arxiv: 2605.04921 · v1 · submitted 2026-05-06 · 📊 stat.ME · stat.AP

Recognition: unknown

A Convolution Process for Sea Surface Temperature Hot-Spot Identification in the Mediterranean Sea

Alessandra Menafoglio, Leonardo Marchesin, Piercesare Secchi

Pith reviewed 2026-05-08 16:45 UTC · model grok-4.3

classification 📊 stat.ME stat.AP

keywords sea surface temperaturehot-spot identificationMediterranean Seaconvolution processdirected linear networkspatial covariancepenalized estimationclimate projections

0 comments

The pith

A directed linear network of ocean currents yields a convolution covariance for Mediterranean sea surface temperatures without land-crossing correlations.

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

This paper proposes a statistical framework for modeling sea surface temperature variations in the Mediterranean Sea. It discretizes the sea into a network of points connected by the direction of currents and defines a covariance structure through a convolution process that incorporates flow directions. The model is then used within Monte Carlo simulations to project future temperatures and locate hot spots where ecological risks may rise. A reader would care because traditional spatial models can imply impossible temperature links across land, leading to flawed forecasts.

Core claim

By representing the marine domain as a directed linear network that preserves ocean current orientations, the authors define a moving-average stochastic process whose covariance arises from a spatial kernel combined with flow-dependent weights derived from a Markovian transition matrix on the network. A penalized estimator is introduced to regularize the parameters while enforcing hydrodynamic consistency, allowing the model to be used for identifying thermal hot spots in RCP-based SST projections.

What carries the argument

The convolution-based moving-average process on the directed linear network using a Markovian transition-probability matrix to encode flow dynamics.

If this is right

Identifies SST hot spots of ecological risk in a way consistent with physical barriers.
Quantifies uncertainty in future SST forecasts under climate scenarios.
Prevents unphysical correlations such as those across land barriers.
Supports targeted environmental assessments through consistent covariance modeling.

Where Pith is reading between the lines

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

Similar network discretizations could improve modeling of other flow-constrained variables like salinity or nutrient transport.
The approach may be tested by checking if it better predicts observed SST patterns in areas with strong currents compared to standard geostatistical models.
Extending the network to account for time-varying currents could refine long-term projections.

Load-bearing premise

Discretizing the continuous sea domain into a directed linear network that follows ocean current directions sufficiently captures the relevant flow dynamics for covariance modeling.

What would settle it

If the model assigns correlations close to zero between SST locations separated by land while real data or alternative models show significant positive correlations, the framework's advantage would be questioned.

Figures

Figures reproduced from arXiv: 2605.04921 by Alessandra Menafoglio, Leonardo Marchesin, Piercesare Secchi.

**Figure 1.** Figure 1: The spatial domain of the northern Tyrrhenian Sea is depicted with the point estimates view at source ↗

**Figure 1.** Figure 1: In this work, velocity and temperature data are co-located on the same regular rectangular grid. We exploit this co-location in constructing the network representation. More generally, when observations are not co-located or are irregularly spaced, one could define a sufficiently fine grid such that the observations lie on it. Extension to such non-co-located settings will be explored in future work. The c… view at source ↗

**Figure 2.** Figure 2: Construction of the edges. – namely, proportional to the ratio between the Euclidean distance of the edge and the magnitude of the corresponding water velocity vector. Note that this change would not affect the following construction. We do not impose restrictive boundary conditions on the perimeter vertices; in particular, we allow water to both exit and enter the domain. This choice reflects the physical… view at source ↗

**Figure 3.** Figure 3: The linear network in the whole domain of analysis. In blue the directed edges composing view at source ↗

**Figure 5.** Figure 5: (b) view at source ↗

**Figure 7.** Figure 7: Spatial heatmaps for different covariance models. The color gradient indicates the covari view at source ↗

**Figure 8.** Figure 8: Empirical covariance functions for the period 2006–2022 and point-wise mean. The view at source ↗

