Pith. sign in

REVIEW 4 major objections 6 minor 34 references

Walking Through Complex Spatial Patterns of Climate and Conflict-Induced Displacements

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A flow-weighted random walk on Somalia's directed settlement network reconstructs likely displacement journeys and quantifies hazard exposure along them, with drought probable on nearly every route and conflict receding at each step.

desk verdict A transparent but methodologically flawed proof-of-concept: walk statistics are not conditioned on reaching the labeled destination, so the headline hazard-decay result may be an artifact of failed walks. read the letter →

arxiv 2506.22120 v2 pith:B37ENCKH submitted 2025-06-27 physics.soc-ph nlin.AO

David Carranza , Devansh Sharma , Francisco Malveiro , Gustavo Kohlrausch , Jisha Mariyam John , Kaloyan Danovski , Malvina Bozhidarova , Rui Zheng
show 1 more author
Sandro Sousa
This is my paper
classification physics.soc-phnlin.AO
keywords internaldisplacementrandomwalksdiffusiononnetworksmigrationSomaliahazardexposureclimateconflict
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 argues that the missing middle of a displacement journey—the route actually taken between origin and destination—can be reconstructed from aggregate origin–destination data by modelling movement as a diffusion process. On a directed network whose edges carry the observed numbers of people moving between Somali settlements, a random walk biased by those flow volumes produces a distribution of likely trajectories, and each settlement's recorded displacement causes assign each path a hazard-exposure profile. Applied to 20,220 recorded displacements in Somalia from February to June 2025, the simulation finds that drought is a probable hazard on paths of every length, while conflict exposure is strongest in the first steps and declines at every subsequent step. The claim matters because humanitarian planning generally lacks route-level information; if flow data alone can estimate where hazards are encountered, agencies can anticipate exposure in regions where route tracking does not exist.

What carries the argument

The central object is the biased random walk on the directed, weighted displacement network: settlements are nodes, observed moves are directed edges, and the weight of an edge is the fraction of all departures from its source settlement that went to that target. Each step follows that flow-weighted probability, making the walk memoryless: the next settlement depends only on the current one. The exposure calculation then converts the walk into a hazard profile: summing, for each hazard, the hazard-specific displacement counts of every node visited, then normalizing by total displacement along the path, gives the path's exposure probability for that hazard; averaging these over all walks and source nodes yields the step-dependent hazard likelihoods reported in the results.

What would settle it

Collect route-level data for a sample of the same displacement episodes—GPS traces, phone records, or follow-up surveys asking which settlements were passed through—and compare actual waypoint sequences with the most probable simulated walks; if real routes diverge systematically from flow-weighted paths, or their stepwise hazard exposure differs systematically from the simulated curves, the reconstruction claim fails.

Watch

Extended reading notes

Core claim

The central discovery is that a flow-weighted random walk on the settlement network is enough to produce plausible displacement trajectories and meaningful hazard-exposure statistics from data that record only origins and destinations. The transition probability from settlement i to settlement j is set to the observed share of people who left i for j, each of 1,000 simulated walks per origin–destination pair moves step by step under that rule until it reaches the target, a sink node, or 1,000 steps, and every visited settlement contributes its hazard-specific displacement counts to the path's exposure vector. Aggregating across all simulated walks, drought and conflict dominate exposure, drought appears on paths of all lengths, conflict exposure diminishes step by step, and exposure probabilities are widely dispersed rather than clustered around a typical profile.

Load-bearing premise

The reconstruction rests on treating each move as a memoryless, flow-weighted choice: a displaced person's next settlement depends only on the current settlement and the observed outflow shares, not on the final destination, the path already traveled, the timing of hazards, or social networks.

Editorial extensions

If this is right

  • For any settlement pair, the method yields a full distribution of plausible routes and a hazard-exposure vector, so agencies can estimate route-level risk for corridors that have no route observations.
  • Because drought exposure is likely on paths of every length, drought preparedness becomes a default component of humanitarian response across the displacement network rather than a region-specific concern.
  • Because conflict and flood exposure concentrate in the first few steps, evacuation planning and assistance should target the earliest phase of movement and the immediate transit hubs.
  • The wide dispersion of exposure probabilities means corridor averages hide dangerous outliers, so operational planning should use exposure quantiles by origin and path length.
  • The framework transfers to any origin–destination displacement matrix with per-location hazard counts, offering a route-exposure estimation method for regions expected to face intensifying hazards.

