AGN STORM: Multiwavelength AGN Mapping

• AGN STORM is a multi-wavelength reverberation mapping campaign that uses time delays from X-rays to infrared to spatially resolve AGN components such as the disk, BLR, winds, and hot dust.
• Its coordinated observations reveal scaling relations, anomalies in disk size, and obscuration effects that challenge conventional thin-disk models.
• The program links dynamic phenomena like the BLR holiday and variable obscuring winds to structural insights, refining our understanding of AGN energy output and inner geometry.

AGN STORM, short for Space Telescope and Optical Reverberation Mapping, is a series of intensive, multi-wavelength reverberation-mapping campaigns designed to use AGN variability as a spatial probe of the central engine. Its working principle is that rapid continuum variations from the accretion disk drive delayed responses in gas and dust at larger radii, so that measured lags between continuum, emission-line, and infrared signals can be converted into structural constraints through $R \sim c\,\tau$. Across its major campaigns on NGC 5548 and Mrk 817, AGN STORM has developed into a program for jointly mapping the accretion disk, the broad-line region (BLR), obscuring outflows, and the dusty torus with dense, coordinated coverage from X-rays through the near-infrared (Cackett et al., 2023).

1. Programmatic conception and reverberation framework

AGN STORM treats AGN variability as a tomographic signal rather than as stochastic noise. The campaigns are built around simultaneous or near-simultaneous monitoring from X-rays through UV, optical, and infrared bands, with the explicit goal of recovering the radial structure and coupling of the corona, disk, BLR, and dust. In this framework, continuum-continuum lags probe the thermal accretion flow, continuum-line lags probe photoionized gas, and continuum-dust lags probe the torus. The lag-wavelength relation for a centrally illuminated geometrically thin, optically thick disk is expected to scale approximately as $\tau(\lambda)\propto \lambda^{4/3}$, while departures from that behavior can signal additional reprocessing components or changes in the effective ionizing spectral energy distribution (Cackett et al., 2023).

The original campaign targeted NGC 5548 and combined HST UV spectroscopy with dense ground-based optical monitoring, later supplemented by near-infrared spectroscopy. AGN STORM 2 extended the same strategy to Mrk 817, but with a fully coordinated X-ray–UV–optical–near-infrared campaign from the outset. This broader architecture is central to the program’s logic: reverberation results are interpreted together with contemporaneous spectroscopy of absorption, emission, and continuum shape, rather than as lag measurements in isolation (Landt et al., 24 Oct 2025).

2. NGC 5548 and the first AGN STORM campaign

AGN STORM 1 on NGC 5548 established both the power and the complexity of high-cadence reverberation mapping. It found a UV/optical lag-wavelength relation roughly consistent with $\tau \propto \lambda^{4/3}$, but with disk sizes approximately three times larger than thin-disk predictions. It also revealed the “BLR holiday,” an interval during which strong broad emission lines no longer tracked the observed UV-optical continuum in the usual way. Those results made clear that continuum reverberation could not always be interpreted as a clean measure of a passive, centrally illuminated disk (Cackett et al., 2023).

The absorption-line counterpart of that holiday was analyzed by Dehghanian et al., who showed that variations in an obscurer’s line-of-sight covering factor can modify the soft X-ray continuum and thereby change the ionization of helium gas in the BLR. Ionizing radiation produced by recombining helium then affects the ionization of other species, providing a physical route to the anomalous absorption-line and emission-line behavior observed during the holiday (Dehghanian et al., 2018).

A later modeling synthesis classified disk winds into three regimes. Case 1 winds are transparent, fully ionized, and have minimal effects on the intrinsic SED; Case 2 winds have a He$^{++}$-He$^+$ ionization front and block part of the XUV continuum while transmitting much of the Lyman continuum; Case 3 winds have an H$^+$ ionization front and block much of the Lyman continuum. In that taxonomy, Case 2 winds lead to the observed abnormal behavior, while Case 3 winds provide a route to more extreme shielding states. This established obscuring disk winds as a structural component that must be incorporated into reverberation interpretations rather than treated as a secondary complication (Dehghanian et al., 2020).

3. AGN STORM 2 and the choice of Mrk 817

AGN STORM 2 targeted Mrk 817 because it is a relatively nearby, bright type-1 AGN with $z \approx 0.03145$, a well-measured black hole mass of approximately $3.9\times10^7\,M_\odot$, a barred spiral host with a compact bulge, and previous optical and near-infrared reverberation measurements, including a K-band dust lag of $92\pm 9$ days in an unobscured state. It also lies at the low-luminosity end of existing dust-reverberation samples, which is important for testing radius-luminosity relations (Landt et al., 24 Oct 2025).