**Figure 9.** Figure 9: Joint exceedance probabilities for neighborhoods view at source ↗

**Figure 10.** Figure 10: Hotspot estimates for RCP 4.5 – Year 2050. Top row: New framework. Bottom row: view at source ↗

read the original abstract

Sea surface temperature (SST) is a fundamental determinant of global climate dynamics and economic activity. Reliable projections of future SST patterns depend critically on a rigorous characterization of the underlying spatial random field. In this study, we introduce a novel convolution-based covariance framework tailored to geostatistical domains constrained by physical barriers and influenced by vector-driven flows. By discretizing the continuous marine domain into a directed linear network that preserves the orientation of ocean currents, we construct a moving-average stochastic process whose dynamic is encoded via a Markovian transition-probability matrix on the network's vertices. The induced covariance structure emerges as a weighted combination of a spatial kernel and flow-dependent weights, giving rise to a complex estimation problem. To stabilize inference, we propose a penalized estimator that regularizes covariance parameters while enforcing consistency with known hydrodynamic properties. We then embed this covariance model into a Monte Carlo simulation framework to refine RCP-based SST projections and to identify thermal 'hot spots' of heightened ecological risk. Our approach delivers a statistically principled framework that prevents physical inconsistencies -- such as correlations across land barriers -- providing a robust basis for quantifying uncertainty in future SST forecasts and for guiding targeted environmental assessments.

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

The paper tries to embed ocean flows and barriers into a convolution covariance for Mediterranean SST but the barrier-blocking claim does not follow from the described construction.

read the letter

The core move is to discretize the sea into a directed linear network that follows current directions, build a moving-average process via Markov transitions on that network, and form the covariance as a mix of a spatial kernel and flow weights. They add a penalized estimator to keep parameters consistent with known hydrodynamics and then run Monte Carlo projections under RCP scenarios to flag thermal hot spots. That network step is a concrete way to bring vector-driven transport into the random field, and the Mediterranean setting with its islands and gyres makes the application sensible rather than generic. The penalized estimator and the simulation pipeline are also standard tools applied in a targeted way. The main weakness is exactly the one the stress-test flags. The abstract says the induced covariance prevents correlations across land barriers, yet it is explicitly a weighted sum of a spatial kernel (presumably Euclidean) and the flow term. The kernel term will generally assign positive covariance to points on opposite sides of land, and the flow weights cannot remove it. Nothing in the abstract indicates they replace the kernel with a network-distance version or zero it out across barriers, so the central physical-consistency claim is not supported by the stated construction. Without seeing the actual derivations or any cross-validation against held-out SST data, it is also impossible to judge whether the penalty actually enforces the desired properties or just smooths the fit. This is the sort of paper a spatial-stats group working on constrained domains would want to read for the network idea, even if the barrier guarantee needs fixing. It is worth sending to referees because the application is real and the modeling direction is worth testing, but the authors should expect pointed questions on the covariance math and on whether the results change when the kernel respects barriers.

Referee Report

2 major / 2 minor

Summary. The paper proposes a convolution-based covariance framework for sea surface temperature (SST) modeling in the Mediterranean Sea. It discretizes the continuous marine domain into a directed linear network preserving ocean current orientations, encodes dynamics via a Markovian transition-probability matrix on network vertices, and induces a covariance as a weighted combination of a spatial kernel and flow-dependent weights. A penalized estimator regularizes covariance parameters while enforcing hydrodynamic consistency, which is then embedded in a Monte Carlo simulation to refine RCP-based SST projections and identify thermal hot spots.

Significance. If the covariance construction rigorously prevents cross-land correlations and the penalized estimator yields consistent estimates without circularity, the framework would provide a statistically principled approach for incorporating physical barriers and flows into geostatistical models. This could strengthen uncertainty quantification in climate projections and support targeted environmental risk assessments in constrained domains.

major comments (2)