Reading between the lines

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

  • A natural validation would compare simulated routes against GPS traces or mobile-phone location data from displaced households; strong path-dependence in real traces would indicate that target-aware or history-dependent transition rules are needed.
  • The dominance of drought may be inflated by its spatial breadth, since drought appears as the dominant hazard on many nodes; a null model that randomly shuffles hazard labels across settlements could separate network-topology effects from genuine hazard prevalence.
  • By reweighting edges with projected future displacement flows, the reconstruction framework could be turned into a forward-looking exposure early-warning tool for regions under anticipated climate or conflict stress.
  • The short observed chains in the data mean the step-by-step decline of conflict exposure is measured over few steps; testing on multi-year or longer-chain data would show whether the decline reflects hazard avoidance or simply the brevity of recorded moves.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes a diffusion-based method to reconstruct likely displacement routes from origin-destination data. Using Somalia's DTM dataset (20,220 movements, Feb-Jun 2025), the authors build a directed, weighted network of settlement-to-settlement flows and simulate biased random walks whose transition probabilities are the observed flow proportions (Eq. 1). Each node carries hazard-specific displacement counts, and path exposure to a hazard is defined as the ratio of hazard counts to total counts along the path (Eqs. 3-5). The authors report that drought and conflict are the dominant hazards along simulated trajectories, that drought is highly probable along any path regardless of length, and that conflict exposure becomes less prominent at each subsequent step. They frame the contribution as a scalable, parameter-lean framework for estimating hazard exposure along displacement trajectories from origin-destination data alone.

Significance. If the central claims were fully supported, the framework would be a practically valuable addition to displacement analysis: it couples network-based mobility modeling with multi-hazard exposure attribution using only routinely collected origin-destination data. The use of real IOM DTM data, the explicit multi-hazard perspective, and the transparent discussion of limitations are strengths. However, the main headline results—drought dominance and conflict step-decay—are currently computed with definitions that do not condition walks on reaching their labeled target, and the exposure metric in the early-step analysis is not a properly normalized probability. These issues undermine the route-reconstruction interpretation and need to be fixed before the paper can support its claims. The paper does not provide code or a data availability statement, which limits reproducibility.

major comments (4)
  1. [Section 2.4, Eqs. (1) and (6)-(7)] The target node g never enters the transition rule; it appears only as a stopping condition. A walk generated for pair (s,g) that reaches a sink or the 1000-step cap still contributes to Pi(h|s) and to the route-level statistics, so the reported 'probabilistic trajectories' are not conditioned on actually reaching g. Supplementary Figure S2 reports a maximum path length of 24 despite the 1000-step cap, which is consistent with many walks terminating in sinks before reaching their labeled target. Because the headline step-decay result (drought dominates; conflict decays) is computed over all walks, it may be an artifact of averaging over failed walks. Please re-estimate Eqs. (6)-(7) and all route-level results restricted to walks that successfully hit g before termination, or explicitly reinterpret the objects as unconditional flow-following walks and adjust the claims accordingly.
  2. [Eq. (6)] Pi(h|s) is not a probability or a proportion as claimed. It sums raw hazard counts Dh(v_i) over walks and divides by the number of walks that reached step i, with no normalization by Dtotal(v_i) or by the sum over hazards. Consequently Pi(h|s) is an average hazard count per walk, not a likelihood of encountering hazard h, and the pie charts in Figure 3B are not distributions over hazard types unless an additional, unspecified normalization is applied. This affects the central early-step analysis and the comparison across hazards.
  3. [Section 2.3, Eqs. (3)-(5)] The path exposure Eh(P) is a weighted average of node-level hazard cause shares Dh(v)/Dtotal(v) from the same DTM dataset that defines the flow network. The observed dominance of drought and conflict along simulated paths is therefore a direct algebraic consequence of the cause composition in the input data (Figure S1), not an independent finding about routes. The authors should state this explicitly and provide a validation or comparison—for example, against direct origin-destination hazard shares or independent hazard event data—to show what the path-based analysis adds beyond the node-level composition.
  4. [Section 4, limitations] The paper acknowledges that simulated paths are independent and ignore path dependence, and that the authors 'do not directly predict the path taken by individuals but estimate a path that they could have taken.' This is honest, but it directly limits the central claim of reconstructing 'likely trajectories.' I recommend adding a sensitivity analysis with destination-aware or history-dependent transition rules, or at least quantifying the fraction of walks that reach their assigned target, so readers can calibrate how much the route-reconstruction interpretation depends on the Markov assumption.
minor comments (6)
  1. [Introduction] There is a typo in 'globaly' near the start of Section 1; please correct to 'globally.'
  2. [Eq. (6)] The notation M(i)_s is defined as the total number of walks from s that reached step i, but the sum over g in V and j = 1..M is not fully indexed; clarify how M(i)_s is computed from the realized walks.
  3. [Supplementary Figure S2] The caption says path lengths are 'aggregated over M simulated trajectories for each of the N locations as origin,' while Section 2.4 says walks are generated for each ordered pair (s,g); clarify whether target nodes are used in the path-length statistics.
  4. [Section 2.2] The 'largest connected component' should specify whether it is the largest weakly or strongly connected component; this matters because random walks in a directed, sink-heavy network can become trapped.
  5. [Data availability] No data or code availability statement is provided; for reproducibility, please include the cleaned settlement registry and simulation code, or a clear statement of availability.
  6. [Figure 3A] The x-axis label 'Density of probabilities' is vague; specify that these are kernel density estimates of exposure probabilities over source settlements.

