Pith. sign in

REVIEW 2 major objections 1 minor 49 references

Regularized covariance estimation from partially observed interferometric data

T0 review · 2 major / 1 minor · reviewed 2026-06-26 · grok-4.3

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Covariance from partially observed 2D interferometric data is recovered by Laplacian-regularized matrix completion without stationarity assumptions.

desk verdict The paper defines a fragmented missingness regime for 2D functional data and estimates covariance via Laplacian-regularized matrix completion without stationarity assumptions. read the letter →

arxiv 2606.19065 v1 pith:III6GBUQ submitted 2026-06-17 stat.ME stat.AP

Teresa Bortolotti , Roberta Troilo , Francesco Casu , Simone Vantini , Alessandra Menafoglio This is my paper · ORCID
classification stat.MEstat.AP
keywords covarianceestimationmatrixcompletionLaplacianregularizationfunctionaldataanalysisinterferometricfragmentedregimespatialdependencenonparametric
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper treats ground displacement measurements from the Small BAseline Subset technique as partially observed functional data on a two-dimensional domain. It defines a fragmented regime where large portions of the curves are systematically missing across replicates. The authors formulate covariance estimation as a matrix completion problem with Laplacian regularization to recover spatial dependence. The resulting nonparametric estimator avoids any stationarity or isotropy requirements. Simulations across varied covariance structures show consistently low error, and the method applied to Phlegraean Fields data recovers interpretable dependence patterns useful for risk monitoring.

What carries the argument

Laplacian-regularized matrix completion applied to the covariance matrix of fragmented functional data on a 2D domain.

What would settle it

Apply the estimator to synthetic data generated from a known covariance that cannot be well approximated by Laplacian regularization and check whether the recovered matrix deviates substantially from the generating covariance in Frobenius norm.

Watch

Extended reading notes

Core claim

The paper establishes a nonparametric covariance estimator for functional data observed under the fragmented regime of partial observation. The estimator is obtained by solving a matrix completion problem whose objective includes a Laplacian regularization term; this term is shown to recover the true spatial dependence structure of the data-generating process on the two-dimensional domain.

Load-bearing premise

The true spatial dependence structure is recoverable by Laplacian regularization when the missingness follows the fragmented regime of systematic partial observation.

Editorial extensions

If this is right

  • The estimator produces low error for a wide range of covariance structures in simulation studies.
  • Application to real interferometric data yields spatial dependence patterns that align with geophysical expectations.
  • The approach supports downstream analysis of geophysical processes without requiring stationarity or isotropy.
  • The method extends covariance estimation tools to functional data with systematic large-scale missingness.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same matrix-completion framing could be tested on other remote-sensing modalities that exhibit block-wise missingness.
  • Performance comparisons against kriging or other spatial covariance estimators would clarify when the Laplacian term adds value.
  • If the fragmented regime accurately describes additional data sources, the estimator could be reused directly in those domains.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 1 minor

Summary. The paper introduces the fragmented regime of systematic partial observation for 2D functional data arising from InSAR measurements. It formulates covariance estimation as a matrix-completion problem regularized by the graph Laplacian on the spatial domain, claiming the resulting estimator is nonparametric and free of stationarity or isotropy assumptions. Simulations are reported to yield consistently low error across covariance structures, and the method is applied to Phlegraean Fields displacement data to recover spatial dependence patterns.

Significance. If the Laplacian regularizer recovers the true second-order structure without systematic bias under the fragmented regime, the approach would supply a practical, assumption-light tool for covariance estimation in geophysical remote-sensing applications where large contiguous blocks of data are missing.