[Covariance construction (around the description of the weighted combination)] The central claim that the induced covariance prevents physical inconsistencies such as correlations across land barriers is load-bearing but not demonstrated. The covariance is explicitly a weighted combination of a spatial kernel (typically Euclidean) and flow-dependent weights from the Markovian transition matrix on the directed network. Unless the spatial kernel is redefined on the network metric or the spatial term is explicitly set to zero across barriers, positive covariance can still be assigned to point pairs separated by land; the flow weights modulate only the second term. This requires an explicit proof or numerical check (e.g., via the covariance matrix entries for land-separated pairs) in the covariance construction section.
[Discretization and network construction] The discretization of the continuous domain into a directed linear network is presented as preserving current orientations and sufficient for flow dynamics, but no validation is shown that this approximation captures the relevant hydrodynamic properties without introducing artifacts in the covariance or hot-spot identification. This assumption underpins both the Markovian matrix and the prevention of inconsistencies.

minor comments (2)

[Abstract and Introduction] The abstract and introduction would benefit from a clearer statement of how the penalized estimator differs from standard geostatistical penalties and what specific hydrodynamic properties are enforced.
[Model formulation] Notation for the spatial kernel, transition matrix, and weighting parameters should be introduced with explicit definitions and dimensions to aid readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thorough and insightful comments, which have helped us identify areas where the manuscript can be strengthened. We address each major comment below and outline the revisions we intend to make in the updated version of the paper.

read point-by-point responses

Referee: The central claim that the induced covariance prevents physical inconsistencies such as correlations across land barriers is load-bearing but not demonstrated. The covariance is explicitly a weighted combination of a spatial kernel (typically Euclidean) and flow-dependent weights from the Markovian transition matrix on the directed network. Unless the spatial kernel is redefined on the network metric or the spatial term is explicitly set to zero across barriers, positive covariance can still be assigned to point pairs separated by land; the flow weights modulate only the second term. This requires an explicit proof or numerical check (e.g., via the covariance matrix entries for land-separated pairs) in the covariance construction section.

Authors: We agree that an explicit demonstration is necessary to support the claim. In the original manuscript, the covariance is constructed such that the flow-dependent weights, derived from the Markovian transition matrix on the directed network, ensure no flow across land barriers, and the spatial kernel is applied only within connected marine components. However, to rigorously address this, we will revise the covariance construction section to include a formal proof showing that the covariance is zero for any pair of points separated by land, based on the zero transition probabilities across barriers. Additionally, we will provide numerical examples computing covariance matrix entries for representative land-separated pairs to illustrate this property. This addition will be included in the revised manuscript. revision: yes
Referee: The discretization of the continuous domain into a directed linear network is presented as preserving current orientations and sufficient for flow dynamics, but no validation is shown that this approximation captures the relevant hydrodynamic properties without introducing artifacts in the covariance or hot-spot identification. This assumption underpins both the Markovian matrix and the prevention of inconsistencies.

Authors: The discretization into a directed linear network was performed using high-resolution ocean current data to align with prevailing flow directions in the Mediterranean Sea. While the manuscript focuses on the resulting model, we acknowledge the value of explicit validation. In the revision, we will add a validation subsection that compares the network's transition probabilities against independent hydrodynamic simulations or observational data on current velocities. We will also conduct a sensitivity analysis to assess the impact of discretization resolution on the identified hot spots and covariance estimates, ensuring no significant artifacts are introduced. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation builds from independent geostatistical and network principles

full rationale

The paper defines a new covariance via discretization of the marine domain into a directed linear network, a Markovian transition matrix, and a moving-average process whose covariance is explicitly a weighted sum of a spatial kernel plus flow-dependent terms. The penalized estimator is introduced to regularize parameters while incorporating external hydrodynamic knowledge. No equation or step reduces by construction to its own inputs, no fitted parameter is relabeled as a prediction, and no load-bearing premise rests solely on self-citation. The framework is therefore self-contained against external benchmarks of spatial statistics and network flows.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the ability to discretize the domain and the validity of the convolution process for capturing flow effects, with free parameters in the covariance estimation.

free parameters (1)

covariance parameters
The complex estimation problem for the weighted combination of spatial kernel and flow-dependent weights requires regularization via penalized estimator, implying fitted parameters.

axioms (1)

domain assumption The marine domain can be discretized into a directed linear network preserving ocean current orientations.
Invoked in the construction of the moving-average stochastic process.

