Curvature and confinement: robustness of dryland vegetation patterns on sloped, undulating terrain
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Nonlinear Sciences. The day in nonlinear science: chaos, patterns, networks and adaptation.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Curvature and confinement".
Mira: Here's the gist of the paper. We consider a topographic extension to the well known Klausmeier model describing vegetation on sloped undulating terrain with a primary uphill direction and perpendicular small amplitude undulation forming ridges and valleys The authors investigate how the terrain impacts the resilience of the vegetation by linking undulation amplitude to pattern dynamics, finding that at small undulation, the ridge plays a critical role in determining pattern existence and stability Model Formulation The investigation uses a system of reaction-diffusion-advection equations to model dryland vegetation patterns on nonuniformly sloped terrain.
Kai: The dimensionless equations evolve a water field, W(x, y, t), and a biomass field, B(x, y, t) via these equations.
Mira: The terrain is modeled by an undulating elevation function ζ(x, y) = x − σ cos(kyy), where ky = 2π/Ly.
Kai: The velocity V is proportional to the gradient of the elevation function, V = −V0∇ζ, which depends only on y.
Title and authors: Kai: Moving into the specific title, "Curvature and confinement: robustness of dryland vegetation patterns on sloped, undulating terrain," it really hammers home that the shape of the terrain, specifically its curvature, dictates how stable those vegetation stripes are >
Mira: Right. It’s not just about having a slope; it’s about the way that slope curves, because that curvature seems to affect whether you get stable patterns or if they just fall apart when conditions get tough >
Lev: I wonder what kind of mathematical tools they use to track that curvature effect dynamically, since modeling real systems with those kinds of geometric constraints is always tricky >
Kai: They use analysis and numerical simulations together to reduce the full two-dimensional dynamics down to these ridge and valley models, which they argue gives them insight into the whole picture >
Mira: The summary says that in particular, when they look at sufficiently small terrain undulation, they see a wavy stripe pattern appear right at the start of pattern formation with the same number of bands on both the ridge and valley >
Lev: That suggests a specific relationship between those two features—the ridge and valley—that dictates the initial pattern configuration before things get more complex >
The paper's summary: Kai: So, to summarize what they found, the paper investigates how undulation amplitude affects pattern dynamics by linking it to how the ridge and valley equations behave in those simplified one-dimensional models >
Mira: They’re looking at several pattern types emerging from their simulations, including x-uniform states, patterned uniform hybrid states, rigid traveling waves, shearing waves, and confined traveling waves >
Lev: Those different pattern types sound like they would require very different sets of conditions to sustain them on a real system; what does that imply about the stability landscape >
Kai: When undulation amplitude is small, they find a wavy stripe pattern appears at onset with the same number of bands on the ridge and valley, and these transitions in wavenumber happen almost simultaneously as rainfall decreases >
Mira: That simultaneous tracking of wavenumber changes with precipitation level suggests a strong coupling between those two features under low undulation conditions >
Lev: But what happens when the undulation amplitude gets bigger? The paper mentions that for slightly larger undulation, patterns can dislocate and select different wavenumbers, which is a key detail we need to look at >
The paper's improvements: Kai: So, one of the main points they make about their approach is that by using the ridge and valley equations as approximations—the Busse balloon models—they can capture the downslope water flow along both the ridge and along the valley >
Mira: But they admit those approximations come with a catch because they neglect the transverse diffusive spread of biomass, which is a crucial part of what's happening in two dimensions >
Lev: That means their one-dimensional models are really just capturing the basic advection, but missing the diffusion aspect that governs how things spread sideways across those features >
Kai: To fix that limitation, they complement this analysis with numerical simulations of the full two-dimensional problem to see what happens when you include that transverse diffusion >
Mira: And in those simulations, they find results depend qualitatively on the undulation amplitude of the terrain, and specifically in the case of sufficiently small terrain undulation, a wavy stripe pattern appears at onset with matching band numbers >
Lev: So their improvement is essentially showing that while the simplified models give you a qualitative idea based on amplitude, you need the full simulation to see if those specific initial conditions actually materialize >