major comments (2)
  1. [Simulation study] Simulation study: the claim of low error 'across a range of covariance structures' does not include deliberate stress tests against covariances whose eigenstructure is misaligned with the low-frequency modes of the domain Laplacian (e.g., rough or spatially localized dependence). Without such cases the nonparametric claim remains unverified and the bias concern raised by the regularizer is unaddressed.
  2. [Methods / estimator definition] Estimator definition: the regularization parameter appears as a free hyper-parameter whose selection procedure is not shown to preserve the nonparametric character of the estimator; any data-driven choice must be demonstrated not to re-introduce implicit parametric assumptions.
minor comments (1)
  1. [Introduction / setup] The precise mathematical definition of the 'fragmented regime' should be stated explicitly with a diagram or equation before the estimator is introduced.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their detailed and constructive report. We address each major comment below and outline the revisions we will make to the manuscript.

read point-by-point responses

  1. Referee: [Simulation study] Simulation study: the claim of low error 'across a range of covariance structures' does not include deliberate stress tests against covariances whose eigenstructure is misaligned with the low-frequency modes of the domain Laplacian (e.g., rough or spatially localized dependence). Without such cases the nonparametric claim remains unverified and the bias concern raised by the regularizer is unaddressed.

    Authors: We agree that the existing simulations, while spanning multiple covariance structures, do not contain explicit stress tests for eigenstructures deliberately misaligned with the low-frequency modes of the Laplacian (e.g., rough or spatially localized dependence). To verify the nonparametric claim and directly address potential bias introduced by the regularizer, we will add new simulation scenarios with such covariance structures in the revised manuscript. These additions will include quantitative error comparisons under the fragmented observation regime. revision: yes

  2. Referee: [Methods / estimator definition] Estimator definition: the regularization parameter appears as a free hyper-parameter whose selection procedure is not shown to preserve the nonparametric character of the estimator; any data-driven choice must be demonstrated not to re-introduce implicit parametric assumptions.

    Authors: The regularization parameter is treated as a hyper-parameter whose value is chosen via cross-validation on the observed entries. We acknowledge that the manuscript does not explicitly demonstrate that this data-driven procedure preserves the nonparametric character of the estimator. In the revision we will expand the methods section with a clearer description of the selection algorithm together with additional empirical checks (and, if feasible, a brief theoretical argument) showing that the procedure does not implicitly re-introduce stationarity or other parametric assumptions. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; estimator is a proposed modeling choice

full rationale

The paper proposes a covariance estimator by formulating the task as matrix completion with Laplacian regularization on the 2D domain under a newly defined fragmented missingness regime. This is presented as a nonparametric modeling decision free from stationarity or isotropy assumptions, with performance assessed via simulations across covariance structures and a real-data application. No equations, derivations, or self-citations are described that reduce the estimator to a fitted quantity by construction, invoke a uniqueness theorem from prior author work, or rename a known result. The central claim rests on the modeling choice and empirical validation rather than any self-referential reduction, making the derivation self-contained.

Assumptions & free parameters 1 free parameters · 1 assumptions · 1 invented entities

The central claim rests on treating the interferometric measurements as functional data on a 2D domain under a newly introduced missingness regime and on the suitability of Laplacian regularization for recovering spatial covariance without stationarity assumptions.

free parameters (1)
  • regularization parameter
    Laplacian regularization strength must be chosen; its selection procedure is not detailed in the abstract.
assumptions (1)
  • domain assumption Observations can be modeled as partially observed functional data with two-dimensional domain under the fragmented regime
    Core modeling choice stated in the abstract for handling systematic missingness.
invented entities (1)
  • fragmented regime
    purpose: To describe the pattern where parts of the curves are systematically missing across replicates
    New term introduced to characterize the observation regime.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Regularized covariance estimation from partially observed interferometric data." pith.science (2026). https://pith.science/paper/III6GBUQ

@misc{pith2026260619065,
  author       = {Pith},
  title        = {Pith review of: Regularized covariance estimation from partially observed interferometric data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/III6GBUQ}},
  note         = {Machine review of arXiv:2606.19065}
}
read the original abstract