pith-pipeline@v0.9.0 · 5506 in / 1419 out tokens · 72763 ms · 2026-05-08T16:45:04.552264+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

51 extracted references · 42 canonical work pages

[1]

1993 , publisher=

Statistics for Spatial Data , journal =. 1993 , publisher=. doi:10.1002/9781119115151 , author=

work page doi:10.1002/9781119115151 1993
[2]

Mathematical Geology , author=

On the Use of Non-Euclidean Distance Measures in Geostatistics , volume=. Mathematical Geology , author=. 2006 , pages=. doi:10.1007/s11004-006-9055-7 , number=

work page doi:10.1007/s11004-006-9055-7 2006
[3]

A generalized convolution model and estimation for non-stationary random functions , volume=

Fouedjio, Francky and Desassis, Nicolas and Rivoirard, Jacques , year=. A generalized convolution model and estimation for non-stationary random functions , volume=. doi:10.1016/j.spasta.2016.01.002 , journal=

work page doi:10.1016/j.spasta.2016.01.002 2016
[4]

Environmental Modelling & Software , year=

Extraction of hydrological proximity measures from DEMs using parallel processing , volume=. Environmental Modelling & Software , year=. doi:10.1016/j.envsoft.2011.07.018 , author=

work page doi:10.1016/j.envsoft.2011.07.018 2011
[5]

and Peterson, Erin and Theobald, David , year=

Ver Hoef, Jay M. and Peterson, Erin and Theobald, David , year=. Spatial statistical models that use flow and stream distance , url=. United States Department of Commerce: Staff Publications , DOI =
[6]

2020 , howpublished =

Multi Observation Global Ocean 3D Temperature Salinity Height Geostrophic Current and MLD , author =. 2020 , howpublished =

2020
[7]

Water Resources Research , author=

A new method for the determination of flow directions and upslope areas in grid digital elevation models , volume=. Water Resources Research , author=. 1997 , pages=. doi:10.1029/96WR03137 , number=

work page doi:10.1029/96wr03137 1997
[8]

Journal of Agricultural, Biological, and Environmental Statistics , author=

Spatial prediction on a river network , volume=. Journal of Agricultural, Biological, and Environmental Statistics , author=. 2006 , pages=. doi:10.1198/108571106X110649 , number=

work page doi:10.1198/108571106x110649 2006
[9]

Journal of Computational and Graphical Statistics , author =

Zimmerman, Dale and Ver Hoef, Jay , year=. The Torgegram for Fluvial Variography: Characterizing Spatial Dependence on Stream Networks , volume=. doi:10.1080/10618600.2016.1247006 , journal=

work page doi:10.1080/10618600.2016.1247006 2016
[10]

Risser, Mark D. , year=. Review: Nonstationary Spatial Modeling, with Emphasis on Process Convolution and Covariate-Driven Approaches , url=. doi:10.48550/arXiv.1610.02447 , journal =

work page doi:10.48550/arxiv.1610.02447
[11]

Geostatistical modelling on stream networks: Developing valid covariance matrices based on hydrologic distance and stream flow , volume=

Peterson, Erin and Theobald, David and Ver Hoef, Jay , year=. Geostatistical modelling on stream networks: Developing valid covariance matrices based on hydrologic distance and stream flow , volume=. doi:10.1111/j.1365-2427.2006.01686.x , journal=

work page doi:10.1111/j.1365-2427.2006.01686.x 2006
[12]

and Ver Hoef, Jay M

Peterson, Erin E. and Ver Hoef, Jay M. , year=. A mixed-model moving-average approach to geostatistical modeling in stream networks , volume=. Ecology , publisher=
[13]

Doubly Stochastic Matrices and the Diagonal of a Rotation Matrix , volume=

Horn, Alfred , year=. Doubly Stochastic Matrices and the Diagonal of a Rotation Matrix , volume=. American Journal of Mathematics , publisher=. doi:10.2307/2372705 , number=

work page doi:10.2307/2372705
[14]

