Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Uncovering Insights of Compound Flooding with Data-Driven AI

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper claims that South Florida's compound flooding is driven more by groundwater saturation and the spatial state of neighboring monitoring stations than by immediate rainfall or long histories, and it demonstrates this with…

desk verdict SF2Bench is a valuable new dataset, but the paper's three scientific findings are not supported by its own ablation tables and are internally contradicted by Q4. read the letter →

arxiv 2506.04281 v3 pith:3ILZPFR2 submitted 2025-06-04 cs.LG

Xu Zheng , Chaohao Lin , Sipeng Chen , Zhuomin Chen , Jimeng Shi , Jayantha Obeysekera , Jingchao Ni , Wei Cheng
show 2 more authors
Jason Liu Dongsheng Luo
This is my paper · ORCID
classification cs.LG
keywords compoundfloodingSouthFloridagroundwaterdata-drivenforecastingfeatureablationspatialcontextextremeeventsSF2Bench
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

This paper tries to establish that compound flooding in South Florida is governed less by the familiar rain-driven sequence and more by the coupled state of the surrounding hydrological system, particularly how saturated the ground already is. It builds SF2Bench, an hourly dataset spanning 1985 to 2024 from 2,452 stations that record water stage, groundwater, rainfall, pump activity, and gate operations, and then uses a suite of forecasting architectures as analytical probes. The reported results are that removing groundwater degrades forecasts more than removing rainfall, that adding neighboring stations improves accuracy until a finite radius is reached, and that extending the temporal look-back window gives diminishing returns during extremes. If these findings hold, flood forecasting and early-warning design for porous coastal regions should re-center around subsurface storage and spatial coupling rather than rainfall intensity and long sequences alone.

What carries the argument

The central instrument is SF2Bench, a multi-modal, hourly aligned dataset covering 2,452 stations with water stage, groundwater, rainfall, pump, and gate observations across a 67,349 square kilometer region from 1985 to 2024, organized into eight temporal splits. The central mechanism is the analytical probe protocol: trained forecasting models are treated as instruments, and their error changes under controlled input removal, spatial-radius expansion, and look-back-window variation are read as evidence about which physical factors matter. The tail-focused SEDI metric is the third component, used to separate extreme-event dependence from average accuracy.

What would settle it

Run the same ablation with shuffled or time-lagged groundwater inputs: if permuted groundwater degrades forecasts as much as true groundwater does, or if removing groundwater changes MAE by less than run-to-run variability, then groundwater dominance is an artifact. A stronger check would compare the ablation ordering to a physically based water-budget simulation of the same period, asking whether measured flood severity drops as the ablation ranking predicts when groundwater storage is depleted.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the state variables describing system saturation, especially groundwater stage, carry more predictive and causal weight for flood severity in this managed, porous coastal region than the immediate meteorological forcing, and that the spatial configuration of nearby monitoring stations supplies context that longer temporal histories cannot. This is established through feature ablation (removing rainfall, groundwater, and human-control inputs, alone and in pairs), spatial expansion (growing the input radius around an anchor area), and temporal-window experiments across multiple model families. The paper also finds a decoupling between average-error improvements and extreme-event skill, with some models performing well on the tail-sensitive SEDI metric while lagging on MAE and MSE. The overall conclusion is that compound flooding here is better described as a spatially coupled system state than as a long temporal sequence.

Load-bearing premise

The central findings treat a forecasting model's loss of accuracy when an input is removed as proof of that input's physical importance in causing floods; if ablation sensitivity is not a faithful proxy for physical causality, the groundwater-dominance and spatial-context conclusions are not established.

Editorial extensions

If this is right

  • Forecasting systems for porous coastal regions should include groundwater stage as a first-class input, since removing it consistently degrades predictions more than removing rainfall does.
  • Spatial context from nearby monitoring stations within a finite radius should be built into models, because error drops as that context is added and saturates once the local catchment is covered.
  • Extending the look-back history beyond about one day gives diminishing returns for extreme events, so observational effort is better spent widening spatial coverage than lengthening history.
  • Benchmarks for compound-flood models should report tail-focused metrics such as SEDI alongside MAE and MSE, because average-error gains do not guarantee better extreme-event prediction.
  • Channel-dependent architectures are better probes for multi-factor compound-flood questions than channel-independent ones, since they show the largest error reductions when auxiliary factors are added.

Reading between the lines

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

  • If groundwater dominance holds up under causal validation, groundwater-well density becomes a priority investment for coastal flood early warning in karst and sandy regions, not just rain-gauge coverage.
  • The finite effective radius result implies spatially aware flood models could be structured by hydrological proximity with a cutoff rather than fully connected graphs; this is a testable architectural consequence the paper does not itself implement.
  • The same probe protocol could be carried to other compound hazards, such as storm surge with river discharge, rain-on-snow, or urban flash floods, wherever multi-factor monitoring networks exist, to rank driver importance before building predictive models.
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 / 5 minor

Summary. The paper introduces SF2Bench, a large observational dataset for compound flooding in South Florida, and benchmarks 14 time-series forecasting models across MLP, CNN, RNN, GNN, Transformer, and LLM categories. The authors use feature-ablation and context-scaling experiments to draw three advertised conclusions: (i) temporal-only models fail to capture multi-factor interactions; (ii) groundwater stage is a dominant predictor, often outweighing rainfall; and (iii) spatial context from nearby stations is more important than long temporal history. The paper also includes an extreme-event metric (SEDI) and a Hurricane Ian case study. The dataset and code are publicly released.

