Pith. sign in

REVIEW 4 major objections 5 minor 39 references

Spatio-temporal Causal Learning for Streamflow Forecasting

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper reports that feeding a river flow graph into a graph neural network improves streamflow forecasts at all horizons on 73 Texas stations, with short-range NSE rising from 0.78 to 0.86.

desk verdict A sensible composite model with an honest ablation, but the missing temporal train/test split makes the headline performance claim unverifiable as written. read the letter →

arxiv 2411.17937 v1 pith:7QEZO5JR submitted 2024-11-26 cs.LG cs.AI

Shu Wan , Reepal Shah , Qi Deng , John Sabo , Huan Liu , K. Selçuk This is my paper · ORCID
classification cs.LGcs.AI
keywords streamflowforecastingspatio-temporalgraphneuralnetworkcausalriverflowrunoffembeddingvariationalautoencoderhydrologicconnectivityBrazosbasin
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

Streamflow forecasts matter for flood warnings, reservoir operations, and water allocation, but standard graph neural networks treat nearby stations symmetrically and can let downstream information leak into upstream predictions. This paper tries to fix that by using a river flow graph—an explicit map of which stations are upstream of which—as a causal mask inside a spatio-temporal graph neural network. The proposed model, Causal Streamflow Forecasting (CSF), reports the best scores among all tested methods on 73 Brazos River basin stations for short, medium, and long forecast horizons. If the result holds, it gives hydrologists a practical way to inject known physical connectivity into deep learning without giving up computational speed.

What carries the argument

The load-bearing object is the river flow graph, a directed adjacency matrix derived from topographic and watershed-boundary data that records which stations are upstream of which. CSF feeds this graph into a two-stage model: a Variational Autoencoder at each station learns a latent 'runoff embedding' from meteorological and static forcing variables, and a Spatio-Temporal Graph Convolutional Network uses the river flow graph as a causal mask so each station aggregates information only from causally relevant upstream neighbors. A hierarchical clustering of stations into sub-drainage groups makes training scalable by batching at the group level.

What would settle it

Retrain CSF on the same 73 stations with the causal adjacency matrix randomly rewired while preserving each station's number of upstream neighbors; if short-range NSE stays near 0.86 instead of falling toward the 0.78 of the plain STGCN, the reported improvement is not caused by the river flow graph.

Watch

Extended reading notes

Core claim

The paper claims that streamflow forecasting at a set of river stations improves when a spatio-temporal graph neural network is constrained by a river flow graph rather than by geographic proximity. In the proposed Causal Streamflow Forecasting (CSF) model, the adjacency matrix is built from Digital Elevation Models and Hydrologic Unit Codes so that message passing follows upstream-to-downstream connectivity and ignores stations in different drainage areas. Tested on daily data from 73 stations in the Brazos River basin, CSF reports the highest scores in NSE, KGE, VE, and Pearson correlation across short (1-day), medium (3-day), and long (7-day) forecast horizons; for short range, NSE is 0.86 versus 0.78 for the strongest baseline, STGCN. The ablation study attributes the gain to the combination of the river graph and a hierarchical grouping of stations.

Load-bearing premise

The river flow graph is assumed to be the true causal routing of water, even though engineered structures such as dams and diversions can change which stations actually influence which.

Editorial extensions

If this is right

  • If the central claim is correct, causal masking with hydrologically defined upstream-downstream edges is a reliable accuracy improvement over proximity-based graphs, not just an interpretability aid.
  • The reported short-range NSE gain (0.86 vs 0.78) means the design lifts daily forecast skill at a scale useful for water management decisions.
  • The two-stage decomposition—local runoff embedding plus basin-level routing—offers a faster alternative to full process simulation while retaining physical structure.
  • The ablation results imply that hierarchical batching and the causal graph are complementary; dropping either one costs accuracy, most visibly at longer horizons.
  • The same graph-guided recipe should transfer to other basins wherever elevation and watershed-boundary data exist.

