arxiv: 2604.22721 · v1 · submitted 2026-04-24 · ⚛️ physics.ao-ph · cs.NA· math.NA· physics.data-an

Recognition: unknown

Spectral-Domain Local Statistics with Missing-Data Support for Cartesian and Polar Grids

Jairo M. Valdivia-Prado , William E. Chapman , Katja Friedrich

Authors on Pith no claims yet

Pith reviewed 2026-05-08 08:49 UTC · model grok-4.3

classification ⚛️ physics.ao-ph cs.NAmath.NAphysics.data-an

keywords local statisticsmissing datanormalized convolutionspectral operatorsCartesian gridspolar gridsboundary conditionsradar data

0 comments

The pith

A spectral framework computes local mean, variance, and standard deviation on incomplete Cartesian and polar grids by combining normalized convolution with boundary-aware transforms.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes a method to derive local statistics from gridded data that contain missing values. It pairs normalized convolution with reflective Discrete Cosine Transforms on Cartesian axes and periodic Real Fast Fourier Transforms on polar azimuth to respect actual data boundaries. This matters for applications such as radar or atmospheric fields where gaps are common and standard Fourier methods create wrap-around errors. Stability rules keep results bounded when sample support is low. The result is statistics that reflect the true local data density rather than assuming complete periodic coverage.

Core claim

The central claim is that normalized convolution performed with reflective DCT boundary handling on non-periodic Cartesian directions and periodic RFFT handling on circular azimuth directions, together with explicit denominator floors, prefill fallbacks, and variance clamps, produces bounded, support-aware local mean, variance, standard deviation, and effective sample count on incomplete grids.

What carries the argument

Normalized convolution realized through boundary-aware spectral operators: reflective DCT for Cartesian non-periodicity and periodic RFFT for polar circularity, augmented by denominator-floor, prefill-fallback, and variance-clamp safeguards.

If this is right

Cartesian boundary checks show elimination of wrap-around artifacts that appear with standard periodic assumptions.
Synthetic 3D outlier tests yield support-aware variance that correctly flags anomalies without false positives near data gaps.
Real-radar polar applications produce bounded local statistics suitable for interpretation even in sparsely sampled azimuth sectors.
An open-source implementation supplies a concise, reproducible path for applying the operators to new datasets.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same boundary-aware spectral construction could be tested on other incomplete gridded fields such as satellite imagery or climate reanalyses to check transferability.
Extending the safeguards to time-varying data would allow direct application to sequential radar volumes without re-deriving the operators.
Comparing the method against convolution kernels of varying support sizes on the same incomplete grid would quantify how much the spectral route reduces edge bias.

Load-bearing premise

The chosen reflective and periodic boundary conditions plus the stability safeguards produce accurate statistics without introducing new artifacts on real incomplete data.

What would settle it

Running the operators on a controlled synthetic Cartesian or polar grid whose true local mean and variance are known in advance and then checking whether the recovered values stay bounded and match the ground truth in under-supported regions.

Figures

Figures reproduced from arXiv: 2604.22721 by Jairo M. Valdivia-Prado, Katja Friedrich, William E. Chapman.

**Figure 1.** Figure 1: Spectral normalized-convolution pipeline illustrated on a 2-D synthetic field with missing values (gray). The view at source ↗

**Figure 2.** Figure 2: Boundary-condition sanity check on a 1D synthetic signal ( view at source ↗

**Figure 3.** Figure 3: Synthetic 3D cyclone-wind outlier-identification example using sparse quiver rendering. (a) Observed vector view at source ↗

**Figure 4.** Figure 4: Threshold sweep for 3D cyclone-wind outlier identification using view at source ↗

**Figure 5.** Figure 5: Real-radar polar demonstration on uncorrected DOW6 X-band reflectivity from WINTRE-MIX (2022-02-23), view at source ↗

read the original abstract

This paper presents a method for computing local mean, variance, standard deviation, and effective sample count on incomplete gridded data using boundary-aware spectral operators. The framework combines normalized convolution with explicit boundary-condition modeling: reflective Discrete Cosine Transform (DCT) for non-periodic Cartesian axes and periodic Real Fast Fourier Transform (RFFT) for circular azimuth processing in polar geometry. Stability safeguards (denominator floor, prefill fallback, and variance clamp) are specified for under-supported regions. We evaluate the framework across three targeted scenarios: a Cartesian boundary-condition check demonstrating the mitigation of wrap-around artifacts, a synthetic 3D outlier-identification test, and a real-radar polar application. Results establish bounded, support-aware interpretation of local statistics while preserving a concise reproducibility path through the open-source 'dct\_toolkit' implementation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper gives a practical way to do local mean/variance stats on incomplete Cartesian and polar grids via spectral normalized convolution plus boundary operators and safeguards, but the safeguards' ability to avoid new artifacts needs checking against the actual results.