Circularity Check

1 steps flagged · score 6.0 of 10

Hazard-dominance result is a reweighted restatement of the input cause composition, but the step-decay result is a network-derived property.

  1. renaming known result [Section 2.4, Eq. (5); Section 3, Fig. 3; Supplementary Fig. S1]
    "The exposure probability for the path P to hazard h is defined as E_h(P) = D_h(P)/D_total(P) ... First, we note that these results confirm our primary observation that droughts and conflicts are predominantly responsible for driving the displacement of individuals. Drought risk, in particular, is a highly probable hazard along any path, regardless of its length, starting from any location."

    Eq. (5) defines path exposure as a ratio of the same node-level hazard counts D_h(v) that are taken directly from the displacement dataset used to construct the network (Section 2.2). The walk's transition probabilities are set equal to the observed flow proportions (Eq. 1). Hence the aggregate P_i(h|s) of Eqs. (6)-(7) is a flow-weighted average of those input hazard counts. The raw data already show that drought and conflict predominate (Fig. S1), so the headline result that drought and conflict dominate exposure along simulated paths is a reweighted restatement of the input cause composition, not an independent prediction. The step-dependent decay of conflict in Fig. 3B, however, depends on network topology and is not purely circular.

full rationale

The central hazard-dominance claim is partially circular because the hazard exposure measure E_h(P) is defined as a ratio of the same node-level hazard counts D_h(v) that are input from the displacement dataset, and the random walk samples nodes according to observed flow proportions. Thus the result that drought and conflict are predominant along most paths largely restates the cause composition in Fig. S1. However, the paper also reports a step-dependent decay of conflict and flood risk that emerges from the network topology and is not directly present in the raw cause counts; that part is an independent model output. No self-citation is load-bearing. The unvalidated Markov assumption and the failure to condition on successful arrivals are correctness risks, not circularity. Overall score 6 reflects the partial reduction of the main exposure result to its inputs.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central quantities (hazard exposure along paths) are deterministic functions of the input data: the edge weights in Eq. (1) and hazard counts in Eq. (3) come from the same DTM dataset. No new entities are introduced. The apparent 'predictions' (drought-dominated exposure, conflict decay) are weighted averages of the input cause distribution on the flow network; this is why the circularity burden is not negligible.

free parameters (3)
  • M (number of simulated walks per source-target pair) = 1000
    Chosen without a convergence or sensitivity analysis; the reported exposure distributions in Fig. 3 may depend on this value.
  • k (number of early steps for step-dependent analysis) = 5
    Arbitrary cutoff for classifying 'early phases' of displacement; Figure 3B is specific to this horizon.
  • Max walk length = 1000 steps
    Termination rule used to keep simulations finite; it can truncate long walks and affect statistics for regions that are weakly connected to the target.
assumptions (5)
  • domain assumption The flow-weighted transition probabilities define a first-order Markov process for individual routes; the next settlement depends only on the current one and not on history, intended target, or transport mode.
    Invoked in Eq. (1) and used throughout the simulation. The paper itself later states path dependence is not captured (Section 4).
  • ad hoc to paper Observed direct origin-destination displacements can be chained into viable multi-step routes that approximate real journeys.
    This is the core assumption of the route-reconstruction idea; no route-level or GPS data is used to support it (Section 4: 'We assume that trips passing by settlements between these two regions are a viable route').
  • ad hoc to paper Hazard counts at a settlement Dh(v) indicate the hazard exposure of a traveler passing through that settlement.
    Dh(v) is the number of people displaced from v due to hazard h, but a person merely passing through v may not face that hazard; this equates origin-level hazards with en-route hazards (Eq. 5).
  • domain assumption The recorded movements in the 25 districts are representative of the actual displacement system for all of Somalia during the period.
    The dataset covers only 25 districts and the largest connected component; unobserved or informal routes are omitted, as the authors acknowledge in Section 4.
  • domain assumption Restricting to the largest connected component does not bias the global hazard exposure estimates.
    Isolated node pairs are excluded from simulation (Section 2.2); the effect of this exclusion is not quantified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Walking Through Complex Spatial Patterns of Climate and Conflict-Induced Displacements." pith.science (2026). https://pith.science/paper/B37ENCKH