Reading between the lines

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

  • Beyond the paper's claims, the river flow graph omits dams, reservoirs, and diversions, so the causal mask may be wrong in regulated basins; a stress test on a dam-heavy river would show whether the method's advantage survives when the graph misstates actual water routing.
  • Beyond the paper's claims, the runoff embedding is validated only by similarity to VIC-simulated runoff; one could test whether the embedding also predicts observed soil moisture or evapotranspiration to support a physical reading.
  • Beyond the paper's claims, a direct way to separate the causal-graph contribution from the hierarchical-batching contribution is to rewire the adjacency matrix randomly; if accuracy holds, the gain is not from causality but from regularization or grouping.
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 proposes Causal Streamflow Forecasting (CSF), a two-stage model that first learns latent runoff embeddings from meteorological forcings via a variational autoencoder (VAE), then feeds these embeddings into a spatio-temporal graph convolutional network (STGCN) whose spatial aggregation is guided by a river flow graph derived from DEM and HUC data. The model is evaluated on daily streamflow prediction at 73 stations in the Brazos River basin over a 3,650-day period (1967–1977), with short-range (7-day input, 1-day output), medium-range (14-day input, 3-day output), and long-range (28-day input, 7-day output) forecasting tasks. The reported results (Table I) show CSF outperforming MLP, CNN, Conv-LSTM, TCDF, MTGNN, and STGCN on NSE, KGE, VE, and ρ across all horizons; an ablation study (Table II) attributes the gains to the hierarchical network (HN) and river graph (RG) modules.

Significance. If the empirical results are reproducible, the paper makes a useful contribution by demonstrating that a hydrologically informed causal adjacency matrix can guide STGNNs to better streamflow forecasts than purely data-driven graph structures. The two-stage design that disentangles local runoff generation from basin-scale routing is conceptually appealing and mirrors the VIC-CaMa-Flood process chain. The ablation study and the runoff-embedding alignment analysis are valuable components. However, the lack of a clearly defined out-of-sample temporal split, the absence of uncertainty quantification, and the untested efficiency claim substantially limit the strength of the conclusions as currently written.

major comments (4)
  1. [§5.C (Experiment Settings) and Table I] The manuscript never defines the temporal train/validation/test split. The description of the 'rolling training pipeline' is ambiguous: it could describe expanding-window retraining, sliding-window validation, or in-sample reconstruction, and the text never states which time periods are held out during evaluation. Because the central claim in Section V.D.1 ("CSF consistently outperforms all baseline models") rests entirely on the metrics in Table I, these numbers must be computed on a held-out temporal period to be evidence of forecasting skill. Please state explicitly how the 3,650-day record is partitioned, and report results for a fixed test period that is never used for training or validation.
  2. [§5.D.1 (Forecasting Accuracy) and Table I] No error bars, confidence intervals, or significance tests are reported for any table. The reported margins (e.g., short-range NSE 0.86 for CSF vs. 0.78 for STGCN) could lie within run-to-run variation for neural network training, which is stochastic. The claim of consistent outperformance requires either multiple independent runs with variance reporting, or a paired statistical test across stations or time blocks. Without this, the robustness of the comparison cannot be assessed.
  3. [Abstract and §2 (Background)] The abstract claims that CSF "achieves higher computational efficiency compared to traditional simulation methods," but no experiment in Section V compares runtime, memory usage, or any efficiency measure against VIC-CaMa-Flood or any process-based model. This is a load-bearing part of the stated contributions; please either add a direct efficiency comparison or remove the claim from the abstract and the contributions list.
  4. [§3.B (Causal Adjacency Matrix) and Figure 3b] The river flow graph is treated as the true causal adjacency matrix, but it is generated purely from DEM and HUC data and does not account for artificial structures such as dams, reservoirs, and diversions, which are known to alter actual water routing in the Brazos basin. This assumption is structurally distinct from the performance claim, but it underlies the causal interpretation and the design rationale. The authors should explicitly acknowledge this limitation and, if possible, test sensitivity of the results to the presence of such structures (e.g., by removing or re-routing edges at known dam locations).
minor comments (5)
  1. [§5.B.1 (Runoff Embedding)] The validation of the learned runoff embeddings uses simulated runoff from the VIC model calibrated by the authors' prior work (reference [6]). This introduces a potential circularity because the same prior calibration informs the design. Please clarify that this is a proxy validation, not an independent ground-truth comparison, and consider an alternative validation (e.g., against observed streamflow-derived runoff proxies).
  2. [§4.A (Two-Stage Physically Aware Hierarchical Modelling), Eq. (4)] The total loss in Eq. (4) combines the station-level and prediction losses with a hyperparameter λ, but the manuscript never reports the chosen value of λ or any sensitivity analysis for it. Please provide the hyperparameter settings used in the experiments, including λ, embedding size (mentioned as 8 in Figure 6), and the STGCN architecture details (number of layers, hidden units, learning rate, etc.).
  3. [§5.C (Experiment Settings)] The description of the rolling training pipeline is insufficiently detailed for reproducing the multi-step forecasts. The sentence "the model is trained to predict streamflow for the next day, and its prediction is then used as input for forecasting subsequent days" suggests a recursive strategy, but the medium-range and long-range tasks use output windows of 3 and 7 days, respectively. Please clarify how the 3-day and 7-day outputs are produced (recursive vs. direct multi-step) and whether teacher forcing is used during training.
  4. [Throughout] There are several typographical errors and awkward phrasings, e.g., "forfor similating" in Section II, "hydrologicly" in Section I, and "wold" in Section I. A careful proofreading pass is recommended.
  5. [§5.D.2 (Runoff Embedding Similarities)] Figure 6 shows the kNN similarity versus embedding size, and Figure 7 shows similarity versus NSE. The text states that embedding size 8 "provides the best trade-off" and that higher similarity correlates with better NSE, but no quantitative evidence (e.g., correlation coefficient or standard deviation across runs) is provided. Please add the quantitative relationship or temper the interpretive claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central forecasting claim is an external benchmark comparison, the river-flow graph is a fixed prior rather than a fitted output, and the only self-citation is not load-bearing; the main weakness is an underspecified temporal split, which is a reporting/correctness risk, not a circular reduction.