read the letter

The main thing here is a numerical recipe for local statistics on grids with gaps. It uses normalized convolution in the spectral domain, reflective DCT to avoid wrap-around on Cartesian axes, and periodic RFFT for the circular azimuth in polar data, with three simple stability rules for low-support spots. The work targets atmospheric and radar applications where missing data is routine. What is new is the exact pairing of those transforms with the convolution for both grid types plus the listed safeguards; prior normalized convolution work exists, but this specific integration and the polar handling appear as a direct extension. The paper does well by keeping the implementation concise, releasing the dct_toolkit code, and running three focused tests: a Cartesian boundary check, a synthetic 3D outlier case, and a real polar radar example. That gives a clear reproducibility path. The soft spot is the safeguards themselves. The denominator floor, prefill fallback, and variance clamp are meant to keep things bounded, yet it is not obvious they commute with the spectral operators or fully cancel leakage from the prefill when the mask is patchy. The abstract claims the results show support-aware stats without new artifacts, so the full evaluations should contain the error metrics and comparisons; if those numbers hold up under scrutiny the concern is minor, but if they only show bounded outputs the bias issue could remain. This is a tool paper for people who already work with incomplete gridded radar or atmospheric data and want an efficient spectral option rather than spatial-domain loops. A reader in that subfield gets immediate value from the code and the targeted tests. It deserves peer review because the problem is real, the method is reproducible, and the evaluations cover the intended use cases; revisions would likely focus on more explicit checks of the safeguard behavior.

Referee Report

2 major / 1 minor

Summary. The paper presents a spectral-domain framework for computing local mean, variance, standard deviation, and effective sample count on incomplete gridded data. It combines normalized convolution with reflective Discrete Cosine Transform (DCT) boundary conditions for non-periodic Cartesian axes and periodic Real Fast Fourier Transform (RFFT) for circular azimuths in polar geometry, plus three explicit stability safeguards (denominator floor, prefill fallback, variance clamp) for under-supported regions. The method is evaluated in three scenarios: a Cartesian boundary-condition check, a synthetic 3D outlier-identification test, and a real-radar polar application, with results claimed to establish bounded, support-aware statistics and supported by an open-source 'dct_toolkit' implementation.

Significance. If the central claims hold, the approach offers an efficient, reproducible alternative to spatial-domain methods for handling missing data in geophysical grids (e.g., radar, satellite), with explicit boundary modeling that avoids wrap-around artifacts. The open-source implementation and focus on polar geometry are strengths that could aid adoption in atmospheric and oceanic data processing.

major comments (2)

[Abstract] Abstract: the claim that the three safeguards 'establish bounded, support-aware interpretation' without new artifacts is not supported by any derivation showing that the denominator floor, prefill fallback, and variance clamp commute with the DCT/RFFT operators or preserve the exact normalized-convolution identity when the support mask is non-uniform; in regimes where support falls below the implicit window size, prefill can inject values whose spectral leakage is not corrected by the denominator term.
[Abstract] Abstract: the three evaluation scenarios are described but no quantitative error metrics, bias measurements, or comparisons to spatial-domain normalized convolution baselines are provided, preventing verification that the spectral implementation remains faithful to the available data support.

minor comments (1)

The reproducibility path via 'dct_toolkit' is noted but the manuscript would benefit from explicit pseudocode or a small worked example showing how the safeguards interact with the spectral operators.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful and constructive review. We address each major comment point by point below. Where the comments correctly identify gaps in the original manuscript, we have revised the text accordingly.

read point-by-point responses

Referee: [Abstract] Abstract: the claim that the three safeguards 'establish bounded, support-aware interpretation' without new artifacts is not supported by any derivation showing that the denominator floor, prefill fallback, and variance clamp commute with the DCT/RFFT operators or preserve the exact normalized-convolution identity when the support mask is non-uniform; in regimes where support falls below the implicit window size, prefill can inject values whose spectral leakage is not corrected by the denominator term.