@misc{pith2026250622120,
  author       = {Pith},
  title        = {Pith review of: Walking Through Complex Spatial Patterns of Climate and Conflict-Induced Displacements},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B37ENCKH}},
  note         = {Machine review of arXiv:2506.22120}
}
read the original abstract

Extreme weather events are projected to intensify global migration, increase resource competition, and amplify socio-spatial phenomena, including intergroup conflicts, socioeconomic inequalities, and unplanned displacements, among others. Addressing these challenges requires consolidating heterogeneous data to identify, estimate, and predict the dynamical process behind climate-induced movements. We propose a novel hybrid approach to reconstruct hazard-induced displacements by analysing the statistical properties of a diffusion process (walks) that explores the spatial network constructed from real displacements. The likely trajectories produced by the walks inform the typical journey of individuals, identifying potential hazards that may be encountered when fleeing high-risk areas. As a proof of concept, we apply this method to Somalia's detailed displacement tracking matrix, containing 20,220 movements dating from February 8 to June 18, 2025. We reconstruct the likely routes that displaced persons could have taken when fleeing areas affected by conflict or climate hazards. We find that individuals using the most likely paths based on current flows would experience mainly droughts and conflicts, while the latter becomes less prominent at every subsequent step. We also find that the probability of conflict and drought across all trajectories is widely dispersed, meaning that there is no typical exposure. This work provides an understanding of the mechanisms underlying displacement patterns and a framework for estimating future movements in areas expected to face increasing hazards.

Figures

Figures reproduced from arXiv: 2506.22120 by the authors.

Figure 1
Figure 1. Descriptive summary and constructed network of Somalia’s displacement flows. (A) Descriptive statistics of displacements, including demographic group, number of displacements, and top regions of origin. The majority of movements recorded in the dataset are associated with the first displacement of the population. Children are the most represented demographic group across all displacements. (B) Graphical representati… view at source ↗
Figure 2
Figure 2. Path reconstruction and spatial displacement flows. (A) Reconstruction of the path from an origin to a destination point using the displacement matrix. Starting with an origin-destination pair (top panel), running simulations allows us to estimate a potential path between those points (bottom panel). Numbers in the figure correspond to node ID. Such paths enable us to identify the hazards people are likely to encoun… view at source ↗
Figure 3
Figure 3. Hazard probabilities and their time dependency. (A) Density of probabilities of hazard exposure for paths starting from distinct settlements. Each dot in the scatterplot represents the probability of encountering the respective hazard over all simulated trajectories, starting from a single location. Conflict and drought display a wider dispersion of probability of occurrence across the path compared to the other haz… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

34 extracted references · 32 canonical work pages

  1. [1]

    World Migration Report 2024,

    M. McAuliffe and L. Oucho, “World Migration Report 2024,” International Organization for Migration (IOM), Geneva, Tech. Rep. PUB2023/047/L*, 2024, p. 384. [Online]. Available:https://publications.iom.int/ books/world-migration-report-2024

    work page 2024

  2. [2]

    Global Report on Internal Displacement (GRID),

    “Global Report on Internal Displacement (GRID),” Internal Displacement Monitoring Centre (IDMC), Tech. Rep., May 2025. DOI: 10.55363/IDMC.XTGW2833. (visited on 07/02/2025)

    work page doi:10.55363/idmc.xtgw2833 2025

  3. [3]

    Civil conflicts are associated with the global climate,

    S. M. Hsiang, K. C. Meng, and M. A. Cane, “Civil conflicts are associated with the global climate,” Nature, vol. 476, no. 7361, pp. 438–441, 2011

    work page 2011

  4. [4]