full rationale

The paper's central claim (Section V.D.1, Table I) is an empirical benchmark comparison: CSF is trained under the same rolling pipeline as MLP, CNN, Conv-LSTM, TCDF, MTGNN, and STGCN, and the reported NSE, KGE, VE, and rho values are computed against observed streamflow, not reconstructed from the model's own fitted inputs. The causal adjacency matrix (Section III.B) is a fixed prior graph derived from DEM and HUC data, not a parameter learned from streamflow, so the forecast does not reduce to an input by construction. The VAE runoff embedding (Section IV.A.1) is trained with a reconstruction/ELBO loss plus the downstream prediction loss, and it is validated against VIC-simulated runoff (Section V.B.1); the simulated runoff comes from the authors' prior calibrated model (ref [6]), but it is not a training target of the VAE and does not enter the Table I comparison, so this self-citation is not load-bearing. The experiment section (Section V.C) describes a 'rolling training pipeline' without specifying the temporal train/validation/test partition; this is an unverified precondition and a reproducibility/correctness risk, but no quoted equation or fitted value shows that the reported predictions are equivalent to training targets by construction. Therefore no demonstrable circular step was found.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The model's contribution sits on a fixed hydrologic graph and a set of hyperparameters chosen without a separate validation set. No new physical entities are postulated; the 'runoff embedding' is a learned latent representation, not a new object. The main external support is the public DEM/HUC and USGS/Livneh data, while the VIC reference runoff comes from the authors' prior calibration.

free parameters (4)
  • VAE embedding size = 8 (selected from {4, 8, 16})
    Figure 6 shows embedding size 8 gives best balance across forecasting ranges; selected on the same dataset without a separate validation split (Section V.D.2).
  • Loss weight lambda = not reported
    Equation 4 introduces lambda to combine station-level and basin-level losses, but experiments do not report its value (Section IV.A.2).
  • Streamflow cap percentile = 99th percentile
    Streamflow data capped at 99th percentile to limit influence of extremes (Section V.A, Figure 4), a hand-set threshold affecting the task.
  • STGCN architecture hyperparameters = unspecified
    Number of layers, hidden units, learning rate, and temporal kernel sizes are not stated, leaving the model under-specified (Section IV.A.2).
assumptions (4)
  • domain assumption The river flow graph generated from DEM and HUC data correctly encodes causal hydrologic connectivity between stations (Section III.B, Figure 3b).
    The entire 'causal adjacency matrix' A is defined from this graph, and the STGCN message passing is masked by it. If the graph misses engineered flow modifications or drainage errors, the causal claim fails.
  • domain assumption VIC simulated runoff, calibrated in prior work [6], is a valid ground truth for evaluating the learned runoff embeddings (Section V.B.1).
    The mutual kNN alignment metric uses VIC runoff as the reference. VIC is a model, not an observation, and the calibration is from the authors' earlier paper.
  • standard math STGCN spatial and temporal convolution equations and VAE ELBO are valid and applicable as written (Section IV.A.2, Eq. 2).
    The paper relies on standard results from graph convolution and variational autoencoders without re-deriving them.
  • ad hoc to paper The rolling training pipeline produces valid forecasts for multi-step horizons (Section V.C).
    The model is trained to predict the next day and its prediction is used as input for subsequent days. This choice is not justified or compared with direct multi-step training, and it can accumulate errors.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Spatio-temporal Causal Learning for Streamflow Forecasting." pith.science (2026). https://pith.science/paper/7QEZO5JR