Significance. If the advertised findings were established, the paper would provide a useful counterpoint to rain-centric and sequence-only flood forecasting, and the public dataset would be a community resource for compound-flood research. The paper explicitly credits the public dataset/code release, includes a wide span of architectures, and makes a good-faith attempt to evaluate tail behavior. However, the central causal claims are not supported by the evidence actually reported: the Q3 ablation table is inconsistent across models and within rounding noise for several entries, the ablation-to-causality proxy is unvalidated, and the Q4 result directly contradicts the abstract's spatial-dominance claim. The scientific contribution therefore currently rests on load-bearing conclusions that the manuscript's own numbers undermine.

major comments (4)
  1. [Section 4.2, Q3 and Table 4] The claim that groundwater 'consistently emerges as the most informative auxiliary factor' is not supported by Table 4, which is the only ablation table and is reported for a single split (S6). For iTransformer the MAE change on removing groundwater is +0.0001 (0.1406→0.1407); for PatchTST it is +0.0015 (0.1376→0.1391); for TSMixer removing groundwater improves MAE (0.1596→0.1419); and for TimesNet it also improves MAE (0.1642→0.1599). Similarly, removing rainfall improves NLinear (0.1546→0.1483) and leaves iTransformer essentially unchanged (0.1406→0.1406). Only PatchTST and NLinear move in the advertised direction, and the deltas for the transformer-based models are within rounding/noise. No confidence intervals, no repeated-seed variance, and no aggregation across the eight splits are provided. This is the core evidence for finding (ii), so the headline claim is unsupported.
  2. [Section 3.3, Data-Driven AI Analysis] The paper frames data-driven models as 'analytical probes' and interprets leave-one-input-out ablation differences as physical causal importance, but this proxy is never validated. Because water stage and groundwater are strongly correlated (Figure 3c/d), a collinear but physically important driver can be dropped with little or no accuracy loss, so ablation sensitivity is not a consistent estimator of causal contribution. The paper offers no synthetic experiments with known ground-truth drivers, no comparison against a physically based benchmark, and no statistical test to distinguish genuine sensitivity from noise. Every Q3 and Q4 causal statement inherits this unvalidated assumption.
  3. [Section 4.2, Q4 and Figure 4c] There is a direct internal contradiction between the advertised findings and the results section. The abstract and Introduction state that 'the spatial state of surrounding monitoring stations ... provides critical causal context' and that compound flooding is 'governed more by spatially coupled system states than by long-term temporal dependencies.' However, Figure 4c and the text in Q4 conclude that 'the temporal context is the primary driver of accuracy,' with spatial benefits (Figure 4b) saturating quickly. The paper needs either to reconcile these operationalizations or to revise the abstract and Introduction to match the actual results.
  4. [Section 3.5, Eq. (1)] The metric defined in Eq. (1) is not the Symmetric Extremal Dependence Index (SEDI). SEDI is a skill score based on the hit rate and false-alarm rate, typically written as a log-odds ratio; Eq. (1) instead computes the fraction of observed extremes that are correctly predicted. Consequently, the extreme-event conclusions in Q2 and the SEDI columns of Tables 3 and 6 are based on a mislabeled metric. Please either correct the formula and reference to the actual SEDI, or rename the metric to something like 'extreme hit proportion' and avoid comparing it to SEDI.
minor comments (5)
  1. [Section 4.3, Q5] The phrase 'vibrant behavior' appears to be a typo for 'erratic behavior' in the description of the iTransformer results during the Hurricane Ian case study.
  2. [Section 3.5 and Appendix B] Appendix B states that all reported metrics are computed on normalized data, but this fact is not stated in the main text where Table 4 is discussed; it should be moved earlier because ablation deltas of order 0.0001 are otherwise hard to interpret.
  3. [Figure 3 caption] The caption says the temporal patterns use 'data from split S5 and S7 as a representative example,' but it is unclear which panel corresponds to which split; please clarify.
  4. [Table 2] The units in Table 2 are phrased inconsistently ('Feet Water Stage', 'Stage of Groundwater', 'Inches Rainfall'); please use a consistent notation (e.g., ft, in, RPM, dimensionless).
  5. [Section 4.2, Q4, Figure 4a] The spatial ablation is defined by 'an anchor area of radius R' that is expanded by scale factors 1.0 to 1.8, but R itself is never specified; please report the actual radii in kilometers or degrees.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the paper is an empirical model-ablation study whose headline claims are reports of measured forecast behavior, not consequences entailed by the definitions or by self-citations.