The campaign assembled daily or near-daily Swift monitoring, dense HST/COS UV spectroscopy, NICER, XMM-Newton, NuSTAR, extensive ground-based photometry and spectroscopy, and near-infrared follow-up. Early observations showed that Mrk 817 had entered a heavily obscured X-ray state, despite having been selected in part for its historically unobscured nature. NICER time-resolved spectroscopy attributed much of the X-ray variability to changes in the column density of a dust-free, ionized obscurer by at least one order of magnitude, from $$2.85^{+0.48}_{-0.33} \times 10^{22}\,\mathrm{cm^{-2}}$$ to $\tau(\lambda)\propto \lambda^{4/3}$0, together with changes in the intrinsic continuum brightness. Correlation between the X-ray obscurer’s column density and the strength of UV broad absorption lines suggested that the X-ray and UV continua are both affected by the same obscuration, consistent with a clumpy disk wind launched from the inner BLR (Partington et al., 2023).

High-resolution X-ray spectroscopy then showed that the obscurer is a multi-phase ionized wind with an outflow velocity of approximately $\tau(\lambda)\propto \lambda^{4/3}$1 km s$\tau(\lambda)\propto \lambda^{4/3}$2. Variability in the shape of the absorption lines on timescales of hours placed the variable component at roughly $\tau(\lambda)\propto \lambda^{4/3}$3 if attributed to transverse motion along the line of sight, aligning with independent UV measurements that place the obscurer at the inner BLR. The inferred travel time of roughly 200 days from the disk to the line of sight matched the timescale of the outflow’s column-density variations through the campaign. This made Mrk 817 a direct laboratory for studying the dynamical connection between disk winds and reverberation phenomenology (Zaidouni et al., 2024).

4. Disk reverberation, BLR response, and the origin of continuum delays

The Swift campaign established that the UV/optical light curves of Mrk 817 are well correlated with one another and that their lags increase with wavelength roughly following $\tau(\lambda)\propto \lambda^{4/3}$4, as expected for a geometrically thin, optically thick, centrally illuminated disk. At the same time, the observed X-rays were largely uncorrelated with the UV/optical continuum, and the light curves were not simple shifted and scaled copies of one another over the full campaign. In particular, an interval of suppressed continuum response coincided with high UV line and X-ray absorption and with reduced emission-line variability amplitudes, suggesting a significant contribution to the continuum from BLR gas that sees an absorbed ionizing continuum (Cackett et al., 2023).

Frequency-resolved analysis sharpened that picture. Lewin et al. reported the first simultaneous measurement of X-ray reverberation lags and UVOIR disk reprocessing lags in the same AGN. The XMM-Newton data showed a soft X-ray lag of $\tau(\lambda)\propto \lambda^{4/3}$5 s, consistent with reverberation from the innermost accretion flow, while the UVOIR lags at low temporal frequencies were longer than standard disk predictions by factors of $\tau(\lambda)\propto \lambda^{4/3}$6–$\tau(\lambda)\propto \lambda^{4/3}$7. When variability was restricted to timescales shorter than the contemporaneous H$\tau(\lambda)\propto \lambda^{4/3}$8 lag, the UVOIR lags became consistent with thin-disk expectations. Modeling required an additional reprocessor with a median radius of $\tau(\lambda)\propto \lambda^{4/3}$9 days, consistent with the BLR size scale inferred from H$\tau \propto \lambda^{4/3}$0, and the BLR fraction increased when the obscurer column density was high (Lewin et al., 2024).

Netzer et al. extended this to a full continuum-origin model in which diffuse continuum emission, with additional contributions from strong and broad emission lines, explains the continuum lags observed in Mrk 817 during both high and low luminosity phases. Their BLR models assume radiation pressure-confined clouds distributed over $\tau \propto \lambda^{4/3}$1–$\tau \propto \lambda^{4/3}$2 light days, and they conclude that disk illumination by the variable X-ray corona contributes only a small fraction of the observed continuum lags. They also identify the signatures of lower column density winds and show that torus dust emission affects the observed lags in the $\tau \propto \lambda^{4/3}$3 and $\tau \propto \lambda^{4/3}$4 bands (Netzer et al., 2024).

A complementary inversion of the UV/optical light curves into $\tau \propto \lambda^{4/3}$5 maps found that the temperature fluctuations are dominated by coherent radial structures that move slowly inwards and outwards, with $\tau \propto \lambda^{4/3}$6, implying that much of the continuum variability is intrinsic to the disk rather than purely reverberative. Detrending isolates the shorter-timescale reverberation component, and simulated BLR contamination indicates that only approximately $\tau \propto \lambda^{4/3}$7 of the variable flux in the $\tau \propto \lambda^{4/3}$8 and $\tau \propto \lambda^{4/3}$9 light curves can be due to BLR contamination. This suggests that AGN STORM 2 does not reduce to a single lag spectrum: it resolves a superposition of prompt disk reprocessing, slow intrinsic disk fluctuations, and extended BLR or wind reprocessing (Neustadt et al., 2023).