@misc{pith2026241117937,
  author       = {Pith},
  title        = {Pith review of: Spatio-temporal Causal Learning for Streamflow Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7QEZO5JR}},
  note         = {Machine review of arXiv:2411.17937}
}
read the original abstract

Streamflow plays an essential role in the sustainable planning and management of national water resources. Traditional hydrologic modeling approaches simulate streamflow by establishing connections across multiple physical processes, such as rainfall and runoff. These data, inherently connected both spatially and temporally, possess intrinsic causal relations that can be leveraged for robust and accurate forecasting. Recently, spatio-temporal graph neural networks (STGNNs) have been adopted, excelling in various domains, such as urban traffic management, weather forecasting, and pandemic control, and they also promise advances in streamflow management. However, learning causal relationships directly from vast observational data is theoretically and computationally challenging. In this study, we employ a river flow graph as prior knowledge to facilitate the learning of the causal structure and then use the learned causal graph to predict streamflow at targeted sites. The proposed model, Causal Streamflow Forecasting (CSF) is tested in a real-world study in the Brazos River basin in Texas. Our results demonstrate that our method outperforms regular spatio-temporal graph neural networks and achieves higher computational efficiency compared to traditional simulation methods. By effectively integrating river flow graphs with STGNNs, this research offers a novel approach to streamflow prediction, showcasing the potential of combining advanced neural network techniques with domain-specific knowledge for enhanced performance in hydrologic modeling.

Figures

Figures reproduced from arXiv: 2411.17937 by the authors.

Figure 1
Figure 1. The Brazos River Basin, TX. Each dot represents one monitoring station. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. VIC-CaMa-Flood model. Process-based hydrologic models are traditional ap￾proaches that simulate various physical processes in hydrol￾ogy. The Variably Infiltration Capacity (VIC) model [4] is a widely-used, large-scale distributed model that accounts for key hydrologic processes, including precipitation-runoff dynamics, evapotranspiration, soil moisture behavior, and river flow mechanics. VIC can be coupled with hyd… view at source ↗
Figure 3
Figure 3. Causal Streamflow Forecasting (CSF) model diagram. a regularization term ensuring the latent variables follow the prior distribution. 2) Stage 2: Basin-level Modeling: In the second stage, a basin-level model is utilized to account to capture the broader spatio-temporal dynamics across the entire river basin. The model uses a STGCN as the backbone, leveraging the river flow graph as the causal graph. The STGCN is de… view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: River flow graph. Each color represent a HUC8 subordination. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: The runoff embedding size versus the kNN similarity in different forecasting tasks. 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 Similarity 0.65 0.70 0.75 0.80 0.85 NSE Short Medium Long [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 30 canonical work pages

  1. [1]

    Streamflow forecasting: overview of ad- vances in data-driven techniques,

    P. Sharma and D. Machiwal, “Streamflow forecasting: overview of ad- vances in data-driven techniques,” Advances in Streamflow Forecasting, pp. 1–50, 2021

    work page 2021

  2. [2]

    Spatio-temporal graph convolutional net- works: a deep learning framework for traffic forecasting,

    B. Yu, H. Yin, and Z. Zhu, “Spatio-temporal graph convolutional net- works: a deep learning framework for traffic forecasting,” inProceedings of the 27th International Joint Conference on Artificial Intelligence , ser. IJCAI’18. AAAI Press, 2018, p. 3634–3640

    work page 2018

  3. [3]

    What role does hydrological science play in the age of machine learning?