Conclusion: Kai: So, wrapping up the paper "Curvature and confinement: robustness of dryland vegetation patterns on sloped, undulating terrain," they conclude that for small undulation, the increased sensitivity of vegetation on the ridge to aridity stress causes it to be less resilient than models assuming a uniformly sloped terrain would predict >
Mira: And because of that sensitivity, this leads to earlier pattern onset localized right on the ridge itself when rainfall is low >
Lev: That’s a big implication for resilience, suggesting that if you build hardware or model systems with these topographical features, you need to account for that extra vulnerability at the ridge >
Kai: Then they say that for slightly larger undulation amplitudes, patterns can actually dislocate and select different wavenumbers, which allows stripes to confine into the valleys where they can survive longer under protection from the terrain >
Mira: They also found something interesting about extinction points: for small undulation, extinction of patterns occurs at higher rainfall levels than on uniformly sloped terrain >
Lev: And for larger undulation, there’s an abrupt change where that extinction point decreases close to linearly with a slope very close to the one found for the valley extinction curve >
Kai: So what they suggest is that coupling the ridge and valley systems into a system of four reaction diffusion equations could explain why confined stripes reach instabilities before the boundary of the valley balloons >
Mira: It’s a lot to digest, but essentially, it shows that small undulation has an outsized impact on pattern formation and transition, and larger undulation allows for better confinement in the valleys >
Lev: I just think we need to see how those four coupled equations behave when you try to implement them on real hardware with finite precision or noise added, because that's where the real test of these robustness claims lies >
Timothy V Roberts, Punit Gandhi, Mary Silber
Department of Statistics and Committee on Computational and Applied Mathematics, The University of Chicago · Department of Mathematics and Statistics, Virginia Commonwealth University · NSF-Simons National Institute for Theory and Mathematics in Biology
nlin.PS
Submitted: 2026-10-30
Updated: 2026-10-30
Comments: 28 pages, 10 figures, 1 table, 2 appendices
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 87/100
The gist: We consider a topographic extension to the well known Klausmeier model describing vegetation on sloped undulating terrain with a primary uphill direction and perpendicular small amplitude undulation…
Key concepts
- Reaction-Diffusion-Advection Equations
- These are mathematical equations used to describe how two things—water and vegetation biomass—change over space and time. The 'advection' part specifically accounts for movement driven by the slope of the terrain, which influences where the vegetation grows.
- Ridge and Valley Line Models
- The authors simplify complex 2D dynamics by looking at one-dimensional models along specific lines: ridges (high points) and valleys (low points). This simplification helps reveal how terrain features specifically influence pattern behavior in these key topographical areas.
- Pattern Confinement
- This refers to the phenomenon where vegetation stripes, which might otherwise cover the entire landscape, become restricted or 'confined' to the valleys created by the terrain undulations. This confinement is shown to increase survival chances under dry conditions when undulation amplitudes are slightly larger.
Terminology
Summary
We consider a topographic extension to the well known Klausmeier model describing vegetation on sloped undulating terrain with a primary uphill direction and perpendicular small amplitude undulation forming ridges and valleys The authors investigate how the terrain impacts the resilience of the vegetation by linking undulation amplitude to pattern dynamics, finding that at small undulation, the ridge plays a critical role in determining pattern existence and stability
Model Formulation
The investigation uses a system of reaction-diffusion-advection equations to model dryland vegetation patterns on nonuniformly sloped terrain. The dimensionless equations evolve a water field, W(x, y, t), and a biomass field, B(x, y, t) via these equations. p. 2 The terrain is modeled by an undulating elevation function ζ(x, y) = x − σ cos(kyy), where ky = 2π/Ly. p. 3 The velocity V is proportional to the gradient of the elevation function, V = −V0∇ζ, which depends only on y. p. 3
Pattern Dynamics and Ridge/Valley Equations
The analysis reduces the full 2-dimensional dynamics to two related 1-dimensional problems by considering approximate ridge and valley line models. These one-dimensional models are derived by setting y to the values of the valley (y=0) and ridge lines (y = ±Ly/2), resulting in modified Klausmeier models. p. 4 The sign of s in the loss term of equation (2. p. 44a) is interpreted as a curvature effect due to the terrain, where water is lost to the slope on the ridge and accumulates in the valley. p. 4