5. Spectroscopic reverberation mapping of the hot dust in Mrk 817

The eleventh AGN STORM 2 paper, by Landt et al., shifted the program outward to the innermost hot dust. It used 157 cross-dispersed near-infrared spectra obtained over 608 days with ARC/TripleSpec, Gemini/GNIRS, and IRTF/SpeX, all in modes that provide simultaneous $^{++}$0–$^{++}$1–$^{++}$2–$^{++}$3 coverage. Flux calibration was anchored to the narrow [S III] $^{++}$4 line, yielding near-infrared continuum light curves with typical absolute flux uncertainties of approximately $^{++}$5, adequate for dust reverberation work (Landt et al., 24 Oct 2025).

The central methodological advance was spectroscopic rather than photometric dust reverberation. At each epoch, the spectra were decomposed into a variable outer accretion-disk continuum, hot-dust blackbody emission, a constant extended dust component, and host-galaxy starlight. After subtraction of the disk component, the residual longward of $^{++}$6 was fit with a single blackbody. The hot dust temperatures clustered tightly around $^{++}$7–$^{++}$8 K, with dispersions of about $^{++}$9 K per instrument, and the blackbody description gave reduced $^+$0 over $^+$1–$^+$2 (Landt et al., 24 Oct 2025).

A key result was that dust-flux light curves did not show a clean reverberation signal, whereas the dust-temperature light curve did. The reverberation lags of the dust temperature were $^+$3 days relative to $^+$4, $^+$5 days relative to $^+$6, and $^+$7 days relative to UVW2. Interpreting these lags as light-travel times gave a characteristic dust radius of approximately $^+$8 light-days, consistent with the previous photometric K-band lag of $^+$9 days measured when Mrk 817 was in an unobscured state. The temperature and radius therefore remained effectively unchanged between obscured and unobscured epochs (Landt et al., 24 Oct 2025).

Landt et al. interpreted that stability as evidence that the inner dusty structure behaves not as an instantaneous sublimation boundary but as a luminosity-invariant “dusty wall” of carbonaceous composition. Under the assumption of thermal equilibrium for large grains that are optically thick to the incident radiation, they used

$^+$0

with $^+$1 light-days and $^+$2 K to infer $^+$3. They considered two disk-heating scenarios, corresponding roughly to $^+$4 for an obscured SED and $^+$5 for an unobscured lower-luminosity disk, and preferred the latter because it allows the UV/optical continuum lags in the high-obscuration state to be dominated by diffuse BLR emission (Landt et al., 24 Oct 2025).

The same analysis also revealed an extended hot-dust component on scales $^+$6–$^+$7 pc, inferred from slit-dependent excess $^+$8- and $^+$9-band flux. Gemini/NIFS [S III] $z \approx 0.03145$0 imaging showed a spatially resolved rotating disk of ionized gas on comparable scales, and the extended hot dust was interpreted as residing in the compact bulge of the barred spiral host galaxy, probably heated by a nuclear starburst rather than by the AGN itself. This explicitly separated nuclear torus dust from host-galaxy hot dust within the AGN STORM framework (Landt et al., 24 Oct 2025).

6. Broader significance and interpretive consequences

Across AGN STORM and AGN STORM 2, a consistent structural picture has emerged in which disk reverberation, BLR diffuse continuum, obscuring winds, and dust reverberation are entangled rather than separable by default. In Mrk 817, continuum lags can follow a $z \approx 0.03145$1 form while still exceeding thin-disk predictions by factors of approximately $z \approx 0.03145$2–$z \approx 0.03145$3, and the lag normalization can vary by as much as a factor of $z \approx 0.03145$4 between epochs. The ground-based optical campaign found ICCF centroid lags from $z \approx 0.03145$5 days in $z \approx 0.03145$6 to $z \approx 0.03145$7 days in $z \approx 0.03145$8, with shorter lags from JAVELIN and PyROA, and interpreted the epoch dependence as evidence that changes in ionizing luminosity can alter the diffuse continuum luminosity emitted by the BLR and/or obscuring outflow (Montano et al., 4 May 2026).

A plausible implication is that the long-standing “disk size discrepancy” identified in AGN STORM 1 and related continuum campaigns is not a single-parameter problem. In NGC 5548, non-blackbody scattering atmospheres were shown to enlarge UV/optical-emitting radii relative to a blackbody disk, but the same models push the SED peak to shorter wavelengths than observed, implying that scattering atmospheres can contribute to the explanation for large inferred AGN accretion-disk sizes but are unlikely to be the only contributor (Hall et al., 2017).

The latest hot-dust results extend that same caution to the torus. Spectroscopic dust reverberation in Mrk 817 shows that dust temperature is a more robust reverberation observable than dust flux when extended hot dust and calibration systematics are present, and it supports a carbon-dominated, non-sublimating inner dust boundary that remains stable across very different obscuration states. This suggests that AGN STORM has progressively shifted from a program that measures single characteristic radii to one that resolves coupled transfer functions across the disk, BLR, wind, torus, and host galaxy (Landt et al., 24 Oct 2025).