    G. S. Nearing, F. Kratzert, A. K. Sampson, C. S. Pelissier, D. Klotz, J. M. Frame, C. Prieto, and H. V . Gupta, “What role does hydrological science play in the age of machine learning?” Water Resources Research, vol. 57, no. 3, p. e2020WR028091, 2021

    work page 2021

  4. [4]

    A simple hydrologically based model of land surface water and energy fluxes for general circulation models,

    X. Liang, D. P. Lettenmaier, E. F. Wood, and S. J. Burges, “A simple hydrologically based model of land surface water and energy fluxes for general circulation models,” Journal of Geophysical Research: Atmospheres, vol. 99, no. D7, pp. 14 415–14 428, 1994

    work page 1994

  5. [5]

    A physically based description of floodplain inundation dynamics in a global river routing model,

    D. Yamazaki, S. Kanae, H. Kim, and T. Oki, “A physically based description of floodplain inundation dynamics in a global river routing model,” Water Resources Research, vol. 47, no. 4, 2011

    work page 2011

  6. [6]

    Design principles for engineering wetlands to improve resilience of coupled built and natural water infrastructure,

    R. Shah, Y . Tsai, D. Stampoulis, H. G. Damavandi, and J. Sabo, “Design principles for engineering wetlands to improve resilience of coupled built and natural water infrastructure,” Environmental Research Letters, vol. 18, no. 11, p. 114045, 2023

    work page 2023

  7. [7]

    Modeling daily floods in the lancang-mekong river basin using an improved hydrological-hydrodynamic model,

    J. Wang, X. Yun, Y . Pokhrel, D. Yamazaki, Q. Zhao, A. Chen, and Q. Tang, “Modeling daily floods in the lancang-mekong river basin using an improved hydrological-hydrodynamic model,” Water Resources Research, vol. 57, no. 8, p. e2021WR029734, 2021

    work page 2021

  8. [8]

    Convolutional lstm network: A machine learning approach for precipitation nowcasting,

    X. Shi, Z. Chen, H. Wang, D.-Y . Yeung, W.-K. Wong, and W.-c. Woo, “Convolutional lstm network: A machine learning approach for precipitation nowcasting,” Advances in neural information processing systems, vol. 28, 2015

    2015

Show all 39 references
  1. [9]

    Rainfall–runoff modelling using long short-term memory (lstm) net- works,

    F. Kratzert, D. Klotz, C. Brenner, K. Schulz, and M. Herrnegger, “Rainfall–runoff modelling using long short-term memory (lstm) net- works,” Hydrology and Earth System Sciences , vol. 22, no. 11, pp. 6005–6022, 2018

    2018

  2. [10]

    Toward improved predictions in ungauged basins: Exploiting the power of machine learning,

    F. Kratzert, D. Klotz, M. Herrnegger, A. K. Sampson, S. Hochreiter, and G. S. Nearing, “Toward improved predictions in ungauged basins: Exploiting the power of machine learning,” Water Resources Research, vol. 55, no. 12, pp. 11 344–11 354, 2019

    2019

  3. [11]

    Spatiotemporal deep learning rainfall-runoff forecasting combined with remote sensing pre- cipitation products in large scale basins,

    S. Zhu, J. Wei, H. Zhang, Y . Xu, and H. Qin, “Spatiotemporal deep learning rainfall-runoff forecasting combined with remote sensing pre- cipitation products in large scale basins,” Journal of Hydrology, vol. 616, p. 128727, 2023

    2023

  4. [12]

    Fully distributed rainfall-runoff modeling using spatial-temporal graph neural network,

    Z. Xiang and I. Demir, “Fully distributed rainfall-runoff modeling using spatial-temporal graph neural network,” EarthArXiv eprints, p. X57P74, Jan. 2022

    2022

  5. [13]

    Neural general circulation models,

    D. Kochkov, J. Yuval, I. Langmore, P. Norgaard, J. Smith, G. Mooers, J. Lottes, S. Rasp, P. D ¨uben, M. Kl ¨ower et al. , “Neural general circulation models,” arXiv preprint arXiv:2311.07222 , 2023

    2023 arXiv

  6. [14]

    A bayesian neural network for an accurate representation and transformation of runoff dynamics: A case study of the brazos river basin in texas,

    H. G. Damavandi, D. Stampoulis, J. Sabo, R. Shah, L. Huang, Y . Wei, Y . Tsai, J. Srinivasan, T. Sinha, D. Boscovic, and G. Low, “A bayesian neural network for an accurate representation and transformation of runoff dynamics: A case study of the brazos river basin in texas,” J...