full rationale

The paper contains no derivation chain that reduces to its own inputs. The central findings about groundwater influence, spatial context, and temporal context are inferred from controlled forecasting experiments (Table 4, Figure 4) in which models are trained on historical observations and evaluated on temporally disjoint test periods. The claim that groundwater is a 'dominant predictor' is an empirical statement about measured ablation sensitivity, not a tautology: the operational definition of informativeness (drop in MAE/MSE when a factor is removed) is stated in Section 3.3, but the numerical outcomes are contingent on the data and models, so the finding is not forced by construction. Similarly, the spatial-context and temporal-context conclusions summarize observed error curves rather than being encoded in the experimental setup. The few self-citations involving the authors, most notably [27] (Jane, Cadavid, Obeysekera, Wahl) used to motivate South Florida as a compound-flooding study region and [60] (Shi et al.) cited when discussing existing datasets, are background motivation and are not load-bearing for any claimed result; neither is invoked as a uniqueness theorem or as a justification that forbids alternative interpretations. Concerns raised in review, such as the small magnitude of ablation deltas, the lack of confidence intervals, the inconsistency across models in Table 4, and the apparent tension between the abstract and the Q4 temporal-context conclusion, are evidentiary and soundness issues about whether the measurements support the causal language, not instances of circular reasoning. Accordingly, the circularity score is low.

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

The central claims about groundwater dominance and spatial-over-temporal context rest on two proxies: ablation-based importance and a non-standard extreme-event hit-rate metric. Neither is validated against an external physical benchmark, and effect sizes are near noise.

assumptions (3)
  • domain assumption Change in forecasting error under input ablation is a valid proxy for the physical causal importance of the ablated driver.
    Used to convert benchmark ablations into scientific findings (ii) and (iii). Declared in Section 3.3 ('analytical probes') but never validated against independent physical evidence.
  • domain assumption Missing rainfall and hydraulic-control records can be filled with zero without distorting the system state.
    Section 3.2 'Handling Missing Observations' treats missing rainfall, pump, and gate entries as absences and fills with zero; assumes sensor downtime does not coincide with active events.
  • domain assumption The co-occurrence rate defined in Eq. (1), here named SEDI, is an adequate measure of extreme-event dependence.
    Eq. (1) computes the fraction of observed extremes that are also predicted as extremes, not the standard SEDI from the cited weather literature; conclusions about model extreme-event skill rest on this non-standard metric.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Uncovering Insights of Compound Flooding with Data-Driven AI." pith.science (2026). https://pith.science/paper/3ILZPFR2

@misc{pith2026250604281,
  author       = {Pith},
  title        = {Pith review of: Uncovering Insights of Compound Flooding with Data-Driven AI},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3ILZPFR2}},
  note         = {Machine review of arXiv:2506.04281}
}
read the original abstract

Compound flooding, driven by nonlinear interactions between multiple hydrometeorological factors, poses a significant challenge to hazard prevention. Existing forecasting approaches, whether physics-based or data-driven, often emphasize temporal patterns while underexploring how multiple interacting factors jointly shape flood dynamics. To address this problem, we conduct a large-scale data-driven analysis of compound flooding in South Florida, a typical area for compound flooding, by integrating tidal conditions, rainfall, groundwater stage, and human water management activities. Our analysis reveals three key findings: (i) models that capture temporal dynamics alone fail to represent multi-factor interactions during compound events; (ii) subsurface saturation, as reflected by groundwater levels, emerges as a dominant predictor of flood severity, often outweighing immediate rainfall intensity in this porous coastal region; and (iii) the spatial state of surrounding monitoring stations within a finite effective radius provides critical causal context for flooding, while extending temporal history yields diminishing returns during extreme events. These findings suggest that compound flooding is governed more by spatially coupled system states than by long-term temporal dependencies, challenging rain-centric and sequence-dominated forecasting paradigms. By framing data-driven models as tools for scientific inquiry rather than prediction alone, this study offers new insights into the mechanisms of compound flooding and informs the design of more physically grounded early-warning systems for coastal environments. Our dataset and code are publicly available at https://github.com/AslanDing/SFBench.

Figures

Figures reproduced from arXiv: 2506.04281 by the authors.

Figure 1
Figure 1. Global historical statistics for 5 types of natural [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. (a). Location distribution of monitor stations. (b). Flood observation location distribution. In (a) and (b), we highlight [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. (a).The different interest areas with a radius for the ablation study on spatial information. (b). Study on spatial [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: (a) The interest area of the case study. (b) The com [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Multi-Source Dynamic Graph Learning for Compound-Flood Forecasting in Managed Coastal Systems

    cs.LG 2026-08 conditional novelty 6.0 of 10

    An anchored forecaster that adds bounded, regime-gated corrections from a dynamic multi-station graph to a local temporal forecast improves sustained high-water plateau prediction in South Florida without harming rout...

  2. Exploring Robust Multi-Agent Workflows for Environmental Data Management

    cs.AI 2026-04 conditional novelty 6.0 of 10

    A role-separated multi-agent workflow with deterministic validation gates blocked a coordinate-transformation error before publication and completed a 2,452-station dataset release in two days, based on two non-contro...