Numerical Simulations and Pattern Types
Numerical simulations reveal various pattern types depending on rainfall level (a) and undulation amplitude (σ). These include x-uniform states, patterned-uniform hybrid states, rigid traveling waves, shearing waves, and confined traveling waves. p. 5 For sufficiently small terrain undulation, a wavy stripe pattern appears at onset with the same number of bands on the ridge and valley. p. 2
Resilience and Extinction Points
The study examines how pattern formation affects ecosystem resilience under aridity stress. For small undulation, the increased sensitivity of the ridge to aridity causes vegetation to be less resilient than indicated by uniformly sloped models, leading to earlier pattern onset on the ridge. p. 0 Conversely, for undulation amplitudes above a threshold, vegetation can survive longer by becoming localized in the valleys. p. 0 The best case extinction point for patterns is found at higher rainfall levels than predicted in the uniform terrain case when undulation is small. p. 0
Conclusion and Future Directions
The authors conclude that for small undulation, the increased sensitivity of the ridge to aridity stress has an outsized impact on pattern formation and transition, leading to earlier pattern onset localized on the ridge. For slightly larger undulation, patterns can dislocate and select different wavenumbers, allowing stripes to confine to valleys where they can survive longer under protection from the terrain. p. 0 The authors suggest that coupling the ridge and valley systems into a system of four reaction diffusion equations could explain why confined stripes reach instabilities before the boundary of the valley balloon. p. 3 p. 3
Acknowledgements
This research was supported in part by grants from the NSF (DMS-2235451) and Simons Foundation (MPSNITMB-00005320) to the NSF-Simons National Institute for Theory and Mathematics in Biology. PG also gratefully acknowledges support of the Simons Foundation Collaboration Grant for Mathematicians 848615. p. 0 Computing resources were provided by the High Performance Research Computing (HPRC) Core Facility at Virginia Commonwealth University. p. 0 PG and MS would like to thank Justin Tzou for their discussions on patterns in heterogeneous systems. p. 1 PG and TVR would like to thank David Lloyd and Jens Rademacher for their suggestions on numerical computation of patterns and stability. p. 2 Arjen Doelman for their discussions on the development and interpretations of our results to resilience. p. 0
References
[1] R. Bastiaansen, O. Jaïbi, V. Deblauwe, M. B. Eppinga, K. Siteur, E. Siero, S. Mermoz, A. Bouvet, A Doelman and M Rietkerk
[2] Robbin Bastiaansen, Paul Carter and Arjen Doelman
[3] Mara Baudena, Jost Von Hardenberg and Antonello Provenzale
[4] Jamie J. R Bennett, Bidesh K Bera, Michel Ferré, Hezi Yizhaq, Stephan Getzin and Ehud Meron
[5] Bidesh K Bera, Omer Tzuk, Jamie JR Bennett and Ehud Meron
[6] S. B Boaler and C A Hodge
[7] F Borgogno, P D’Odorico, F Laio and L Ridolfi
[8] F H Busse
[9] Paul Carter and Arjen Doelman
[10] Octavia V Crompton and Sally E Thompson
[11] V Deblauwe, N Barbier, P Couteron, O Lejeune and J Bogaert
[12] V Deblauwe, P Couteron, J Bogaert and N Barbier
[13] E Doedel, R C Paffenroth, A R Champneys, T F Fairgrieve, Yuri A Kuznetsov, Bart E Oldeman Björn Sandstede and X Wang
[14] Arjen Doelman, Björn Sandstede, Arnd Scheel and Guido Schneider
[15] P Gandhi, S Iams, S Bonetti and M Silber
[16] Punit Gandhi, Sara Bonetti, Sarah Iams Amilcare Porporato and Mary Silber
[17] Punit Gandhi, Lily Liu and Mary Silber
[18] Punit Gandhi, Matthew Oline and Mary Silber
[19] Punit Gandhi, Lucien Werner, Sarah Iams Karna Gowda and Mary Silber
[20] R Gardner
[21] K Gowda, S Iams and M Silber
[22] J E G W Greenwood
[23] C F Hemming
[24] A Kassam and L N Trefethen
[25] C A Klausmeier
[26] Alexandra G Konings, Stefan C Dekker, Max Rietkerk and Gabriel G Katul
[27] WA Macfadyen
[28] Gavan S McGrath, Kyungrock Paik and Christoph Hinz
[29] E Meron
[30] Ehud Meron
[31] Ehud Meron
[32] Athanasios Paschalis, Gabriel G Katul, Simone Fatichi Gabriele Manoli and Peter Molnar
[33] David Pinto-Ramos and Ricardo Martinez-Garcia
[34] Jens D M Rademacher, Björn Sandstede and Arnd Scheel
[35] Max Rietkerk, Robbin Bastiaansen, Arjen Banerjee Johan van de Koppel Mara Baudena and Arjen Doelman
[36] Björn Sandstede and Arnd Scheel
[37] Jonathan A Sherratt
[38] Jonathan A Sherratt
[39] Jonathan A Sherratt
[40] Jonathan A Sherratt and Gabriel J Lord
[41] E Siero, A Doelman, M B Eppinga, J D M Rademacher, M Rietkerk and K Siteur
[42] Koen Siteur, Eric Siero Maarten B Eppinga Jens D.M Rademacher Arjen Doelman and Max Rietkerk
[43] RO Slatyer
[44] Justin C Tzou and Leo Tzou
[45] C Valentin, J M d’Herbés and J Poesen
[46] Sjors van der Stelt, Arjen Doelman, Geertje Hek and Jens D M Rademacher
[47] H Yizhaq, S Sela, T Svoray, S Assouline and G Bel
How the training signal is built