The European Physical Journal Special Topics 174(1):157--179

Complex networks in climate dynamics , volume=. The European Physical Journal Special Topics , author=. 2009 , pages=. doi:10.1140/epjst/e2009-01098-2 , number=

work page doi:10.1140/epjst/e2009-01098-2 2009
[15]

Europhysics Letters , author=

The backbone of the climate network , volume=. Europhysics Letters , author=. 2009 , pages=. doi:10.1209/0295-5075/87/48007 , number=

work page doi:10.1209/0295-5075/87/48007 2009
[16]

Network-based forecasting of climate phenomena , volume=

Ludescher, Josef and Martin, Maria and Boers, Niklas and Bunde, Armin and Ciemer, Catrin and Fan, Jingfang and Havlin, Shlomo and Kretschmer, Marlene and Kurths, Jürgen and Runge, Jakob and Stolbova, Veronika and Surovyatkina, Elena and Schellnhuber, Hans Joachim , year=. Network-based forecasting of climate phenomena , volume=. Proceedings of the Nationa...

work page doi:10.1073/pnas.1922872118
[17]

Newman, Mark E. J. , year=. The structure and function of complex networks , url=. SIAM Review , DOI=
[18]

Nonparametric estimators over metric graphs , ISSN=

Clemente, Aldo and Arnone, Eleonora and Mateu, Jorge and Sangalli, Laura M , year=. Nonparametric estimators over metric graphs , ISSN=. doi:10.1093/biomet/asag029 , journal=

work page doi:10.1093/biomet/asag029
[19]

Environmetrics , author=

Modeling Anisotropy and Non-Stationarity Through Physics-Informed Spatial Regression , volume=. Environmetrics , author=. 2024 , pages=. doi:10.1002/env.2889 , number=

work page doi:10.1002/env.2889 2024
[20]

Karrer, Brian and Newman, Mark E. J. , year=. Random graph models for directed acyclic networks , volume=. Physical Review E , publisher=. doi:10.1103/PhysRevE.80.046110 , number=

work page doi:10.1103/physreve.80.046110
[21]

and Peterson, Erin and Clifford, David and Shah, Rohan , year=

Ver Hoef, Jay M. and Peterson, Erin and Clifford, David and Shah, Rohan , year=. SSN: An R Package for Spatial Statistical Modeling on Stream Networks , volume=. doi:10.18637/jss.v056.i03 , journal=

work page doi:10.18637/jss.v056.i03
[22]

Newman, M. E. J. , year=. The structure and function of complex networks , url=. doi:10.1137/S003614450342480 , journal=

work page doi:10.1137/s003614450342480
[23]

A Moving Average Approach for Spatial Statistical Models of Stream Networks , volume=

Ver Hoef, Jay and Peterson, Erin , year=. A Moving Average Approach for Spatial Statistical Models of Stream Networks , volume=. doi:10.1198/jasa.2009.ap08248 , journal=

work page doi:10.1198/jasa.2009.ap08248 2009
[24]

Marine heat waves in the Mediterranean Sea: An assessment from the surface to the subsurface to meet national needs , volume=

Dayan, Hugo and McAdam, Ronan and Juza, Mélanie and Masina, Simona and Speich, Sabrina , year=. Marine heat waves in the Mediterranean Sea: An assessment from the surface to the subsurface to meet national needs , volume=. doi:10.3389/fmars.2023.1045138 , journal=

work page doi:10.3389/fmars.2023.1045138 2023
[25]

Geophysical Research Letters , author=

Climate change hot-spots , volume=. Geophysical Research Letters , author=. 2006 , language=. doi:10.1029/2006GL025734 , number=

work page doi:10.1029/2006gl025734 2006
[26]

and Burrows, Michael T

Smith, Kathryn E. and Burrows, Michael T. and Hobday, Alistair J. and King, Nathan G. and Moore, Pippa J. and Gupta, Alex Sen and Thomsen, Mads S. and Wernberg, Thomas and Smale, Dan A. , year=. Biological Impacts of Marine Heatwaves , volume=. doi:10.1146/annurev-marine-032122-121437 , journal=

work page doi:10.1146/annurev-marine-032122-121437
[27]

