consistent with optically thin synchrotron or a transition toward the optically thin regime. Source 2 has a rising spectrum from 95 to 150 GHz and an approximately flat spectrum from 150 to 220 GHz, which is described as suggesting partially self-absorbed synchrotron at 95–150 GHz. The lack of detected linear polarization, with \(p<10\%\) at 95 and 150 GHz, is stated to be consistent with self-absorbed emission and/or tangled magnetic fields in reconnection-driven plasmoids; it does not rule out synchrotron [2509.08962].

Both flares are associated with long-period accreting white dwarf systems, one a cataclysmic variable with a subgiant donor and the other a symbiotic system with a K giant donor. The paper argues statistically that it is unlikely, at approximately 1% combined probability, that both mm flares would arise from donor stars alone rather than from systems containing accreting white dwarfs. This favors an accretion-flow origin. The proposed mechanism is magnetic reconnection in the accretion disk or magnetically threaded inner flow, by analogy with reconnection flares in young stellar object disks and with optical superflares suggested in some cataclysmic variables.

The paper also contrasts these events with cm-band flares from cataclysmic variables and related systems, which often show circular polarization and are attributed either to donor-star flaring or to coherent emission processes. The mm events reported here show no detected linear polarization and have spectra consistent with synchrotron, suggesting different opacity regimes and possibly different emission zones. An alternative interpretation in terms of localized thermonuclear “micronova” events on the white dwarf surface is considered but regarded as less natural for the mm data: generating day-scale mm synchrotron would require shocks and rapid nonthermal particle acceleration, and the observed brightness temperature and spectral behavior are stated to fit reconnection-driven synchrotron in the accretion flow more naturally than thermal ejecta. This does not exclude other channels, but it sets the preferred interpretation within the framework of the reported observations.

## 6. Selection effects, survey limitations, and expansion of parameter space

The survey design deliberately prioritizes robustness over completeness. The 220 GHz band is too noisy for primary detection, and the dual-band \(5\sigma\) requirement suppresses false alarms from instrumental glitches, atmospheric artifacts, balloon contamination, and leakage near bright Galactic structures, but it also reduces sensitivity to fainter flares and to events detectable in only one band. Similarly, the use of 20-minute observations and Gaussian flare fitting favors durations from roughly hours to days and is less sensitive to events much shorter than an observation or much longer than a few days. The paper explicitly notes that Galactic-plane noise is anisotropic and that a full statistical trials-factor false-alarm analysis is deferred to future work [2509.08962].

Within those conservative cuts, the two detections already define a scientifically important region of parameter space. The observed rate—two events in approximately 500 hours over \(100~\mathrm{deg}^2\)—is reported to exceed naive expectations based on extragalactic-field rates. The detected flares have \(\nu L_\nu(150)\approx10^{31}~\mathrm{erg}~\mathrm{s}^{-1}\), near the high-luminosity end of mm flares found in SPT-3G extragalactic fields, while the distances of approximately 1–1.5 kpc show the advantage of Galactic Plane observations for sampling luminous sources over larger Galactic volumes. A plausible implication is that the Plane survey accesses a source population and luminosity function not well represented in higher-latitude mm surveys.

Planned enhancements include lowering the detection threshold from \(5\sigma\) to \(3\sigma\) once the noise is better characterized, allowing single-band detections with more robust artifact rejection, and expanding the time-domain analysis in both directions: faster searches on minute timescales using single-scan maps and slower searches on month-to-year timescales using coadds. Additional observing seasons are planned at roughly one month per year through 2028. The expected consequence, as stated in the paper, is substantial growth in sample size and broader sensitivity to luminous accretion-related flares and rare Galactic mm transients, including X-ray binaries, novae, and magnetars. In that sense, the SPT-3G Galactic Plane Survey functions both as a transient discovery program and as a survey of Galactic compact-object variability in a wavelength regime that has only recently become accessible to blind wide-field searches.

Source: https://www.emergentmind.com/topics/spt-3g-galactic-plane-survey