The authors use a combination of analysis and numerical simulations to investigate the links between undulation amplitude and pattern dynamics. They take advantage of features of the domain, specifically the ridge and valley, to reduce the analysis to two related 1-dimensional problems. p. 0 They find that a combination of the ridge and valley equations provides insight into the full 2-dimensional dynamics. p. 0
Where the gains come from
The key findings are that at small undulation, the added sensitivity of vegetation on the ridge to aridity stress causes vegetation to be less resilient than indicated by uniformly sloped models. For undulation amplitudes above a threshold, vegetation is able to survive longer by becoming localized in the valleys, but this requires increasing the amplitude further to mitigate detrimental effects. p. 0 The investigation also shows that for small undulation, the ridge and valley patterns remain strongly locked together, leading to a loss in resilience because instabilities on the ridge cause extinction at higher rainfall levels than on uniformly sloped terrain. p. 0 For slightly larger undulation, patterns can dislocate and select different wavenumbers, allowing stripes to confine to the valleys where they can survive to lower rainfall levels under protection from the terrain. p. 0
Pruning redundant reasoning tokens
The paper uses phenomenological observations of [19] to reduce the study of full 2-dimensional patterns to uncoupled 1-dimensional patterns on the ridge and valley. The authors exploit existing techniques like Busse-balloons and linear stability analysis of x-uniform solutions. p. 9 They find that for small undulation, the increased sensitivity of the ridge to aridity stress has an outsized impact on pattern formation and transition. p. 0 The paper also notes that confined stripes can go through transverse instabilities, phenomena not captured by uncoupled ridge and valley equations. p. 2
Discussion
The authors argue that for small undulation, the increased sensitivity of the ridge to aridity stress has an outsized impact on pattern formation and transition. They confirm that this leads to earlier pattern onset, starting with patterns localized on the ridge. p. 2 For slightly larger undulation, the strong locking is lost, allowing the ridge and valley patterns to dislocate and select different wavenumbers. p. 4 This leads to an enhancement of vegetation resilience when stripes confine to the valleys at low rainfall levels. p. 0 The authors suggest that coupling the ridge and valley systems into a system of four reaction diffusion equations could explain why confined stripes reach instabilities before the boundary of the valley balloon.
Extinction of vegetation at low rainfall
The study shows that at small undulation extinction of patterns occurs at higher rainfall levels than it would for the uniformly sloped terrain. For moderate undulation, extinction of confined patterns occurs for values of a slightly below the uniformly sloped extinction point. p. 2 The authors estimate the true 2-dimensional extinction point by ramping experiments and find that for small undulation, vegetation is less robust than on uniformly sloped terrain. p. 0 For larger undulation, there is an abrupt change where the extinction point decreases close to linearly with a slope very close to that found for the valley extinction curve. The authors conclude that for small undulation, stripes cover the full domain, but for larger undulation, solutions confine to the valleys. p. 0
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved AI can predict vegetation pattern formation based on terrain undulation amplitude by determining if it falls below a well-defined threshold, as found in
For undulation amplitudes smaller than a well-defined threshold, the added sensitivity of vegetation on the ridge to aridity stress causes vegetation to be less resilient than indicated by uniformly sloped models.
-
Improved AI can assess ecosystem resilience by comparing predicted extinction points under different terrain undulations against those from uniformly sloped models, as demonstrated in
extinction of patterns occurs at higher rainfall levels than it would for the uniformly sloped terrain.
-
Improved AI can distinguish between pattern types such as rigid traveling waves and shearing solutions by analyzing the relative speeds of ridge and valley stripes, which can be tracked using methods like computing the wavenumber by counting bands on the domain.
-
Improved AI can predict whether vegetation will be confined to valleys or cover the full domain based on rainfall levels and undulation amplitude, noting that
For small undulation, we find stripes that cover the full domain.
Related papers
- Periodic orbits underlying optical turbulence in the Lugiato-Lefever equation
- Rate-selective synchronization in switching networks: when timing meets structure
- Thouless Pumping of One-Dimensional Dark Solitons and Domain Walls
- Phenotype-structured nonlocal aggregation models I: signal production and response
- Dark solitons in the fractional NLS equation
- Near-Threshold Kink Trapping Enhances Defect Healing in an Adaptive Chain