The Small BAseline Subset technique provides remote measurements of ground displacement with high spatial resolution, making it a key tool for monitoring geophysical processes in hazard-prone areas. An effective analysis of this type of data requires reliable estimation of their second-order structure, which is difficult to achieve because the measurements are systematically missing over relatively large portions of the investigated areas. We tackle the problem from a functional data analysis perspective and treat the observations as partially observed functional data with two-dimensional domain. To properly characterize the data, we introduce the fragmented regime of partial observation, where parts of the curves are systematically missing across replicates. For this regime, we propose a novel method for covariance estimation, formulating the task as a matrix completion problem with Laplacian regularization. The estimator is nonparametric and free from stationarity or isotropy assumptions. Extensive simulations show that our method achieves consistently low estimation error across a range of covariance structures. Application to ground displacement data relative to the Phlegraean Fields demonstrates its ability to recover meaningful spatial dependence patterns, highlighting its potential for environmental risk assessment and monitoring.

Figures

Figures reproduced from arXiv: 2606.19065 by the authors.

Figure 1
Figure 1. Illustration of the regimes of partial observation: [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Covariance between three selected locations (black dots) and all other points in the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Example of covariance kernel estimation with state-of-the-art methods for partially ob [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Comparison of the performance in terms of RMSE [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Functional boxplots of the curves of RMSE generated per each value of the logarithm of [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Functional boxplots of the curves of RMSE generated for a grid of values of [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: Example of covariance kernel estimation with state-of-the-art methods for partially ob [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Temporal coherence values per pixel in the Phlegraean Fields (Italy). Pixels with coher [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]
Figure 9
Figure 9. Figure 9: Illustrative covariance estimation for three consecutive columns of the 101 [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]
Figure 10
Figure 10. Figure 10: Comparison of per-pixel covariance estimates. Each row is referred to a reference pixel [PITH_FULL_IMAGE:figures/full_fig_p019_10.png]
Figure 11
Figure 11. Figure 11: Incomplete and reconstructed ground displacement in images relative to an area of [PITH_FULL_IMAGE:figures/full_fig_p020_11.png]
Figure 12
Figure 12. Figure 12: Simulated scenarios generated by sampling from a Gaussian process with second-order [PITH_FULL_IMAGE:figures/full_fig_p021_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references

  1. [1]

    2005 , publisher=

    Functional Data Analysis , author=. 2005 , publisher=

    2005

  2. [2]

    2012 , publisher=

    Inference for functional data with applications , author=. 2012 , publisher=

    2012

  3. [3]

    2003 , publisher=

    Hierarchical modeling and analysis for spatial data , author=. 2003 , publisher=

    2003

  4. [4]

    2000 , publisher=

    Linear processes in function spaces: theory and applications , author=. 2000 , publisher=

    2000

  5. [5]

    2015 , publisher=

    Statistics for spatial data , author=. 2015 , publisher=

    2015

  6. [6]

    2012 , publisher=

    Geostatistics: modeling spatial uncertainty , author=. 2012 , publisher=

    2012

  7. [7]

    2006 , publisher=

    Numerical mathematics , author=. 2006 , publisher=

    2006

  8. [8]

    2007 , publisher=

    Finite difference methods for ordinary and partial differential equations: steady-state and time-dependent problems , author=. 2007 , publisher=

    2007

Show all 49 references
  1. [9]

    2004 , publisher=

    Convex optimization , author=. 2004 , publisher=

    2004

  2. [10]

    Horn and Charles R

    Roger A. Horn and Charles R. Johnson , title =. 2012 , publisher =

    2012

  3. [11]

    Computational Statistics & Data Analysis , pages=

    Estimating a Smooth Covariance for Functional Data , author=. Computational Statistics & Data Analysis , pages=. 2025 , publisher=

    2025

  4. [12]

    Journal of the American Statistical Association , volume=

    Estimating the covariance of fragmented and other related types of functional data , author=. Journal of the American Statistical Association , volume=. 2021 , publisher=

    2021

  5. [13]

    Biometrika , volume=

    Recovering covariance from functional fragments , author=. Biometrika , volume=. 2019 , publisher=