    Climate change and conflict,

    V . Koubi, “Climate change and conflict,” Annual review of political science, vol. 22, no. 1, pp. 343–360, 2019

    work page 2019

  5. [5]

    Climate change and catastrophes,

    N. Ospanova, “Climate change and catastrophes,” Earth and Environmental Science Research and Reviews. Kansas, vol. 7, no. 1, pp. 1–10, 2024

    work page 2024

  6. [6]

    Modeling recent climate change induced extreme events in bangladesh: A review,

    M. R. Dastagir, “Modeling recent climate change induced extreme events in bangladesh: A review,”Weather and Climate Extremes, vol. 7, pp. 49–60, 2015

    work page 2015

  7. [7]

    How climate change affects extreme weather events,

    P. Stott, “How climate change affects extreme weather events,”Science, vol. 352, no. 6293, pp. 1517–1518, 2016

    work page 2016

  8. [8]

    Data and research to inform global policy: The global compact for safe, orderly and regular migration,

    E. P. Kraly and B. Hovy, “Data and research to inform global policy: The global compact for safe, orderly and regular migration,” Comparative Migration Studies, vol. 8, no. 1, p. 11, 2020

    work page 2020

Show all 34 references
  1. [9]

    Laczko, S

    F. Laczko, S. Hieronimus, and L. Hartenstein, More than numbers: How migration data can deliver real-life benefits for migrants and governments, 2018

    2018

  2. [10]

    Measuring global migration flows using online data,

    G. Chi, G. J. Abel, D. Johnston, E. Giraudy, and M. Bailey, “Measuring global migration flows using online data,” Proceedings of the National Academy of Sciences , vol. 122, no. 18, e2409418122, May 2025. DOI: 10.1073/pnas.2409418122. (visited on 06/16/2025)

    2025 doi

  3. [11]

    Gravity models: A tool for migration analysis,

    R. Ramos, “Gravity models: A tool for migration analysis,” IZA World of Labor, 2016

    2016

  4. [12]

    A universal model for mobility and migration patterns,

    F. Simini, M. C. González, A. Maritan, and A. -L. Barabási, “A universal model for mobility and migration patterns,” Nature, vol. 484, no. 7392, pp. 96–100, 2012

    2012

  5. [13]

    Gravity models do not explain, and cannot predict, international migration dynamics,

    R. M. Beyer, J. Schewe, and H. Lotze-Campen, “Gravity models do not explain, and cannot predict, international migration dynamics,” Humanities and Social Sciences Communications, vol. 9, no. 1, pp. 1–10, 2022

    2022

  6. [14]

    Migrability: A diffusion model of migration,

    F. Rajulton, “Migrability: A diffusion model of migration,” Genus, pp. 31–48, 1991

    1991

  7. [15]

    Random walks on complex networks,

    J. D. Noh and H. Rieger, “Random walks on complex networks,” Physical review letters , vol. 92, no. 11, p. 118 701, 2004

    2004

  8. [16]

    Biased random walks in complex networks: The role of local navigation rules,

    A. Fronczak and P. Fronczak, “Biased random walks in complex networks: The role of local navigation rules,” Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, vol. 80, no. 1, p. 016 107, 2009

    2009

  9. [17]

    Quantifying ethnic segregation in cities through random walks,

    S. Sousa and V . Nicosia, “Quantifying ethnic segregation in cities through random walks,” Nature Communica- tions, vol. 13, no. 1, p. 5809, 2022

    2022

  10. [18]

    Human mobility: Models and applications,

    H. Barbosa, M. Barthelemy, G. Ghoshal, et al., “Human mobility: Models and applications,” Physics Reports, vol. 734, pp. 1–74, 2018

    2018

  11. [19]

    Multi-scale random walk models for reactive transport processes in fracture-matrix systems,

    D. Roubinet, P. Gouze, A. Puyguiraud, and M. Dentz, “Multi-scale random walk models for reactive transport processes in fracture-matrix systems,” Advances in Water Resources, vol. 164, p. 104 183, 2022

    2022

  12. [20]

    Diffusion segregation and the disproportionate incidence of covid-19 in african american communities,

    A. Bassolas, S. Sousa, and V . Nicosia, “Diffusion segregation and the disproportionate incidence of covid-19 in african american communities,” Journal of The Royal Society Interface, vol. 18, no. 174, p. 20 200 961, 2021

    2021

  13. [21]

    From random walks to epidemic spreading: Compartment model with mortality for vector transmitted diseases,

