Larson–Sekanina Rotational Gradient Filter

• The Larson–Sekanina rotational gradient filter is a nucleus-centered imaging technique that suppresses the smooth coma and highlights anisotropic jet structures.
• It operates by subtracting a rotated copy of the original image using a fixed angular offset of 31° and a small radial shift, refining comet morphology.
• The filter’s output, though qualitative, enables extraction of jet position angles that are crucial for inferring comet rotation periods and potential precession.

The Larson–Sekanina rotational gradient filter is a nucleus-centered image-enhancement method used in cometary morphology studies to suppress the quasi-radial or nearly azimuthally symmetric coma component and amplify azimuthal intensity gradients associated with jets, fans, and other anisotropic outflow features. In recent analyses of the post-perihelion morphology of 3I/ATLAS, it functions as the key bridge between imaging and rotational inference: Hubble Space Telescope images processed with the filter yield measurable jet position angles, and those position-angle time series are then modeled to infer a characteristic post-perihelion period near $7.1$–$7.2$ h (Scarmato et al., 19 Feb 2026).

1. Functional definition and observational role

In the 3I/ATLAS studies, the filter is defined operationally rather than as a methodological topic in itself. One paper states that Larson–Sekanina processing is “widely used in cometary morphology studies to suppress the quasi-radial coma component and amplify azimuthal intensity gradients associated with jets, fans, and other anisotropic outflow features,” while the other describes it as the image-enhancement step that turns largely coma-dominated HST views into morphology maps in which narrow, repeatable jet-like features can be identified and tracked in position angle (Scarmato et al., 15 Jan 2026).

This framing is important. The filter is not treated as a generic sharpening operator, nor as a photometric transform. Its purpose is specifically morphological: to remove much of the low-order smooth coma and make angular asymmetries around the nucleus stand out. In the cited work, those asymmetries are the observables of interest because jet position angles and their time variability can constrain rotation, precession, and the distribution of active areas (Scarmato et al., 19 Feb 2026).

A recurrent misconception is to regard the filter output as a direct physical map of intensity. The 3I/ATLAS studies do not use it in that way. They use the output qualitatively to reveal the jet visually and quantitatively only in the limited sense that the revealed ridge line is converted into a numerical position angle. The filter output is not used directly for flux analysis or physical inversion of the coma brightness field (Scarmato et al., 15 Jan 2026).

2. Mathematical form and parameterization

One of the papers provides an explicit rotational-subtraction form for the filter,

$$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$

thereby identifying the implementation as a subtraction variant rather than a ratio image or normalized difference (Scarmato et al., 15 Jan 2026). In this form, the image is compared with a rotated copy at fixed radius. Smooth coma components that vary slowly with $\theta$ tend to subtract away, whereas localized azimuthal features survive as enhanced positive and negative structures.

The same paper reports that the adopted rotation increment was $31^\circ$, and that the authors experimented with a narrow range of $\Delta\theta$ values before selecting the setting that produced the most stable morphology between frames within the same epoch while avoiding “spurious multiple-lobe artifacts” (Scarmato et al., 15 Jan 2026). The second paper likewise reports “typical settings of $\delta r=0.1$ and $\Delta\theta=31^\circ$” (Scarmato et al., 19 Feb 2026).

The parameterization is, however, not fully documented. The appendix of one paper states “with the rotation increment $\theta = 31^\circ$ and $\delta r = 0.1$,” but the displayed equation contains no radial-shift term, and the papers do not define the units of $7.2$0 (Scarmato et al., 15 Jan 2026). This leaves an important ambiguity: a radial offset is mentioned as part of the practical setup, but its implementation is not reproducible from the equation alone. The other paper describes the filter as involving a small radial offset $7.2$1 and an angular rotational offset $7.2$2, but also notes that no explicit formula is written down there (Scarmato et al., 19 Feb 2026).

A plausible synthesis is that the 3I/ATLAS analyses use the Larson–Sekanina filter as a rotational differential operator centered on the comet photocenter, with a documented angular offset of $7.2$3 and a reported but incompletely specified radial offset parameter $7.2$4. The papers support the subtraction form explicitly and the radial offset only incompletely.