    2020

  7. [15]

    Accurate prediction of streamflow using long short-term memory network: a case study in the brazos river basin in texas,

    H. G. Damavandi, R. Shah, D. Stampoulis, Y . Wei, D. Boscovic, J. Sabo et al., “Accurate prediction of streamflow using long short-term memory network: a case study in the brazos river basin in texas,” International Journal of Environmental Science and Development, vol. 10, no...

    2019

  8. [16]

    Applying graph neural networks to improve the data resolution of stream water quality monitoring networks,

    T. Liu, Q. Deng, K. Ding, J. L. Sabo, H. Liu, K. S. Candan, and R. L. Muenich, “Applying graph neural networks to improve the data resolution of stream water quality monitoring networks,” in AGU Fall Meeting Abstracts, vol. 2023, 2023, pp. H42E–03

    2023

  9. [17]

    A deep learning approach to distributed rainfall-runoff modeling with spatiotemporal interpretation

    Q. Deng, T. Liu, Y . Wei, and J. L. Sabo, “A deep learning approach to distributed rainfall-runoff modeling with spatiotemporal interpretation.” in AGU Fall Meeting Abstracts , vol. 2022, 2022, pp. IN12B–0271

    2022

  10. [18]

    Artificial intelligence tech- niques in hydrology and water resources management,

    F.-J. Chang, L.-C. Chang, and J.-F. Chen, “Artificial intelligence tech- niques in hydrology and water resources management,” p. 1846, 2023

    2023

  11. [19]

    A transdisciplinary review of deep learning research and its relevance for water resources scientists,

    C. Shen, “A transdisciplinary review of deep learning research and its relevance for water resources scientists,” Water Resources Research , vol. 54, no. 11, pp. 8558–8593, 2018

    2018

  12. [20]

    Spatio-temporal neural structural causal models for bike flow prediction,

    P. Deng, Y . Zhao, J. Liu, X. Jia, and M. Wang, “Spatio-temporal neural structural causal models for bike flow prediction,” in Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence and Thirty-Fifth Conference on Innovative Applications of Artificial Inte...

  13. [21]

    Deciphering spatio-temporal graph forecasting: A causal lens and treatment,

    Y . Xia, Y . Liang, H. Wen, X. Liu, K. Wang, Z. Zhou, and R. Zimmer- mann, “Deciphering spatio-temporal graph forecasting: A causal lens and treatment,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  14. [22]

    Causal discovery for feature selection in phys- ical process-based hydrological systems,

    P. Sheth, T. Liu, D. Doner, Q. Deng, Y . Wei, R. Muenich, J. Sabo, K. S. Candan, and H. Liu, “Causal discovery for feature selection in phys- ical process-based hydrological systems,” in 2022 IEEE International Conference on Big Data (Big Data) . IEEE, 2022, pp. 5568–5577

    2022

  15. [23]

    Stcd: A spatio- temporal causal discovery framework for hydrological systems,

    P. Sheth, R. Shah, J. Sabo, K. S. Candan, and H. Liu, “Stcd: A spatio- temporal causal discovery framework for hydrological systems,” in 2022 IEEE International Conference on Big Data (Big Data) . IEEE, 2022, pp. 5578–5583

    2022

  16. [24]

    Causal discovery with attention- based convolutional neural networks,

    M. Nauta, D. Bucur, and C. Seifert, “Causal discovery with attention- based convolutional neural networks,” Machine Learning and Knowl- edge Extraction, vol. 1, no. 1, p. 19, 2019

    2019

  17. [25]

    Streams: Towards spatio-temporal causal discovery with reinforcement learning for streamflow rate prediction,

    P. Sheth, A. Mosallanezhad, K. Ding, R. Shah, J. Sabo, H. Liu, and K. S. Candan, “Streams: Towards spatio-temporal causal discovery with reinforcement learning for streamflow rate prediction,” in Proceedings of the 32nd ACM International Conference on Information and Knowledge...