Reference graph

Works this paper leans on

91 extracted references · 49 canonical work pages · cited by 2 Pith papers

  1. [1]

    Nans Addor, Andrew J Newman, Naoki Mizukami, and Martyn P Clark. 2017. The CAMELS data set: catchment attributes and meteorology for large-sample studies.Hydrology and Earth System Sciences21, 10 (2017), 5293–5313

    2017

  2. [2]

    Adikari, Sangam Shrestha, Dhanika T

    Kasuni E. Adikari, Sangam Shrestha, Dhanika T. Ratnayake, Aakanchya Bud- hathoki, S. Mohanasundaram, and Matthew N. Dailey. 2021. Evaluation of artificial intelligence models for flood and drought forecasting in arid and tropical regions.Environmental Modelling & Software144 (2021), 105136. doi:10.1016/j.envsoft.2021.105136

    work page arXiv 2021

  3. [3]

    Camila Alvarez-Garreton, Pablo A Mendoza, Juan Pablo Boisier, Nans Addor, Mauricio Galleguillos, Mauricio Zambrano-Bigiarini, Antonio Lara, Cristóbal Puelma, Gonzalo Cortes, Rene Garreaud, et al. 2018. The CAMELS-CL dataset: catchment attributes and meteorology for large sample studies–Chile dataset. Hydrology and Earth System Sciences22, 11 (2018), 5817–5846

    2018

  4. [4]

    Donato Amitrano, Gerardo Di Martino, Alessio Di Simone, and Pasquale Imper- atore. 2024. Flood Detection with SAR: A Review of Techniques and Datasets. Remote Sensing16, 4 (2024). doi:10.3390/rs16040656

    doi:10.3390/rs16040656 2024

  5. [5]

    Shaojie Bai, J Zico Kolter, and Vladlen Koltun. 2018. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling.arXiv preprint arXiv:1803.01271(2018)

    arXiv 2018

  6. [6]

    Emanuele Bevacqua, Michalis I Vousdoukas, Giuseppe Zappa, Kevin Hodges, Theodore G Shepherd, Douglas Maraun, Lorenzo Mentaschi, and Luc Feyen. 2020. More meteorological events that drive compound coastal flooding are projected under climate change.Communications earth & environment1, 1 (2020), 47

    2020

  7. [7]

    Derrick Bonafilia, Beth Tellman, Tyler Anderson, and Erica Issenberg. 2020. Sen1Floods11: A georeferenced dataset to train and test deep learning flood algorithms for sentinel-1. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops. 210–211

    2020

  8. [8]

    Samuel D Brody, Sammy Zahran, Praveen Maghelal, Himanshu Grover, and Wesley E Highfield. 2007. The rising costs of floods: Examining the impact of planning and development decisions on property damage in Florida.Journal of the American Planning Association73, 3 (2007), 330–345

    2007

Show all 91 references
  1. [9]

    Gary W Brunner. 1997. HEC-RAS river analysis system. Hydraulic reference manual. Version 1.0. (1997)

    1997

  2. [10]

    Wanlin Cai, Yuxuan Liang, Xianggen Liu, Jianshuai Feng, and Yuankai Wu. 2023. MSGNet: Learning Multi-Scale Inter-Series Correlations for Multivariate Time Series Forecasting.arXiv preprint arXiv:2401.00423(2023)

    2023 arXiv

  3. [11]

    Defu Cao, Yujing Wang, Juanyong Duan, Ce Zhang, Xia Zhu, Conguri Huang, Yunhai Tong, Bixiong Xu, Jing Bai, Jie Tong, and Qi Zhang. 2020. Spectral temporal graph neural network for multivariate time-series forecasting. InProceedings of the 34th International Conference on Neura...

    2020

  4. [12]

    Vinícius BP Chagas, Pedro LB Chaffe, Nans Addor, Fernando M Fan, Ayan S Fleischmann, Rodrigo CD Paiva, and Vinícius A Siqueira. 2020. CAMELS-BR: hydrometeorological time series and landscape attributes for 897 catchments in Brazil.Earth System Science Data12, 3 (2020), 2075–2096

    2020

  5. [13]

    Shiyu Chang, Yang Zhang, Wei Han, Mo Yu, Xiaoxiao Guo, Wei Tan, Xiaodong Cui, Michael Witbrock, Mark A Hasegawa-Johnson, and Thomas S Huang. 2017. Dilated recurrent neural networks.Advances in neural information processing systems30 (2017)

    2017

  6. [14]

    Si-An Chen, Chun-Liang Li, Sercan O Arik, Nathanael Christian Yoder, and Tomas Pfister. 2023. TSMixer: An All-MLP Architecture for Time Series Forecast- ing.Transactions on Machine Learning Research(2023). https://openreview.net/ forum?id=wbpxTuXgm0

    2023

  7. [15]

    Mingyue Cheng, Jiqian Yang, Tingyue Pan, Qi Liu, Zhi Li, and Shijin Wang. 2025. Convtimenet: A deep hierarchical fully convolutional model for multivariate time series analysis. InCompanion Proceedings of the ACM on Web Conference

    2025

  8. [16]

    Gemma Coxon, Nans Addor, John P Bloomfield, Jim Freer, Matt Fry, Jamie Han- naford, Nicholas JK Howden, Rosanna Lane, Melinda Lewis, Emma L Robinson, et al. 2020. CAMELS-GB: hydrometeorological time series and landscape at- tributes for 671 catchments in Great Britain.Earth Sy...