Authors: We agree that the original abstract overstates the guarantees provided by the safeguards. The manuscript does not contain a formal derivation proving commutation with the DCT/RFFT operators or exact preservation of the normalized-convolution identity for arbitrary non-uniform masks. The safeguards are practical post-processing steps applied after the spectral computation to enforce non-negative variances and minimum effective sample counts; prefill is used only as a fallback when the denominator falls below a threshold. In low-support regimes this can indeed introduce uncorrected spectral leakage. In the revised manuscript we have replaced the abstract claim with a more precise statement and added a short subsection in the Methods section that discusses the limitations of the safeguards together with a small-scale numerical comparison between the spectral and direct spatial implementations on a test grid containing both well-supported and under-supported regions. revision: yes
Referee: [Abstract] Abstract: the three evaluation scenarios are described but no quantitative error metrics, bias measurements, or comparisons to spatial-domain normalized convolution baselines are provided, preventing verification that the spectral implementation remains faithful to the available data support.

Authors: The original evaluation section emphasized qualitative demonstration of boundary handling and practical utility rather than quantitative benchmarking. We accept that this leaves the fidelity claim insufficiently verified. In the revised manuscript we have added a new quantitative subsection that reports (i) root-mean-square difference and bias between the spectral local statistics and a reference spatial-domain normalized-convolution implementation on the synthetic 3D test case, and (ii) error statistics for the polar-radar example computed against a high-resolution reference obtained from denser sampling periods. These metrics are restricted to regions where the local support exceeds the implicit window size, consistent with the method’s design. revision: yes

Circularity Check

0 steps flagged

No circularity: framework uses standard transforms and explicit definitions

full rationale

The paper defines its method by combining normalized convolution with established boundary-aware spectral operators (reflective DCT for Cartesian axes, periodic RFFT for polar azimuth). These are standard, externally verifiable transforms rather than quantities defined in terms of the target statistics. Stability safeguards (denominator floor, prefill fallback, variance clamp) are presented as explicit, user-specified components of the implementation for under-supported regions, not as outputs derived from or fitted to the statistics themselves. No equations reduce predictions to inputs by construction, no self-citations form load-bearing uniqueness arguments, and no ansatz is smuggled via prior work. The derivation chain is self-contained against external benchmarks (standard DCT/RFFT libraries and normalized convolution literature) and does not exhibit any of the enumerated circular patterns.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

Relies on standard properties of DCT and RFFT transforms plus the definition of normalized convolution; introduces three explicit stability rules whose exact thresholds are not quantified in the abstract.

free parameters (2)

denominator floor
Minimum value to avoid division by zero in low-support regions; exact value not stated in abstract.
variance clamp
Upper bound on variance to prevent instability; exact threshold not provided.

axioms (2)

domain assumption Reflective DCT correctly models non-periodic boundaries without wrap-around artifacts.
Invoked for Cartesian axes in the boundary-condition modeling section of the abstract.
domain assumption Periodic RFFT correctly handles circular azimuth in polar grids.
Invoked for polar geometry processing.

pith-pipeline@v0.9.0 · 5454 in / 1419 out tokens · 46635 ms · 2026-05-08T08:49:42.595582+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

[1]

doi:10.1175/1520-0426(2002)019<0673:UOTDRS>2.0.CO;2. J. C. Hubbert, M. Dixon, S. M. Ellis, and G. Meymaris. Weather radar ground clutter. part i: Identifica- tion, modeling, and simulation.Journal of Atmospheric and Oceanic Technology, 26(7):1165–1180, 2009. doi:10.1175/2009JTECHA1159.1. Joan Bech, Bernat Codina, Jeroni Lorente, and David Bebbington. The ...

work page doi:10.1175/1520-0426(2002)019 2002
[2]

doi:10.1109/T-C.1974.223784. J. Makhoul. A fast cosine transform in one and two dimensions.IEEE Transactions on Acoustics, Speech, and Signal Processing, 28(1):27–34, 1980. doi:10.1109/TASSP.1980.1163351. Gilbert Strang. The discrete cosine transform.SIAM Review, 41(1):135–147, 1999. doi:10.1137/S0036144598336745. Michael K. Ng, Raymond H. Chan, and Wun-C...

work page doi:10.1109/t-c.1974.223784 1974
[3]

Frédéric Fabry.Radar Meteorology: Principles and Practice

doi:10.1109/PROC.1979.11511. Frédéric Fabry.Radar Meteorology: Principles and Practice. Cambridge University Press, 2015. doi:10.1017/CBO9781107707405. Jairo M. Valdivia, Will Chapman, and Katja Friedrich. Radar data smoothing using the discrete cosine transform: A fast spectral domain algorithm.EGUsphere [preprint], 2026. doi:10.5194/egusphere-2026-415. ...

work page doi:10.1109/proc.1979.11511 1979