    2019

  6. [14]

    The Annals of Statistics , volume=

    Functional data analysis by matrix completion , author=. The Annals of Statistics , volume=

  7. [15]

    Panaretos, Victor M and Tavakoli, Shahin , journal=. Cram. 2013 , publisher=

    2013

  8. [16]

    and Liebl, D

    Kneip, A. and Liebl, D. On the optimal reconstruction of partially observed functional data. The Annals of Statistics. 2020

    2020

  9. [17]

    2007 , publisher=

    Introduction to State Space Time Series Analysis , author=. 2007 , publisher=

    2007

  10. [18]

    Annals of Statistics , volume=

    Evaluation of the Cooling Trend in the Ionosphere Using Functional Regression with Incomplete Curves , author=. Annals of Statistics , volume=. 2017 , publisher=

    2017

  11. [19]

    and M \"u ller, H.-G

    Yao, F. and M \"u ller, H.-G. and Wang, J.-L. Functional Data Analysis for Sparse Longitudinal Data. Journal of the American Statistical Association. 2005

    2005

  12. [20]

    Journal of the American Statistical Association , volume=

    On consistency and sparsity for principal components analysis in high dimensions , author=. Journal of the American Statistical Association , volume=. 2009 , publisher=

    2009

  13. [21]

    Journal of the Royal Statistical Society , volume=

    Components and completion of partially observed functional data , author=. Journal of the Royal Statistical Society , volume=. 2015 , publisher=

    2015

  14. [22]

    and Stefanucci, M

    Kraus, D. and Stefanucci, M. , title = ". Biometrika , volume =

  15. [23]

    and Sangalli, L

    Stefanucci, M. and Sangalli, L. and Brutti, P. , title = ". Statistica Neerlandica , volume =

  16. [24]

    Statistics & Probability Letters , volume=

    Ridge Reconstruction of Partially Observed Functional Data is Asymptotically Optimal , author=. Statistics & Probability Letters , volume=. 2020 , publisher=

    2020

  17. [25]

    RNA Biology , volume=

    Dual Laplacian Regularized Matrix Completion for MicroRNA-Disease Associations Prediction , author=. RNA Biology , volume=. 2019 , publisher=

    2019

  18. [26]

    Remote Sensing of Environment , volume=

    A Quantitative Assessment of the SBAS Algorithm Performance for Surface Deformation Retrieval from DInSAR Data , author=. Remote Sensing of Environment , volume=. 2006 , publisher=

    2006

  19. [27]

    Sankhya A , volume=

    Inference on Covariance Operators via Concentration Inequalities: k-sample Tests, Classification, and Clustering via Rademacher Complexities , author=. Sankhya A , volume=. 2019 , publisher=

    2019

  20. [28]

    Biometrika , volume=

    Distances and Inference for Covariance Operators , author=. Biometrika , volume=. 2014 , publisher=

    2014

  21. [29]

    2023 International Conference on Sampling Theory and Applications (SampTA) , year=

    First-Order Algorithms for Optimization over Graph Laplacians , author=. 2023 International Conference on Sampling Theory and Applications (SampTA) , year=

    2023

  22. [30]

    IEEE Transactions on Image Processing , volume=

    Graph Laplacian Regularization for Image Denoising: Analysis in the Continuous Domain , author=. IEEE Transactions on Image Processing , volume=. 2017 , publisher=

    2017

  23. [31]

    Journal of the American Statistical Association , volume=

    Weighted functional data analysis for the calibration of a ground motion model in Italy , author=. Journal of the American Statistical Association , volume=. 2024 , publisher=

    2024

  24. [32]

    IEEE Transactions on geoscience and remote sensing , volume=

    Permanent scatterers in SAR interferometry , author=. IEEE Transactions on geoscience and remote sensing , volume=. 2001 , publisher=

    2001

  25. [33]

    IEEE Transactions on geoscience and remote sensing , volume=

    A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms , author=. IEEE Transactions on geoscience and remote sensing , volume=. 2002 , publisher=