    2020

  9. [17]

    Damien Delforge, Valentin Wathelet, Regina Below, Cinzia Lanfredi Sofia, Margo Tonnelier, Joris AF van Loenhout, and Niko Speybroeck. 2025. EM-DAT: the emergency events database.International Journal of Disaster Risk Reduction (2025), 105509

    2025

  10. [18]

    Dottori, L

    F. Dottori, L. Alfieri, A. Bianchi, J. Skoien, and P. Salamon. 2022. A new dataset of river flood hazard maps for Europe and the Mediterranean Basin.Earth System Science Data14, 4 (2022), 1549–1569. doi:10.5194/essd-14-1549-2022

    2022 doi

  11. [19]

    K. J. A. Fowler, S. C. Acharya, N. Addor, C. Chou, and M. C. Peel. 2021. CAMELS- AUS: hydrometeorological time series and landscape attributes for 222 catchments in Australia.Earth System Science Data13, 8 (2021), 3847–3867. doi:10.5194/essd- 13-3847-2021

    2021 doi

  12. [20]

    Jean-Yves Franceschi, Aymeric Dieuleveut, and Martin Jaggi. 2019. Unsupervised scalable representation learning for multivariate time series.Advances in neural information processing systems32 (2019)

    2019

  13. [21]

    Joshua Green, Ivan D Haigh, Niall Quinn, Jeff Neal, Thomas Wahl, Melissa Wood, Dirk Eilander, Marleen de Ruiter, Philip Ward, and Paula Camus. 2024. A comprehensive review of coastal compound flooding literature.arXiv preprint arXiv:2404.01321(2024)

    2024 arXiv

  14. [22]

    Nate Gruver, Marc Finzi, Shikai Qiu, and Andrew Gordon Wilson. 2023. Large language models are zero-shot time series forecasters. InProceedings of the 37th International Conference on Neural Information Processing Systems(New Orleans, LA, USA). Curran Associates Inc., Red Hook...

    2023

  15. [23]

    Tao Han, Zhenghao Chen, Song Guo, Wanghan Xu, and Lei Bai. 2024. Cra5: Extreme compression of era5 for portable global climate and weather research via an efficient variational transformer.arXiv preprint arXiv:2405.03376(2024)

    2024 arXiv

  16. [24]

    Tao Han, Song Guo, Zhenghao Chen, Wanghan Xu, and Lei Bai. 2024. How far are today’s time-series models from real-world weather forecasting applications? arXiv preprint arXiv:2406.14399(2024)

    2024 arXiv

  17. [25]

    Yukiko Hirabayashi, Roobavannan Mahendran, Sujan Koirala, Lisako Konoshima, Dai Yamazaki, Satoshi Watanabe, Hyungjun Kim, and Shinjiro Kanae. 2013. Global flood risk under climate change.Nature climate change3, 9 (2013), 816–821

    2013

  18. [26]

    Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory.Neural computation9, 8 (1997), 1735–1780

    1997

  19. [27]

    R. Jane, L. Cadavid, J. Obeysekera, and T. Wahl. 2020. Multivariate statistical modelling of the drivers of compound flood events in south Florida.Natural Hazards and Earth System Sciences20, 10 (2020), 2681–2699. doi:10.5194/nhess- 20-2681-2020

    2020 doi

  20. [28]

    Sheo Yon Jhin, Seojin Kim, and Noseong Park. 2024. Addressing Prediction Delays in Time Series Forecasting: A Continuous GRU Approach with Derivative Regularization. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 1234–1245

    2024

  21. [29]

    Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, and Qingsong Wen

    Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y. Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, and Qingsong Wen

  22. [30]

    Sebastiaan N Jonkman and Johannes K Vrijling. 2008. Loss of life due to floods. Journal of flood risk management1, 1 (2008), 43–56

    2008

  23. [31]

    Syed Kabir, Sandhya Patidar, Xilin Xia, Qiuhua Liang, Jeffrey Neal, and Gareth Pender. 2020. A deep convolutional neural network model for rapid prediction of fluvial flood inundation.Journal of Hydrology590 (2020), 125481

    2020

  24. [32]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. InInternational Conference on Learning Repre- sentations. https://openreview.net/forum?id=SJU4ayYgl

    2017

  25. [33]

    Nikolas Kirschstein and Yixuan Sun. 2024. The Merit of River Network Topology for Neural Flood Forecasting. InInternational Conference on Machine Learning. PMLR, 24713–24725

    2024

  26. [34]

    Christoph Klingler, Karsten Schulz, and Mathew Herrnegger. 2021. Lamah| large- sample data for hydrology and environmental sciences for central europe.Earth System Science Data Discussions2021 (2021), 1–46. KDD ’26, August 09–13, 2026, Jeju Island, Republic of Korea Xu Zheng et al