    T. Granger, T. M. Michelitsch, B. A. Collet, M. Bestehorn, and A. P. Riascos, “From random walks to epidemic spreading: Compartment model with mortality for vector transmitted diseases,”arXiv preprint arXiv:2502.08558, 2025

    2025 arXiv

  14. [22]

    Random walks and their applications: Widely used as mathematical models, random walks play an important role in several areas of physics, chemistry, and biology,

    G. H. Weiss, “Random walks and their applications: Widely used as mathematical models, random walks play an important role in several areas of physics, chemistry, and biology,”American Scientist, vol. 71, no. 1, pp. 65–71, 1983

    1983

  15. [23]

    Information dissemination via random walks in d-dimensional space,

    H. Lam, Z. Liu, M. Mitzenmacher, X. Sun, and Y . Wang, “Information dissemination via random walks in d-dimensional space,” inProceedings of the Twenty-Third Annual ACM-SIAM Symposium on Discrete Algorithms, SIAM, 2012, pp. 1612–1622. 9 Complexity72h 23-27 J UNE 2025 - M ADRID

    2012

  16. [24]

    Maps of random walks on complex networks reveal community structure,

    M. Rosvall and C. T. Bergstrom, “Maps of random walks on complex networks reveal community structure,” Proceedings of the National Academy of Sciences, vol. 105, no. 4, pp. 1118–1123, Jan. 2008, ISSN : 0027-8424, 1091-6490. DOI: 10.1073/pnas.0706851105. (visited on 03/05/2021)

    2008 doi

  17. [25]

    International Organization for Migration, Prepositioning relief: Using a random walk model to predict the distribution of idps in nigeria , https : / / dtm . iom . int / content / prepositioning - relief - using - random-walk-model-predict-distribution-idps-nigeria , Accessed:...

    2025

  18. [26]

    The random walk model of human migration,

    N. Yasuda, “The random walk model of human migration,” Theoretical Population Biology, vol. 7, no. 2, pp. 156–167, 1975

    1975

  19. [27]

    The nexus of climate change, food insecurity, and conflict in somalia: A comprehensive analysis of multifaceted challenges and resilience strategies,

    M. M. Ahmed, H. A. Asowe, N. I. Dirie, et al., “The nexus of climate change, food insecurity, and conflict in somalia: A comprehensive analysis of multifaceted challenges and resilience strategies,”F1000Research, vol. 13, p. 913, 2024

    2024

  20. [28]

    Large weather and conflict effects on internal displacement in somalia with little evidence of feedback onto conflict,

    L. Thalheimer, M. P. Schwarz, and F. Pretis, “Large weather and conflict effects on internal displacement in somalia with little evidence of feedback onto conflict,” Global Environmental Change, vol. 79, p. 102 641, 2023

    2023

  21. [29]

    24, 2025

    International Organization for Migration (IOM), Somalia — emergency trend tracking dataset — (since february, 2025), Jun. 24, 2025. [Online]. Available: https://dtm.iom.int/datasets/somalia-emergency-trend- tracking-dataset-february-2025

    2025

  22. [30]

    Newman, Networks

    M. Newman, Networks. Oxford University Press, 2018

    2018

  23. [31]

    The diaspora model for human migration,

    R. Prieto-Curiel, O. Ali, E. Dervi´c, et al., “The diaspora model for human migration,” PNAS Nexus, vol. 3, no. 5, pgae178, May 2024, ISSN : 2752-6542. DOI: 10.1093/pnasnexus/pgae178 . eprint: https://academic. oup.com/pnasnexus/article-pdf/3/5/pgae178/58202258/pgae178.pdf

    2024 doi

  24. [32]

    A systematic review of climate migration research: Gaps in existing literature,

    R. C. Ghosh and C. Orchiston, “A systematic review of climate migration research: Gaps in existing literature,” SN Social Sciences, vol. 2, no. 5, p. 47, 2022

    2022

  25. [33]

    How do environmental stressors influence migration? a meta-regression analysis of environmental migration literature,

    S. Zhou and G. Chi, “How do environmental stressors influence migration? a meta-regression analysis of environmental migration literature,” Demographic research, vol. 50, p. 41, 2024

    2024

  26. [34]

    Forced migration, social cohesion and conflict,

    E. Albarosa and B. Elsner, “Forced migration, social cohesion and conflict,” 2022. 10 Complexity72h 23-27 J UNE 2025 - M ADRID S1 Supplementary Figures Figure S1: Reasons of displacement for 71 districts in Somalia. Districts are presented in decreasing order of number of reco...

    2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.