    2002

  26. [34]

    Geophysical Research Letters , volume=

    Satellite radar interferometry time series analysis of surface deformation for Los Angeles, California , author=. Geophysical Research Letters , volume=. 2004 , publisher=

    2004

  27. [35]

    Mathematical Geosciences , volume=

    On the use of interferometric synthetic aperture radar data for monitoring and forecasting natural hazards , author=. Mathematical Geosciences , volume=. 2021 , publisher=

    2021

  28. [36]

    Mathematical geology , volume=

    Production of conditional simulations via the LU triangular decomposition of the covariance matrix , author=. Mathematical geology , volume=. 1987 , publisher=

    1987

  29. [37]

    Water Resources Research , volume=

    A fast and exact method for multidimensional Gaussian stochastic simulations , author=. Water Resources Research , volume=. 1993 , publisher=

    1993

  30. [38]

    Computers & Geosciences , volume=

    Gstat: a program for geostatistical modelling, prediction and simulation , author=. Computers & Geosciences , volume=. 1998 , publisher=

    1998

  31. [39]

    Remote sensing , volume=

    A covariance-based approach to merging InSAR and GNSS displacement rate measurements , author=. Remote sensing , volume=. 2020 , publisher=

    2020

  32. [40]

    Geoenvironmental Disasters , volume=

    Investigating the spatial correlations in univariate random fields of peak ground velocity and peak ground displacement considering anisotropy , author=. Geoenvironmental Disasters , volume=. 2021 , publisher=

    2021

  33. [41]

    Journal of Geophysical Research: Solid Earth , volume=

    Mapping small elevation changes over large areas: Differential radar interferometry , author=. Journal of Geophysical Research: Solid Earth , volume=. 1989 , publisher=

    1989

  34. [42]

    , author=

    The Parallel SBAS Approach for Sentinel-1 Interferometric Wide Swath Deformation Time-Series Generation: Algorithm Description and Products Quality Assessment. , author=. IEEE Trans. Geosci. Remote. Sens. , volume=

  35. [43]

    International Journal of Remote Sensing , volume=

    Long-term ERS/ENVISAT deformation time-series generation at full spatial resolution via the extended SBAS technique , author=. International Journal of Remote Sensing , volume=. 2012 , publisher=

    2012

  36. [44]

    Biometrika , volume=

    Testing for serial correlation in least squares regression: I , author=. Biometrika , volume=. 1950 , publisher=

    1950

  37. [45]

    Remote Sensing of Environment , volume=

    Sentinel-1 imagery for wide-scale quantitative landslide vulnerability assessment of buildings , author=. Remote Sensing of Environment , volume=. 2026 , publisher=

    2026

  38. [46]

    Remote Sensing of Environment , volume=

    Joint exploitation of space-borne and ground-based multitemporal InSAR measurements for volcano monitoring: The Stromboli volcano case study , author=. Remote Sensing of Environment , volume=. 2021 , publisher=

    2021

  39. [47]

    Geophysical Research Letters , volume=

    Surface displacements associated with the L'Aquila 2009 Mw 6.3 earthquake (central Italy): New evidence from SBAS-DInSAR time series analysis , author=. Geophysical Research Letters , volume=. 2010 , publisher=

    2009

  40. [48]

    ISPRS Journal of Photogrammetry and Remote Sensing , volume=

    Nation-wide mapping and classification of ground deformation phenomena through the spatial clustering of P-SBAS InSAR measurements: Italy case study , author=. ISPRS Journal of Photogrammetry and Remote Sensing , volume=. 2022 , publisher=

    2022

  41. [49]

    Remote Sensing , volume=

    A global archive of coseismic DInSAR products obtained through unsupervised sentinel-1 data processing , author=. Remote Sensing , volume=. 2020 , publisher=

    2020

Pith tools

Reviewed June 26, 2026 · model on record in the stance chip above.