    2021

  27. [35]

    Kratzert, D

    F. Kratzert, D. Klotz, C. Brenner, K. Schulz, and M. Herrnegger. 2018. Rainfall– runoff modelling using Long Short-Term Memory (LSTM) networks.Hydrology and Earth System Sciences22, 11 (2018), 6005–6022. doi:10.5194/hess-22-6005-2018

    2018 doi

  28. [36]

    Guokun Lai, Wei-Cheng Chang, Yiming Yang, and Hanxiao Liu. 2018. Modeling long-and short-term temporal patterns with deep neural networks. InThe 41st international ACM SIGIR conference on research & development in information retrieval. 95–104

    2018

  29. [37]

    Yann LeCun, Léon Bottou, Yoshua Bengio, and Patrick Haffner. 1998. Gradient- based learning applied to document recognition.Proc. IEEE86, 11 (1998), 2278– 2324

    1998

  30. [38]

    Shengsheng Lin, Weiwei Lin, Xinyi Hu, Wentai Wu, Ruichao Mo, and Haocheng Zhong. 2024. CycleNet: Enhancing Time Series Forecasting through Modeling Periodic Patterns. InAdvances in Neural Information Processing Systems, A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Pa- quet...

    2024

  31. [39]

    Shengsheng Lin, Weiwei Lin, Wentai Wu, Haojun Chen, and Junjie Yang. 2024. SparseTSF: Modeling Long-term Time Series Forecasting with *1k* Parameters. In Proceedings of the 41st International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 235), ...

    2024

  32. [40]

    Liu, and Schahram Dustdar

    Shizhan Liu, Hang Yu, Cong Liao, Jianguo Li, Weiyao Lin, Alex X. Liu, and Schahram Dustdar. 2022. Pyraformer: Low-Complexity Pyramidal Attention for Long-Range Time Series Modeling and Forecasting. InInternational Conference on Learning Representations. https://openreview.net/...

    2022

  33. [41]

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. 2024. iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. InThe Twelfth International Conference on Learning Representations. https://openreview.net/forum?id=JePfAI8fah

    2024

  34. [42]

    Yijing Liu, Qinxian Liu, Jian-Wei Zhang, Haozhe Feng, Zhongwei Wang, Zihan Zhou, and Wei Chen. 2022. Multivariate Time-Series Forecast- ing with Temporal Polynomial Graph Neural Networks. InAdvances in Neural Information Processing Systems, S. Koyejo, S. Mohamed, A. Agar- wal,...

    2022

  35. [43]

    Yong Liu, Guo Qin, Xiangdong Huang, Jianmin Wang, and Mingsheng Long

  36. [44]

    Ilya Loshchilov and Frank Hutter. 2019. Decoupled Weight Decay Regularization. InInternational Conference on Learning Representations. https://openreview.net/ forum?id=Bkg6RiCqY7

    2019

  37. [45]

    InThe Thirty-eighth Annual Conference on Neural Information Processing Systems

    AutoTimes: Autoregressive Time Series Forecasters via Large Language Models. InThe Thirty-eighth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id=FOvZztnp1H

  38. [46]

    Fabio Montello, Edoardo Arnaudo, and Claudio Rossi. 2022. MMFlood: A Mul- timodal Dataset for Flood Delineation From Satellite Imagery.IEEE Access10 (2022), 96774–96787. doi:10.1109/ACCESS.2022.3205419

    2022

  39. [47]

    Donghao Luo and Xue Wang. 2024. Moderntcn: A modern pure convolution structure for general time series analysis. InThe twelfth international conference on learning representations. 1–43

    2024

  40. [48]

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2023. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. InThe Eleventh International Conference on Learning Representations. https: //openreview.net/forum?id=Jbdc0vTOcol

    2023

  41. [49]

    Grey Nearing, Deborah Cohen, Vusumuzi Dube, Martin Gauch, Oren Gilon, Shaun Harrigan, Avinatan Hassidim, Daniel Klotz, Frederik Kratzert, Asher Metzger, et al. 2024. Global prediction of extreme floods in ungauged watersheds.Nature 627, 8004 (2024), 559–563

    2024

  42. [50]

    Zijie Pan, Yushan Jiang, Sahil Garg, Anderson Schneider, Yuriy Nevmyvaka, and Dongjin Song. 2024. 2IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting. InForty-first International Conference on Machine Learning

    2024

  43. [51]

    2025.Resiliency and Flood Protection

    Office of Resilience, South Florida Water Management District. 2025.Resiliency and Flood Protection. West Palm Beach, Florida. https://www.sfwmd.gov/our- work/resiliency-and-flood-protection

    2025

  44. [52]

    Katerina Papagiannaki, Olga Petrucci, Michalis Diakakis, Vassiliki Kotroni, Luigi Aceto, Cinzia Bianchi, Rudolf Brázdil, Miquel Grimalt Gelabert, Moshe Inbar, Abdullah Kahraman, et al. 2022. Developing a large-scale dataset of flood fatalities for territories in the Euro-Medit...