Butenschön, Momme and Clark, James and Aldridge, John N. and Allen, Julian Icarus and Artioli, Yuri and Blackford, Jeremy and Bruggeman, Jorn and Cazenave, Pierre and Ciavatta, Stefano and Kay, Susan and Lessin, Gennadi and van Leeuwen, Sonja and van der Molen, Johan and de Mora, Lee and Polimene, Luca and Sailley, Sevrine and Stephens, Nicholas and Torre...

work page doi:10.5194/gmd-9-1293-2016 2016
[28]

Marine Heat Waves Hazard 3D Maps and the Risk for Low Motility Organisms in a Warming Mediterranean Sea , volume=

Galli, Giovanni and Solidoro, Cosimo and Lovato, Tomas , year=. Marine Heat Waves Hazard 3D Maps and the Risk for Low Motility Organisms in a Warming Mediterranean Sea , volume=. doi:10.3389/fmars.2017.00136 , journal=

work page doi:10.3389/fmars.2017.00136 2017
[29]

and Dupuis, Paul , year=

Kushner, Harold J. and Dupuis, Paul , year=. Numerical Methods for Stochastic Control Problems in Continuous Time , volume=. doi:10.1007/978-1-4613-0007-6 , publisher=

work page doi:10.1007/978-1-4613-0007-6
[30]

An object-oriented approach to the analysis of spatial complex data over stream-network domains , volume=

Barbi, Chiara and Menafoglio, Alessandra and Secchi, Piercesare , year=. An object-oriented approach to the analysis of spatial complex data over stream-network domains , volume=. doi:10.1016/j.spasta.2023.100784 , journal=

work page doi:10.1016/j.spasta.2023.100784 2023
[31]

2020 , howpublished =

Marine biogeochemistry data for the Northwest European Shelf and Mediterranean Sea from 2006 up to 2100 derived from climate projections , author =. 2020 , howpublished =

2006
[32]

Spatial Models Generated by Nested Stochastic Partial Differential Equations, with an Application to Global Ozone Mapping , volume=

Bolin, David and Lindgren, Finn , year=. Spatial Models Generated by Nested Stochastic Partial Differential Equations, with an Application to Global Ozone Mapping , volume=. The Annals of Applied Statistics , publisher=
[33]

Journal of the Royal Statistical Society Series B: Statistical Methodology73(4), 423–498 (Aug 2011)

An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach , volume=. Journal of the Royal Statistical Society: Series B (Statistical Methodology) , author=. 2011 , pages=. doi:10.1111/j.1467-9868.2011.00777.x , number=

work page doi:10.1111/j.1467-9868.2011.00777.x 2011
[34]

The SPDE approach for spatio-temporal datasets with advection and diffusion , volume=

Clarotto, Lucia and Allard, Denis and Romary, Thomas and Desassis, Nicolas , year=. The SPDE approach for spatio-temporal datasets with advection and diffusion , volume=. doi:10.1016/j.spasta.2024.100847 , journal=

work page doi:10.1016/j.spasta.2024.100847 2024
[35]

and Brown, Simon , year=

Richards, Jordan and Tawn, Jonathan A. and Brown, Simon , year=. Modelling extremes of spatial aggregates of precipitation using conditional methods , volume=. The Annals of Applied Statistics , publisher=. doi:10.1214/22-AOAS1609 , number=

work page doi:10.1214/22-aoas1609
[36]

Journal of the Royal Statistical Society Series B: Statistical Methodology , author=

Spline Smoothing over Difficult Regions , volume=. Journal of the Royal Statistical Society Series B: Statistical Methodology , author=. 2002 , pages=. doi:10.1111/1467-9868.00339 , number=

work page doi:10.1111/1467-9868.00339 2002
[37]

European Journal of Operational Research , author=

Statistical analysis of complex and spatially dependent data: A review of Object Oriented Spatial Statistics , volume=. European Journal of Operational Research , author=. 2017 , pages=. doi:10.1016/j.ejor.2016.09.061 , number=

work page doi:10.1016/j.ejor.2016.09.061 2017
[38]