    2023

  18. [26]

    Toward causal inference for spatio-temporal data: Conflict and forest loss in colombia,

    R. Christiansen, M. Baumann, T. Kuemmerle, M. D. Mahecha, and J. Peters, “Toward causal inference for spatio-temporal data: Conflict and forest loss in colombia,” Journal of the American Statistical Association, vol. 117, no. 538, pp. 591–601, 2022

    2022

  19. [27]

    Space: The spatial confounding environment,

    M. Tec, A. Trisovic, M. Audirac, S. M. Woodward, J. K. Hu, N. Khosh- nevis, and F. Dominici, “Space: The spatial confounding environment,” in The Twelfth International Conference on Learning Representations , 2023

    2023

  20. [28]

    Auto-encoding variational bayes,

    D. P. Kingma and M. Welling, “Auto-encoding variational bayes,”

  21. [29]

    Cluster-gcn: An efficient algorithm for training deep and large graph convolutional networks,

    W.-L. Chiang, X. Liu, S. Si, Y . Li, S. Bengio, and C.-J. Hsieh, “Cluster-gcn: An efficient algorithm for training deep and large graph convolutional networks,” in Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining , 2019, pp. 257–266

    2019

  22. [30]

    A long-term hydrologically based dataset of land surface fluxes and states for the conterminous united states: Update and extensions,

    B. Livneh, E. A. Rosenberg, C. Lin, B. Nijssen, V . Mishra, K. M. An- dreadis, E. P. Maurer, and D. P. Lettenmaier, “A long-term hydrologically based dataset of land surface fluxes and states for the conterminous united states: Update and extensions,” Journal of Climate , vol....

    2013

  23. [31]

    Similarity of neural network models: A survey of functional and representational measures,

    M. Klabunde, T. Schumacher, M. Strohmaier, and F. Lemmerich, “Similarity of neural network models: A survey of functional and representational measures,” arXiv preprint arXiv:2305.06329 , 2023

    2023 arXiv

  24. [32]

    Hybrid machine learning framework for hydrological assessment,

    J. Kim, H. Han, L. E. Johnson, S. Lim, and R. Cifelli, “Hybrid machine learning framework for hydrological assessment,” Journal of hydrology, vol. 577, p. 123913, 2019

    2019

  25. [33]

    River flow forecasting through conceptual models part i—a discussion of principles,

    J. E. Nash and J. V . Sutcliffe, “River flow forecasting through conceptual models part i—a discussion of principles,” Journal of hydrology, vol. 10, no. 3, pp. 282–290, 1970

    1970

  26. [34]

    On typical range, sensitivity, and normaliza- tion of mean squared error and nash-sutcliffe efficiency type metrics,

    H. V . Gupta and H. Kling, “On typical range, sensitivity, and normaliza- tion of mean squared error and nash-sutcliffe efficiency type metrics,” Water Resources Research, vol. 47, no. 10, 2011

    2011

  27. [35]

    Do nash values have value? discus- sion and alternate proposals,

    R. E. Criss and W. E. Winston, “Do nash values have value? discus- sion and alternate proposals,” Hydrological Processes: An International Journal, vol. 22, no. 14, pp. 2723–2725, 2008

    2008

  28. [36]

    Gradient-based learning applied to document recognition,

    Y . LeCun, L. Bottou, Y . Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proceedings of the IEEE , vol. 86, no. 11, pp. 2278–2324, 1998

    1998

  29. [37]

    Con- necting the dots: Multivariate time series forecasting with graph neural networks,

    Z. Wu, S. Pan, G. Long, J. Jiang, X. Chang, and C. Zhang, “Con- necting the dots: Multivariate time series forecasting with graph neural networks,” in Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining , 2020, pp. 753–763

    2020

  30. [2022]

    Available: https://arxiv.org/abs/1312.6114

    [Online]. Available: https://arxiv.org/abs/1312.6114

    arXiv

  31. [2023]

    Available: https://doi-org.ezproxy1.lib.asu.edu/10.1609/ aaai.v37i4.25542

    [Online]. Available: https://doi-org.ezproxy1.lib.asu.edu/10.1609/ aaai.v37i4.25542

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.