    2022

  45. [53]

    Claudio Paniconi and Mario Putti. 2015. Physically based modeling in catchment hydrology at 50: Survey and outlook.Water Resources Research51, 9 (2015), 7090–7129

    2015

  46. [54]

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. 2019. Language models are unsupervised multitask learners.OpenAI blog 1, 8 (2019), 9

    2019

  47. [55]

    Yao Qin, Dongjin Song, Haifeng Chen, Wei Cheng, Guofei Jiang, and Garrison W Cottrell. 2017. A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction. InProceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence. Internati...

    2017

  48. [56]

    Jannatul Ferdous Ruma, Mohammed Sarfaraz Gani Adnan, Ashraf Dewan, and Rashedur M. Rahman. 2023. Particle swarm optimization based LSTM networks for water level forecasting: A case study on Bangladesh river network.Results in Engineering17 (2023), 100951. doi:10.1016/j.rineng....

    2023

  49. [57]

    Maryam Rahnemoonfar, Tashnim Chowdhury, Argho Sarkar, Debvrat Varshney, Masoud Yari, and Robin Roberson Murphy. 2021. FloodNet: A High Resolution Aerial Imagery Dataset for Post Flood Scene Understanding.IEEE Access9 (2021), 89644–89654. doi:10.1109/ACCESS.2021.3090981

    2021

  50. [58]

    Christopher C Sampson, Andrew M Smith, Paul D Bates, Jeffrey C Neal, Lorenzo Alfieri, and Jim E Freer. 2015. A high-resolution global flood hazard model.Water resources research51, 9 (2015), 7358–7381

    2015

  51. [59]

    David Salinas, Valentin Flunkert, Jan Gasthaus, and Tim Januschowski. 2020. DeepAR: Probabilistic forecasting with autoregressive recurrent networks.Inter- national journal of forecasting36, 3 (2020), 1181–1191

    2020

  52. [60]

    Jimeng Shi, Zeda Yin, Arturo Leon, Jayantha Obeysekera, and Giri Narasimhan

  53. [61]

    Antonia Sebastian. 2022. Chapter 7 - Compound flooding. InCoastal Flood Risk Reduction, Samuel Brody, Yoonjeong Lee, and Baukje Bee Kothuis (Eds.). Elsevier, 77–88. doi:10.1016/B978-0-323-85251-7.00007-X

    2022 doi

  54. [62]

    Thomas Wahl, Shaleen Jain, Jens Bender, Steven D Meyers, and Mark E Luther

  55. [63]

    Huiqiang Wang, Jian Peng, Feihu Huang, Jince Wang, Junhui Chen, and Yifei Xiao

  56. [64]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

    2017

  57. [65]

    Shiyu Wang, Jiawei LI, Xiaoming Shi, Zhou Ye, Baichuan Mo, Wenze Lin, Ju Shengtong, Zhixuan Chu, and Ming Jin. 2025. TimeMixer++: A General Time Series Pattern Machine for Universal Predictive Analysis. InThe Thirteenth International Conference on Learning Representations. htt...

    2025

  58. [66]

    Zhang, and JUN ZHOU

    Shiyu Wang, Haixu Wu, Xiaoming Shi, Tengge Hu, Huakun Luo, Lintao Ma, James Y. Zhang, and JUN ZHOU. 2024. TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting. InThe Twelfth International Conference on Learning Representations. https://openreview.net/forum?id=...

    2024

  59. [67]

    Yuxuan Wang, Haixu Wu, Jiaxiang Dong, Guo Qin, Haoran Zhang, Yong Liu, Yunzhong Qiu, Jianmin Wang, and Mingsheng Long. 2024. TimeXer: Empowering Transformers for Time Series Forecasting with Exogenous Variables. InThe Thirty-eighth Annual Conference on Neural Information Proce...

    2024

  60. [68]

    Oliver EJ Wing, William Lehman, Paul D Bates, Christopher C Sampson, Niall Quinn, Andrew M Smith, Jeffrey C Neal, Jeremy R Porter, and Carolyn Kousky

  61. [69]

    Jingyuan Wang, Ze Wang, Jianfeng Li, and Junjie Wu. 2018. Multilevel wavelet decomposition network for interpretable time series analysis. InProceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining. 2437–2446

    2018

  62. [70]

    Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. 2021. Autoformer: decomposition transformers with auto-correlation for long-term series forecast- ing. InProceedings of the 35th International Conference on Neural Information Processing Systems. 22419–22430

    2021

  63. [71]

    Zonghan Wu, Shirui Pan, Guodong Long, Jing Jiang, Xiaojun Chang, and Chengqi Zhang. 2020. Connecting the dots: Multivariate time series forecasting with graph neural networks. InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining. 753–763

    2020

  64. [72]

    Zonghan Wu, Shirui Pan, Guodong Long, Jing Jiang, and Chengqi Zhang. 2019. Graph wavenet for deep spatial-temporal graph modeling. InProceedings of the 28th International Joint Conference on Artificial Intelligence. 1907–1913

    2019

  65. [73]

    Kui Xu, Yunchao Zhuang, Lingling Bin, Chenyue Wang, and Fuchang Tian. 2023. Impact assessment of climate change on compound flooding in a coastal city. Journal of Hydrology617 (2023), 129166

    2023

  66. [74]

    Qingsong Xu, Yilei Shi, Jie Zhao, and Xiao Xiang Zhu. 2025. FloodCastBench: A large-scale dataset and foundation models for flood modeling and forecasting. Scientific Data12, 1 (2025), 431. Uncovering Insights of Compound Flooding with Data-Driven AI KDD ’26, August 09–13, 202...