3. Data preparation and processing workflow

The most explicit preprocessing description appears in the appendix of the rotation-period study. For each HST visit, “the calibrated frames were registered to a common centroid and (when multiple exposures were available) combined to improve the signal-to-noise ratio while preserving small-scale morphology.” The stated workflow is therefore: start from calibrated HST frames, register them to a common centroid, combine multiple exposures within a visit when available, apply the Larson–Sekanina rotational differential filter, inspect the enhanced image, identify the dominant jet ridge line, and measure the jet position angle relative to the photocenter (Scarmato et al., 15 Jan 2026).

The broader morphology study describes the same operational chain in less formal terms. HST images processed with the Larson–Sekanina rotational-gradient filter were used to measure the position angles of three main persistent jet-like features between 2025 November 30 and December 27, 2025, and those measurements were then used in a weighted Fourier period scan (Scarmato et al., 19 Feb 2026).

Several implementation details are explicitly absent from the papers. They do not specify the HST camera, detector, filter name, wavelength bandpass, pixel scale, exposure times, or native image quality. They do not describe PSF subtraction, deconvolution, azimuthal median profile subtraction, adaptive smoothing, masking of stars or background objects, edge handling, interpolation scheme, polar-coordinate remapping, or any dedicated artifact-rejection protocol. They also do not name a software package or reduction pipeline for the Larson–Sekanina processing (Scarmato et al., 15 Jan 2026).

These omissions are methodologically significant. The filter is used operationally and centrally, but the papers do not present it as a fully reproducible standalone method paper. A careful reading therefore distinguishes between what is explicitly stated—registration, optional stacking, rotational subtraction, angular tuning, and ridge-line measurement—and what remains inferential, such as the exact stacking algorithm or the numerical implementation of image rotation (Scarmato et al., 15 Jan 2026).

4. Jet morphology and position-angle extraction

The immediate product of the filter in the 3I/ATLAS work is an enhanced inner-coma morphology in which persistent jet-like structures can be recognized. One study reports “three main persistent jet-like features” in the coma, “separated equally from each other, by $7.2$5 in the sky,” plus an anti-tail. Their mean position angles and half-amplitudes are summarized as follows (Scarmato et al., 19 Feb 2026):

Feature	Mean PA	Half-amplitude
Jet 1	$7.2$6	$7.2$7
Jet 2	$7.2$8	$7.2$9
Jet 3	$$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$0	$$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$1

The dominant feature for rotational analysis is Jet 2, described as approximately anti-sunward and near the projected spin-axis direction, with an angular excursion “of order $$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$2” (Scarmato et al., 19 Feb 2026). The companion paper focuses on a persistent anti-sunward jet-like feature whose average position angle is $$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$3, compared with an anti-sunward direction of $$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$4 (Scarmato et al., 15 Jan 2026).

The position-angle convention is explicit: measurements are made east of north, with $$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$5 and $$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$6 (Scarmato et al., 15 Jan 2026). The measurement procedure is also described with useful specificity. The enhanced images were inspected to identify the dominant jet ridge line; that ridge was traced over a fixed radial range from the nucleus, chosen to avoid the innermost region where saturation or PSF residuals could corrupt the morphology and the outermost region where low signal-to-noise would degrade the ridge definition; a straight line was then fit to the ridge in the sky plane; and the resulting orientation was converted to a position angle relative to the photocenter (Scarmato et al., 15 Jan 2026).

At the same time, the papers leave some aspects unresolved. They do not state the actual radial range used, whether ridge identification was manual, semi-manual, or algorithmic, whether the line fit was ordinary least squares or a robust estimator, or whether each visit yielded one combined position angle or multiple values from individual frames or sub-combinations (Scarmato et al., 15 Jan 2026). The morphology study is even terser, simply stating that the authors “measured the PAs” of the jet-like structures without detailing the extraction method (Scarmato et al., 19 Feb 2026).

This combination of explicit geometry and incomplete extraction detail has direct consequences for interpretation. The position angles are clearly measurable and coherent across epochs, but the extraction retains a phenomenological element, especially when emission broadens from a narrow jet into a fan.

5. Use as a rotational diagnostic

The filter becomes scientifically decisive when the enhanced morphology is converted into a time series of jet position angles. In the jet-based rotation analysis, the phase for each trial period $$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$7 is defined by

$$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$8

and the position-angle dependence is modeled with a truncated Fourier series,

$$I_{\rm LS}(r,\theta) = I(r,\theta) - I(r,\theta+\Delta\theta),$$9

fit by weighted least squares while scanning trial periods and evaluating $\theta$0 for $\theta$1 and $\theta$2 (Scarmato et al., 15 Jan 2026). The morphology paper describes the same idea in equivalent terms, stating that the measured jet position angles were fit with a low-order harmonic model in a period scan (Scarmato et al., 19 Feb 2026).