2023 , url =

R: A Language and Environment for Statistical Computing , author =. 2023 , url =

2023
[39]

Estimating high-resolution Red Sea surface temperature hotspots, using a low-rank semiparametric spatial model , volume=

Hazra, Arnab and Huser, Raphaël , year=. Estimating high-resolution Red Sea surface temperature hotspots, using a low-rank semiparametric spatial model , volume=. The Annals of Applied Statistics , publisher=. doi:10.1214/20-AOAS1418 , number=

work page doi:10.1214/20-aoas1418
[40]

Journal of the Royal Statistical Society Series B: Statistical Methodology , author=

Excursion and Contour Uncertainty Regions for Latent Gaussian Models , volume=. Journal of the Royal Statistical Society Series B: Statistical Methodology , author=. 2015 , month=. doi:10.1111/rssb.12055 , number=

work page doi:10.1111/rssb.12055 2015
[41]

and Sain, Stephan R

French, Joshua P. and Sain, Stephan R. , year=. Spatio-temporal exceedance locations and confidence regions , volume=. The Annals of Applied Statistics , publisher=. doi:10.1214/13-AOAS631 , number=

work page doi:10.1214/13-aoas631
[42]

and Stouffer, Ronald J

Taylor, Karl E. and Stouffer, Ronald J. and Meehl, Gerald A. , year=. An Overview of CMIP5 and the Experiment Design , url=. doi:10.1175/BAMS-D-11-00094.1 , journal =

work page doi:10.1175/bams-d-11-00094.1
[43]

Stochastic Environmental Research and Risk Assessment , author=

Linear Gaussian state-space model with irregular sampling: application to sea surface temperature , volume=. Stochastic Environmental Research and Risk Assessment , author=. 2011 , month=. doi:10.1007/s00477-010-0442-8 , number=

work page doi:10.1007/s00477-010-0442-8 2011
[44]

1998 , publisher=

Markov chains , author=. 1998 , publisher=

1998
[45]

2025 , publisher =

Reinhard Diestel , title =. 2025 , publisher =

2025
[46]

Anderes, Ethan and Møller, Jesper and Rasmussen, Jakob G. , year=. Isotropic covariance functions on graphs and their edges , volume=. The Annals of Statistics , publisher=. doi:10.1214/19-AOS1896 , number=

work page doi:10.1214/19-aos1896
[47]

https://doi.org/10.3150/23-bej1647

Bolin, David and Simas, Alexandre B. and Wallin, Jonas , year=. Gaussian Whittle–Matérn fields on metric graphs , volume=. Bernoulli , publisher=. doi:10.3150/23-BEJ1647 , number=

work page doi:10.3150/23-bej1647
[48]

Sun, Ying and Genton, Marc G. , year=. Functional Boxplots , volume=. Journal of Computational and Graphical Statistics , publisher=. doi:10.1198/jcgs.2011.09224 , number=

work page doi:10.1198/jcgs.2011.09224 2011
[49]

Climatic Change , author=

The representative concentration pathways: an overview , volume=. Climatic Change , author=. 2011 , month=. doi:10.1007/s10584-011-0148-z , number=

work page doi:10.1007/s10584-011-0148-z 2011
[50]

Biological Reviews , author=

Climate change and freshwater biodiversity: detected patterns, future trends and adaptations in northern regions , volume=. Biological Reviews , author=. 2009 , pages=. doi:10.1111/j.1469-185X.2008.00060.x , number=

work page doi:10.1111/j.1469-185x.2008.00060.x 2009
[51]

and Secchi, Piercesare and Domanin, Maurizio and Nobile, Fabio , year=

Azzimonti, Laura and Sangalli, Laura M. and Secchi, Piercesare and Domanin, Maurizio and Nobile, Fabio , year=. Blood Flow Velocity Field Estimation Via Spatial Regression With PDE Penalization , volume=. Journal of the American Statistical Association , publisher=. doi:10.1080/01621459.2014.946036 , number=

work page doi:10.1080/01621459.2014.946036 2014