    2025

  67. [75]

    Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. 2023. TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis. InThe Eleventh International Conference on Learning Representations. https://openreview.net/forum?id=ju_Uqw384Oq

    2023

  68. [76]

    Kun Yi, Qi Zhang, Wei Fan, Hui He, Liang Hu, Pengyang Wang, Ning An, Long- bing Cao, and Zhendong Niu. 2023. FourierGNN: rethinking multivariate time series forecasting from a pure graph perspective. InProceedings of the 37th Inter- national Conference on Neural Information Pr...

    2023

  69. [77]

    Jie Yin, Yao Gao, Ruishan Chen, Dapeng Yu, Robert Wilby, Nigel Wright, Yong Ge, Jeremy Bricker, Huili Gong, and Mingfu Guan. 2023. Flash floods: why are more of them devastating the world’s driest regions?Nature615, 7951 (2023), 212–215

    2023

  70. [78]

    Zeda Yin, Linglong Bian, Beichao Hu, Jimeng Shi, and Arturo S Leon. 2023. Physic-informed neural network approach coupled with boundary conditions for solving 1D steady shallow water equations for riverine system. InWorld Environmental and Water Resources Congress 2023. 280–288

    2023

  71. [79]

    Zhihan Yue, Yujing Wang, Juanyong Duan, Tianmeng Yang, Congrui Huang, Yunhai Tong, and Bixiong Xu. 2022. Ts2vec: Towards universal representation of time series. InProceedings of the AAAI conference on artificial intelligence, Vol. 36. 8980–8987

    2022

  72. [80]

    Shuaihong Zang, Zhijia Li, Ke Zhang, Cheng Yao, Zhiyu Liu, Jingfeng Wang, Yingchun Huang, and Sheng Wang. 2021. Improving the flood prediction ca- pability of the Xin’anjiang model by formulating a new physics-based routing framework and a key routing parameter estimation meth...

    2021

  73. [81]

    Wanghan Xu, Kang Chen, Tao Han, Hao Chen, Wanli Ouyang, and Lei Bai. 2024. Extremecast: Boosting extreme value prediction for global weather forecast.arXiv preprint arXiv:2402.01295(2024)

    2024 arXiv

  74. [82]

    Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. 2021. Informer: Beyond efficient transformer for long se- quence time-series forecasting. InProceedings of the AAAI conference on artificial intelligence, Vol. 35. 11106–11115

    2021

  75. [83]

    Tian Zhou, Ziqing Ma, Qingsong Wen, Xue Wang, Liang Sun, and Rong Jin. 2022. Fedformer: Frequency enhanced decomposed transformer for long-term series forecasting. InInternational conference on machine learning. PMLR, 27268–27286

    2022

  76. [84]

    Tian Zhou, Peisong Niu, Xue Wang, Liang Sun, and Rong Jin. 2023. One fits all: power general time series analysis by pretrained LM. InProceedings of the 37th International Conference on Neural Information Processing Systems(New Orleans, LA, USA). Curran Associates Inc., Red Ho...

    2023

  77. [87]

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. 2023. Are transformers effective for time series forecasting?. InProceedings of the AAAI conference on artificial intelligence, Vol. 37. 11121–11128

    2023

  78. [91]

    ModernTCN[ 45]

    The learning rate is1× 10−3, weight decay is1× 10−7, batch size is 256, and training is performed for 50 epochs. ModernTCN[ 45]. ModernTCN introduces a streamlined, fully convolutional architecture that aims to simplify design while en- hancing performance. It incorporates com...

    2026

  79. [2015]

    Increasing risk of compound flooding from storm surge and rainfall for major US cities.Nature Climate Change5, 12 (2015), 1093–1097

    2015

  80. [2022]

    Inequitable patterns of US flood risk in the Anthropocene.Nature Climate Change12, 2 (2022), 156–162

    2022

  81. [2023]

    InThe eleventh international conference on learning representations

    Micn: Multi-scale local and global context modeling for long-term series forecasting. InThe eleventh international conference on learning representations

  82. [2024]

    InThe Twelfth International Conference on Learning Representations

    Time-LLM: Time Series Forecasting by Reprogramming Large Language Models. InThe Twelfth International Conference on Learning Representations. https://openreview.net/forum?id=Unb5CVPtae

  83. [2025]

    In Proceedings of the AAAI Conference on Artificial Intelligence, Vol

    Fidlar: Forecast-informed deep learning architecture for flood mitigation. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 28377–28385

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.