The reported jet-based result is

$\theta$3

with the uncertainty deliberately taken as conservative and dominated by sparse sampling and morphology-dependent systematics (Scarmato et al., 15 Jan 2026). Independent coma-dominated photometry yields

$\theta$4

with semiamplitude $\theta$5 about $\theta$6 mag and scatter $\theta$7 about $\theta$8 mag in one paper’s description (Scarmato et al., 19 Feb 2026). The close agreement between the morphology-derived and photometry-derived periods supports a characteristic post-perihelion period of about $\theta$9 h (Scarmato et al., 19 Feb 2026).

The interpretive step advanced by both papers is that this characteristic period traces an attitude precession or nutation, that is, non-principal-axis rotation. The jet structure is said to precess about the rotation axis with a characteristic angular excursion of order about $31^\circ$0, and the rotation axis is aligned with the sunward direction to within about $31^\circ$1 (Scarmato et al., 19 Feb 2026). One paper explicitly associates the $31^\circ$2 h period with precession of the jet structure around the rotation axis by $31^\circ$3 (Scarmato et al., 15 Jan 2026).

In this usage, the Larson–Sekanina filter is not ancillary. Without enhancement, there would be no clean position-angle time series for the dominant jet, and without that time series the morphology-based period estimate would not exist in its reported form. The filter is therefore the mechanism by which a visual coma structure is converted into a quantitative rotational tracer.

6. Limitations, ambiguities, and interpretive cautions

The 3I/ATLAS papers make clear that the filter’s strengths are accompanied by nontrivial caveats. The most explicit is morphology dependence. By 2025 December 27 the morphology had become more fan-like; one paper states that including that date “degrades phase stationarity relative to the four-night subset, consistent with the appearance of fan-shaped jets and a changing jet-to-photometry transfer function,” and notes that such morphological changes can alter the amplitude, shape, and sometimes the apparent phase of the photometric signal (Scarmato et al., 19 Feb 2026).

For position-angle extraction, the same issue is more acute. When a narrow jet broadens into a fan, the notion of a single jet axis becomes less well defined. The morphology paper notes that apparent shorter periods in secondary jets can arise from “non-sinusoidal projection, multi-lobed emission, or morphology-dependent centroiding of the brightest streamline within a fan” (Scarmato et al., 19 Feb 2026). This directly implies that a Larson–Sekanina-enhanced ridge is not always a unique geometric object.

The adopted uncertainty budget reflects these concerns. One paper states, “we adopt $31^\circ$4 per measurement,” and the appendix explains that the uncertainties were estimated from the dispersion among repeated measurements, including variation of the radial range and Larson–Sekanina parameters within a conservative interval, together with the finite angular resolution of the enhanced jet feature (Scarmato et al., 15 Jan 2026). The other paper remarks that the reported $31^\circ$5 values may correspond to $31^\circ$6 in one interpretation, underscoring limited formal precision and likely morphology-dependent systematics (Scarmato et al., 19 Feb 2026).

Another common misunderstanding is to assume that the filter setup is fully specified once $31^\circ$7 is given. The papers themselves show otherwise. They do not quantify sensitivity to centering error, angular step, or radial range; they do not discuss interpolation artifacts or noise amplification in a systematic way; and they do not compare the method with azimuthal median subtraction, division by radial profiles, adaptive Laplace filtering, or unsharp masking (Scarmato et al., 15 Jan 2026). In this sense, the method is established and central in practice, but incompletely documented at the level needed for strict reproducibility from text alone.

A balanced reading is therefore straightforward. The Larson–Sekanina rotational gradient filter is highly effective for revealing persistent angular structure in a bright coma and for enabling jet position-angle measurements. At the same time, the 3I/ATLAS studies show that its scientific value depends on centering, parameter choice, ridge identification, and the evolving morphology of the coma. These papers treat the filter not as an end in itself but as a morphology-enhancement tool whose outputs acquire quantitative force only when their systematics are acknowledged explicitly (Scarmato et al., 15 Jan 2026).